The Cybernetic Brain
Page 9
GREY WALTER,THE LIVING BRAIN (1953, 51)
The tortoise served as a model of the adaptive brain, but only a primitive one. It lived in real time, reacting to environmental cues (lights, contacts) as they happened to it and never learning anything from its experience. Walter quickly sought to go beyond this limitation by building in a second layer of adaptability, and he concluded his first publication on the tortoise by mentioning that the "more complex models that we are now constructing have memory circuits" (1950a, 45). These more complex models entailed two modifications to the basic tortoise. One was to equip it with more senses—wiring a microphone into its circuits, for example, to give it a sensitivity to sound as well as light. The other was the addition of a clever circuit called CORA, for conditioned reflex analogue (figs. 3.8, 3.9). Wired into the tortoise, CORA converted Machina speculatrix to Machina docilis, as Walter called it—the easily taught machine. CORA detected repeated coincident inputs in different sensory channels and, when a certain threshold of repetition was reached, opened up a link from one sense to the other—so that the tortoise would become "conditioned" to react to a sound, say, in the way that it had hitherto reacted to the contact switch on its body.42
Figure 3.8.CORA. Source: de Latil 1956, facing p. 51.
Figure 3.9.CORA: circuit diagram. Source: Walter 1953, 295.
What can we say about CORA? As its name implies, CORA was intended to emulate classical Pavlovian conditioning in animals. As a student at Cambridge, Walter had worked with a student of Pavlov for more than a year to set up a conditioning laboratory, "mastering the technique and improving it by the introduction of certain electronic devices." When "the great" Pavlov himself visited England, Walter, "as the English exponent of his work . . . had the privilege of discussing it with him on familiar terms. . . . I asked him if he saw any relation between the two methods of observing cerebral activity, his and Berger's [EEG readings]. . . . But Pavlov showed no desire to look behind the scenes. He was not in the least interested in cerebral mechanisms" (Walter 1953, 51–52).43 CORA, in contrast, was explicitly a further scientific attempt to look behind the scenes, to open up the Black Box of the adaptive brain by building a model that could mimic its performances, just like the tortoises before that.
CORA did, indeed, make a connection between the electrical rhythms of the brain and conditioned learning. The key element in connecting one sense to another was precisely the build up of an oscillating voltage in CORA, and Walter laid much store by this, even arguing that CORA displayed the contingent negative variation in electrical potential which was his most important contribution to neurophysiology (1966, 13), but I cannot explore this further here.44 Instead, I want to comment on CORA as brain science from several angles before connecting it to Walter's psychiatric milieu.
Reversing the earlier order, I begin with a quick comment on the relation of CORA to the social basis of cybernetics. CORA was a virtuoso piece of electrical engineering, in both its design and construction. The tortoise was imitable—by Frazer, Brooks, and many others—but CORA was inimitable. I know of no attempts to replicate it, never mind take the development of the tortoise beyond it.45 Even Walter discontinued his robotics after CORA. Machina speculatrixpointed to a difficult but, to some—odd schoolboys like Frazer and Brooks; contributors to the Namur conferences—manageable synthesis of brain science and engineering. Machina docilis upped the ante too far. Nothing grew specifically from it in the cybernetic tradition. In the late 1980s revival of Walter-style robotics, machines that learned were indeed built, but that learning was based on neural networks, not CORA-style electronics, and the oscillatory features that intrigued Walter were lost (at least, temporarily). In that sense, CORA remains an unexploited resource in the history of cybernetics.
_ _ _ _ _
CORA also invites us to extend the discussion of cybernetic ontology to encompass epistemology. The great novelty of M. docilis was that it acquired a sort of knowledge about the world: it learned what to associate with what. How should we think about that? The point I would emphasize is that the knowledge of M. docilis waxed and waned with its performance, integrating over its experience—associations between stimuli would be lost if the robot's expectations did not continue to be reinforced—and thus functioned as a heuristic guide, emerging from and returning to performance, not as any kind of controlling center. Docilis thus offers us a vision of knowledge as engaged in, and as part of, performance, rather than as a thing itself or as some sort of external determinant of action—a vision of knowledge as being in the plane of practice, as I put it in The Mangle of Practice, not above it. Much as speculatrix acted out for us a performative ontology, then, docilis also staged a performative epistemology, as I called it in chapter 2—an appreciation of knowledge not as a hopefully definitive mapping of the world, but as another component of performance. This is the vision of knowledge that goes with the cybernetic ontology, and that we will see elaborated in succeeding chapters.46
Cybernetics and Madness
PSYCHIATRISTS USED TO BE REPROACHED WITH THEIR LACK OF THERAPEUTIC ZEAL; IT WAS SAID THEY WERE RESIGNED WHEN THEY SHOULD HAVE BEEN HOPEFUL AND ENTERPRISING, AND TORPID WHEN ENERGY WAS CALLED FOR. WHETHER THE REPROACH WAS DESERVED OR NOT IT SEEMS TO HAVE STUNG THEM INTO THERAPEUTIC FURY. CONTINUOUS NARCOSIS, INSULIN SHOCK AND CARDIAZOL FITS HAVE PROVED THAT THE PSYCHIATRIST IS AT LEAST AS DARING AS THE SURGEON NOWADAYS. THE PAPERS BY KALINOWSKY IN OUR ISSUE OF DEC. 9, AND BY FLEMING, GOLLA AND WALTER IN THIS RECORD ANOTHER BOLD STEP. . . . BUT WE MUST NOT LET UNCONSCIOUS ASSOCIATIONS WITH WHAT IS DONE PERIODICALLY IN A ROOM AT SING SING PREJUDICE US AGAINST WHAT MAY WELL TURN OUT TO BE A VALUABLE STEP FORWARD.
EDITORIAL,"MORE SHOCKS," LANCET, 234, NO. 6070 (30 DECEMBER 1939): 1373
It is tempting to think of the tortoises and CORA as freestanding scientific and engineering projects, divorced from mundane concerns. Walter may have mentioned the tortoise's potential as an autonomous weapons system, but he did nothing to pursue it. On the other hand, one of the first things he did with CORA was drive his tortoises mad. This points to connections between his cybernetics and psychiatry that we can explore.
If the CORA-equipped tortoise could be understood as a model of a normal brain, Walter was keen to show that it was a model for the pathological brain too. In his first article on the tortoise, Walter (1950a, 45) noted that, with
Figure 3.10."Prefrontal lobotomy, Sir . . ." Source: Beer 1994a, 162. (Courtesy of Cwarel Isaf Institute and Malik Management Zentrum St. Gallen [www.management .kybernetik.com,www.malik-mzsg.ch].)
CORA, "the possibility of a conflict neurosis immediately appears," and in a follow-up article in August 1951 he observed that (63) "it becomes only too easy to establish an experimental neurosis. Thus if the arrangement is such that the sound becomes positively associated both with the attracting light and with the withdrawal from an obstacle, it is possible for both a light and a sound to set up a paradoxical withdrawal. The 'instinctive' attraction to a light is abolished and the model can no longer approach its source of nourishment. This state seems remarkably similar to the neurotic behavior produced in human beings by exposure to conflicting influences or inconsistent education." Or, as he put it more poetically in The Living Brain (1953, 183), "in trying, as it were, to sort out the implications of its dilemma, the model ends up, 'sicklied o'er with the pale cast of thought,' by losing all power of action.47
The idea that mental problems might be precipitated by conflicting patterns of conditioning was not original to Walter. As he acknowledged, its history went back to the induction of "experimental neuroses" in dogs by clashing conditioning regimes, carried out in the laboratory of "the master," Pavlov himself, in 1921 (Gray 1979, 119; Pavlov 1927).48 And in March 1950, for example, two months before Walter's first tortoise article appeared, Scientific American featured an article entitled "Experimental Neuroses" by Jules Masserman, a psychiatrist at Northwestern University, which discussed the induction of pathological symptoms by cross-conditioning
in cats. Drawing upon Auguste Comte's typology, Masserman argued that the experimentalization of neuroses moved psychiatry from the "mystic" and "taxonomic" stages into the ultimate "dynamic" phase of "science" (Masserman 1950). Walter could have made the same point about his experiments with CORA. And one could say he had gone one step beyond Masserman. Not content with simply demonstrating that cross-conditioning could produce pathological behavior, he had, again, produced an electromechanical model which enabled one to grasp the go of this process at the hardware level.
It is thus revealing to think of cybernetics as a science of psychiatry, not in the sense that it could be reduced to psychiatry—even with the tortoise it already overflowed the bounds of the brain—but in the sense that psychiatry was a surface of emergence (Pickering 2005b) for cybernetics: Walter's cybernetics (and Ashby's) grew out of the phenomena and problematics of his psychiatric milieu. And we can take this line of thought further in a couple of directions. One is to note that after driving his tortoises mad, Walter cured them (1953, 184): "When a complex learning model develops an excess of depression or excitement, there are three ways of promoting recovery. After a time the conflicting memories may die away—except in obsessional states. . . . Switching off all circuits and switching on again clears all lines and provides, as it were, a new deal for all hands. Very often it has been necessary to disconnect a circuit altogether—to simplify the whole arrangement." And in case his readers missed the point, Walter went on to analogize these electromechanical procedures to those of psychiatric therapy, adding his cybernetic apologia for the latter:
Psychiatrists also resort to these stratagems—sleep [leaving the machine alone for a long time], shock [switching it off and on again] and surgery [disconnecting electrical circuits within it]. To some people the first seems natural, the second repulsive, and the third abhorrent. Everyone knows the benison of sleep, and many have been shocked into sanity or good sense, but the notion that a mental disorder could be put right by cutting out or isolating a part of the brain was an innovation which roused as much indignation and dispute as any development in mental science. There are volumes of expert testimony from every point of view, but our simple models would indicate that, insofar as the power to learn implies the danger of breakdown, simplification by direct attack may well and truly arrest the accumulation of self-sustaining antagonism and "raze out the written troubles of the brain."
So cybernetics was a science of psychiatry in a double sense, addressing the go of both mental disorder and psychiatric therapy and offering a legitimation of the latter along the way. And since Walter does not use the usual terms for the therapies he mentions, we should note that we are plunged here into the "great and desperate cures"—insulin shock, chemical shock, electroshock (electroconvulsive therapy—ECT), and lobotomy—that arose in psychiatry in the 1930s and had their heyday in the 1940s and early 1950s, the same period as the first flush of cybernetics.49
Figure 3.11.ECT brochure, cover page. Source: Science Museum, London, BNI archive.
And to put some more flesh on this connection, we should note that despite his evident desire to "do science" as he had in his student days at Cambridge, Walter continually found himself entangled with the concerns of the clinic. During his brief stint in London, he wrote in a 1938 report to the Rockefeller Foundation on his EEG work (1938, 16) that "the volume of clinical work which I have been asked to undertake has grown to embarrassing proportions. . . . These examinations are, of course, undertaken most willingly . . . but the clerical and other routine work, to say nothing of the maintenance of apparatus . . . take up so much time that little is left for breaking new ground."50 Walter's later work on flicker (see below) also had a significant clinical element. But more directly to the point here is that the newly established Burden Neurological Institute took the lead in transplanting the new approaches to psychiatry to Britain, claiming an impressive list of firsts, including the first use of ECT in Britain in 1939 and the first prefrontal leucotomy in 1940 (Cooper and Bird 1989). And Walter was very much involved in these achievements. His technical skill and creativity were such that he had a standing relationship with the Ediswan Company in the development of commercial apparatus, and Britain's first commercial ECT machine was developed by Ediswan and carried on the title page of its brochure the statement that it was built "to the specification of Mr. Grey Walter" (fig. 3.11).51 Walter was one of the authors of the first three papers to appear from the Burden on ECT. The earliest of these appeared in the Lancetin December 1939 (Fleming, Golla, and Walter 1939), describes ECT as EEG in reverse, and includes EEG readings of two post-ECT patients.52 During World War II Walter himself performed ECT treatments on American soldiers suffering from "battle fatigue."53
Walter's interest in mental pathologies and therapies was thus by no means that of a detached observer, and if one wanted to identify the worldly matrix from which his cybernetics emerged, it would have to be psychiatry; more specifically the psychiatry of the great and desperate cures; and more specifi- cally still the world of electroshock, electroconvulsive therapy, ECT.54
_ _ _ _ _
Two remarks to end this section. In chapter 1 I said that it was interesting to think of cybernetics as one of Deleuze and Guattari's "nomad sciences" that destabilize the state, and we can come back to that thought now. Earlier in the present chapter I described the nomadic wandering of Walter's cybernetics through inter- and antidisciplinary worlds such as the Ratio Club and the Namur conferences, and yet when it came to real-world psychiatry Walter's work was evidently no threat to the established order. What should we make of this? The obvious remark is that Walter's cybernetics was adjusted to his professional career in a double way: its radical aspects flourished outside psychiatry's gates, and within those gates it was domesticated to conform to the status quo. There is no need to be cynical about this: in the forties and early fifties it was possible to be optimistic about the great and desperate psychiatric cures (compared with what had gone before), and there is no reason to doubt that Walter (and Ashby) were genuinely optimistic. Nevertheless, we can also see how ontology and sociology were correlated here. As will become clearer in chapter 5 when we discuss Bateson and Laing, it was possible to go much further than Walter in developing the cybernetic theme of adaptation in psychiatry, but the price of this was a transformation of the social relations of doctors and patients that did, in a Deleuzian fashion, threaten the established order.
Second, it is interesting to ask where this line of first-generation cybernetic psychiatry went. The answer is: nowhere. In Britain, Walter and Ashby were the leading theorists of mental pathology and therapy in the forties and fifties, with their models offering a new understanding of the brain, madness, and its treatment, but histories of twentieth-century psychiatry give them hardly a mention (try Valenstein 1986 and Shorter 1997). And one can think of several reasons why this should be. The first takes us back to the sheer oddity of cybernetics. Walter remained to a significant degree an outsider to psychiatry in his specifically cybernetic work, and it was also the case that Walter's cybernetics made little constructive contribution to psychiatric therapy—it offered an explanation of the mechanisms of ECT and lobotomy without suggesting new therapeutic approaches. Again, one might imagine that Walter had got as far as he could with CORA. It is not clear what his next step might have been in developing this line of research further, or where he could have found support for what would probably have been a significant engineering effort.
But beyond all that, we need to think about two broader developments bearing on psychiatry as a clinical field. The first was the introduction in the 1950s of psychoactive drugs that proved effective in controlling the symptoms of mental disorder, beginning with chlorpromazine (Shorter 1997, chap. 7; Rose 2003). These drugs had their own unfortunate side effects but did not entail the violence and irreversibility of lobotomy, which went into rapid decline in the mid-1950s. ECT enjoyed a longer history, up to the present, but its use also declined in
the face of drugs, and the technique lost its cutting-edge, if I can say that, status as the most advanced form of psychiatric therapy.55 Cybernetics was thus left high and dry in the later 1950s, as a science of clinical practices which were, if not entirely extinct, at least less prevalent than they had been in the 1940s. It is hardly surprising, then, that Walter found other lines of research more attractive from the mid-1950s onward. Ashby continued developing his own cybernetics as a science of psychiatry into the later 1950s, but, as we shall see, he too abandoned his psychiatrically oriented research from around 1960.
The other development we need to think about here is a growing critique in the 1950s of violent psychiatric therapies and even of the use of antipsychotic drugs, a critique which burst into popular consciousness in the 1960s as what was often called the antipsychiatry movement. This movement was not, despite the name, pure critique. It entailed a different way of conceptualizing and acting upon mental disorders, which was, as it happens, itself cybernetic. This gets us back to Bateson and Laing, and on to chapter 5.
Strange Performances
For the remainder of this chapter I want to come at Walter's work from a different angle. The tortoises and CORA were Walter's most distinctive contribution to the early development of cybernetics, but they occupied him only in the late 1940s and early 1950s, and here we can examine some of his more enduring concerns with the brain—and how they crossed over into the counterculture of the sixties.