The Cybernetic Brain
Page 4
So what? I want to say that cybernetics drew back the veil the modern sciences cast over the performative aspects of the world, including our own being. Early cybernetic machines confront us, instead, with interesting and engaged material performances that do not entail a detour through knowledge. The phrase that runs through my mind at this point is ontological theater. I want to say that cybernetics stageda nonmodern ontology for us in a double sense. Contemplation of thermostats, tortoises, and homeostats helps us, first, to grasp the ontological vision more generally, a vision of the world as a place of continuing interlinked performances. We could think of the tortoise, say, exploring its world as a little model of what the world is like in general, an ontological icon. Going in the other direction, if one grasps this ontological vision, then building tortoises and homeostats stages for us examples of how it might be brought down to earth and played out in practice, as robotics, brain science, psychiatry, and so on. The many cybernetic projects we will examine can all stand as ontological theater in this double sense: as aids to our ontological imagination, and as instances of the sort of endeavors that might go with a nonmodern imagining of the world.6
This modern/nonmodern contrast is a key point for all that follows. I want in particular to show that the consistent thread that ran through the history of British cybernetics was the nonmodern performative ontology I have just sketched out. All of the oddity and fascination of this work hangs together with this unfamiliar vision of the sort of place the world is. And I can immediately add a corollary to that observation. In what follows, I am interested in cybernetics as ontological theater in both of the senses just laid out—as both an aid to our imaginations and as exemplification of the fact that, as I said earlier, ontology makes a difference. I want to show that how we imagine the world and how we act in it reciprocally inform one another. Cybernetic projects, in whatever field, look very different from their modern cognates.
From here we can proceed in several directions. I turn first to the "so what?" question; then we can go into some important nuances; finally, we can go back to the critique of cybernetics and the politics of ontology.
_ _ _ _ _
THE ESSENCE OF LIFE IS ITS CONTINUOUSLY CHANGING CHARACTER; BUT OUR CONCEPTS ARE ALL DISCONTINUOUS AND FIXED, . . . AND YOU CAN NO MORE DIP UP THE SUBSTANCE OF REALITY WITH THEM THAN YOU CAN DIP UP WATER WITH A NET, HOWEVER FINELY MESHED.
WILLIAM JAMES,"BERGSON AND INTELLECTUALISM" (1943 [1909, 1912], 253)
Why should we be interested in cybernetics? Haven't modern science and engineering served us well enough over the past few hundred years? Of course, their achievements have been prodigious. But I can still think of a few reasons why it might be interesting and useful to try understanding the world in a different way:
1. It is an exercise in mental gymnastics: the White Queen (or whoever it was) imagining a dozen impossible things before breakfast. Some of us find it fun to find new ways to think, and sometimes it leads somewhere (Feyerabend 1993).
2. Perhaps modern science has succeeded too well. It has become difficult for us to recognize that much of our being does not have a cognitive and representational aspect. I suppose I could figure out how my doorknob works, but I don't need to. I established a satisfactory performative relation with doorknobs long before I started trying to figure out mechanisms. A science that helped us thematize performance as prior to representation might help us get those aspects of our being into focus. And, of course, beyond the human realm, most of what exists does not have the cognitive detour as an option. It would be good to be able to think explicitly about performative relations between things, too.
3. Perhaps there would be positive fruits from this move beyond the representationalism of modern science. In engineering, the thermostat, the tortoise, the homeostat, and the other nonmodern cybernetic projects we will be looking at all point in this direction.
4. Perhaps in succeeding too well, modern science has, in effect, blinded us to all of those aspects of the world which it fails to get much grip upon. I remember as a physicist trying to figure out why quarks always remained bound to one another and reflecting at the same time that none of us could calculate in any detail how water flowed out of a tap. Contemporary complexity theorists like to argue that the methods of modern science work nicely for a finite class of "linear" systems but fail for "nonlinear" systems—and that actually the latter are in some sense most of the world. Stafford Beer foreshadowed this argument in his first book, Cybernetics and Management,where he argued that we could think of the world as built from three different kinds of entities or systems (Beer 1959, 18). We can go into this in more detail in chapter 6, but, briefly, Beer referred to these as "simple," "complex," and "exceedingly complex" systems. The first two kinds, according to Beer, are in principle knowable and predictable and thus susceptible to the methods of modern science and engineering. Exceedingly complex systems, however, are not. They are systems that are so complex that we can never fully grasp them representationally and that change in time, so that present knowledge is anyway no guarantee of future behavior. Cybernetics, on Beer's definition, was the science of exceedingly complex systems that modern science can never quite grasp.
I will come back repeatedly to Beer's idea of exceedingly complex systems as we go along, and try to put more flesh on it. This is the aspect of cybernetics that interests me most: the aspect that assumes an ontology of unknowability, as one might call it, and tries to address the problematic of getting along performatively with systems that can always surprise us (and this takes us back to the adaptive brain, and, again, to nonhuman systems that do not have the option of the cognitive detour). If there are examples of Beer's exceedingly complex systems to be found in the world, then a nonmodern approach that recognizes this (rather than, or as well as, a modern one that denies it) might be valuable. It is not easy, of course, to say where the dividing line between aspects of the world that are "exceedingly complex" rather than just very complicated is to be drawn. Modern science implicitly assumes that everything in the world will eventually be assimilated to its representational schema, but the time horizon is infinite. Here and now, therefore, a cybernetic stance might be appropriate in many instances. This is where the intellectual gymnastics get serious, and where the history of cybernetics might be needed most as an aid to the imagination.
5. I may as well note that my interest in cybernetics stems originally from a conviction that there is indeed something right about its ontology, especially the ontology of unknowability just mentioned. As I said earlier, I arrived at something very like it through my empirical studies in the history of modern science, though the substance of scientific knowledge speaks to us of a different ontology. I lacked the vocabulary, but I might have described modern science as a complex adaptive system, performatively coming to terms with an always-surprising world. At the time, I thought of this as a purely theoretical conclusion. When pressed about its practical implications, I could not find much to say: modern science seems to get on pretty well, even as it obscures (to my way of thinking) its own ontological condition.7 The history of cybernetics, however, has helped me to see that theory, even at the level of ontology, can return to earth. Cybernetic projects point to the possibility of novel and distinctive constructive work that takes seriously a nonmodern ontology in all sorts of fields. They show, from my perspective, where the mangle might take us. And one further remark is worth making. Theory is not enough. One cannot deducethe homeostat, or Laing's psychiatry, or Pask's Musicolour machine from the cybernetic ontology or the mangle. The specific projects are not somehow already present in the ontological vision. In each instance creative work is needed; something has to be added to the ontological vision to specify it and pin it down. That is why we need to be interested in particular manifestations of cybernetics as well as ontological imaginings. That is how, from my point of view,cybernetics carries us beyond the mangle.8
_ _ _ _ _
Now for the nuances. First, knowledge.
The discussion thus far has emphasized the performative aspect of cybernetics, but it is important to recognize that cybernetics was not simply and straightforwardly antirepresentational. Representational models of a firm's economic environment, for example, were a key part of Beer's viable system model (VSM) of the organization. Once one sees that, the clean split I have made between cybernetics and modern science threatens to blur, but I think it is worth maintaining. On the one hand, I want to note that many cybernetic projects did not have this representational aspect. The great advantage that Beer saw in biological computing was that it was immediately performative, involving no detours through the space of representation. On the other hand, when representations did appear in cybernetic projects, as in the VSM, they figured as immediately geared into performance, as revisable guides to future performance rather than as ends in themselves. Beer valued knowledge, but he was also intensely suspicious of it—especially of our tendency to mistake representations for the world, and to cling to specific representations at the expense of performance. We might thus think of cybernetics as staging for us a performative epistemology, directly engaged with its performative ontology—a vision of knowledge as part of performance rather than as an external controller of it. This is also, as it happens, the appreciation of knowledge that I documented and argued for in The Mangle.
_ _ _ _ _
Now that we have these two philosophical terms on the table—ontology and epistemology—I can say more about my own role in this history. In chapter 1 I said that anyone can have their own history of cybernetics, and this one is mine. I picked the cast of characters and which aspects of their work to dwell upon. But beyond that, the emphasis on ontology is more mine than the cyberneticians'. It is the best way I have found to grasp what is most unfamiliar and valuable about cybernetics, but the fact is that the word "ontology" does not figure prominently in the cybernetic literature. What I call the cybernetic ontology tends to be simply taken for granted in the literature or not labeled as such, while "epistemology" is often explicitly discussed and has come increasingly to the fore over time. Contemporary cyberneticians usually make a distinction between "first-order" cybernetics (Walter and Ashby, say) and "second-order" cybernetics (Bateson, Beer, and Pask), which is often phrased as the difference between the cybernetics of "observed" and "observing" systems, respectively. Second-order cybernetics, that is, seeks to recognize that the scientific observer is part of the system to be studied, and this in turn leads to a recognition that the observer is situated and sees the world from a certain perspective, rather than achieving a detached and omniscient "view from nowhere." Situated knowledge is a puzzling and difficult concept, and hence follows an intensified interest in the problematic of knowledge and epistemology.
What should I say about this? First, I take the cybernetic emphasis on epistemology to be a symptom of the dominance of specifically epistemological inquiry in philosophy of science in the second half of the twentieth century, associated with the so-called linguistic turn in the humanities and social sciences, a dualist insistence that while we have access to our own words, language, and representations, we have no access to things in themselves. Cybernetics thus grew up in a world where epistemology was the thing, and ontology talk was verboten. Second, my own field, science studies, grew in that same matrix, but my own research in science studies has convinced me that we need to undo the linguistic turn and all its works. The shift from a representational to a performative idiom for thinking about science, and from epistemology alone to ontology as well, is the best way I have found to get to grips with the problematic of situated knowledge (and much else).
So I think that second-order cybernetics has talked itself into a corner in its intensified emphasis on epistemology, and this book could therefore be read as an attempt to talk my way out of the trap. Again, of course, the "so what?" question comes up. Words are cheap; what does it matter if I use the word "ontology" more than the cyberneticians? Actually—though it is not my reason for writing the book—something might be at stake. Like ontology itself, ontology talk might make a difference. How one conceives a field hangs together with its research agendas. To see cybernetics as being primarily about epistemology is to invite endless agonizing about the observer's personal responsibility for his or her knowledge claims. Fine. But the other side of this is the disappearance of the performative materiality of the field. All of those wonderful machines, instruments, and artifacts get marginalized if one takes cybernetics to be primarily about knowledge and the situatedness of the observer. Tortoises, homeostats, biological computers, Musicolour machines, adaptive architecture—all of these are just history as far as second-order cybernetics is concerned. We used to do things like that in our youth; now we do serious epistemology.
Evidently, I think this position is a mistake. I am interested in cybernetics as the field that brought nonmodern ontology down to earth, and played it out and staged it for us in real projects. I think we need more of this kind of thing, not less. I did not make the history up; I don't have enough imagination; it has taken me years to find it out and struggle with it. But the chapters that follow invite, in effect, a redirection of cybernetics. I think the field might be far more lively and important in the future if it paid attention to my description of its past.
_ _ _ _ _
Now for the trickiest point in this chapter. I began with Black Boxes and the differing stances toward them of modern science and cybernetics: the former seeking to open them up; the latter imagining a world of performances in which they remained closed. This distinction works nicely if we want to think about the work of the second-generation cyberneticians, Beer and Pask, and also Bateson and Laing. Nothing more needs to be said here to introduce them. But it works less well for the first generation, Walter and Ashby, and this point needs some clarification.
I quoted Ashby earlier defining the problematic of the Black Box in terms of an engineer probing the Box with electrical inputs and and observing its outputs. Unfortunately for the simplicity of my story, the quotation continues: "He is to deduce what he can of its contents." This "deduction" is, needless to say, the hallmark of the modern scientific stance, the impulse to open the box, and a whole wing of Ashby's cybernetics (and that of his students at Illinois in the 1960s) circled around this problematic. Here I am tempted to invoke the author's privilege and say that I am not going to go into this work in any detail in what follows. While technically fascinating, it does not engage much with the ontological concerns which inform this book. But it is not so easy to get off this hook. Besides a general interest in opening Black Boxes, Ashby (and Walter) wanted to open up one particular Black Box, the brain, and it is impossible to avoid a discussion of that specific project here—it was too central to the development of cybernetics.9 I need to observe the following:
Seen from one angle, the tortoise and the homeostat function well as nonmodern ontological theater. These machines interacted with and adapted to their worlds performatively, without any representational detours; their worlds remained unknowable Black Boxes to the machines. This is the picture I want to contemplate. But from another angle, Walter and Ashby remained securely within the space of modern science. As brain scientists, they wanted to open up the brain to our representational understanding by a classically scientific maneuver—building models of its interior. These models were unusual in that they took the form of machines rather than equations on paper, but their impulse was the same: precisely to get inside the Black Box and to illuminate the inner go of the adaptive brain.
What should we make of this? Clearly, this branch of cybernetics was a hybrid of the modern and the nonmodern, staging very different acts of ontological theater depending on the angle one watched them from. I could therefore say that the invitation in what follows is to look at them from the nonmodern angle, since this is the aspect of our imagination most in need of stimulation. But, as we will see in more detail later, it is, in fact, also instructive to look more closely at them fro
m the modern angle too. We can distinguish at least three aspects in which Walter and Ashby's cybernetics in fact departed from the paradigms of modern science.
First, sciences like physics describe a homogeneous field of entities and forces that lacks any outside—a cosmos of point masses interacting via an inverse-square law, say. Cybernetic brain modelling, in contrast, immediately entailed an external other—the unknown world to which the brain adapts. So even if early cybernetic brain models can be placed in a modern lineage, they necessarily carried with them this reference to performative engagement with the unknown, and this is what I will focus on in the following chapters.
Second, we can think not about the outside but about the inside of cybernetic brain models. The tortoise and the homeostat were instances of what theoretical biologist Stuart Kauffman (1971) called "articulation of parts explanation."10 Kauffman's examples of this were taken from work in developmental biology in which one appeals to the properties of single cells, say, to explain morphogenesis at a higher level of cellular aggregation. Ashby's and Walter's brain models had just this quality, integrating atomic parts—valves, capacitors, and so on—to achieve higher-level behavior: adaptation. This is a very different style of explanation from that of modern physics, which aims at a calculable representation of some uniform domain—charged particles responding identically to an electric field, for example. And it is worth noting that articulation of parts explanation immediately thematizes performance. One is more concerned with what entities dothan what they are.Ashby and Walter were not exploring the properties of relays and triodes; they were interested in how they would behave in combination. From this angle, too, cybernetic brain modelling once more dramatized performative engagement, now within the brain.