Circuits of Amnesia

The Logical Calculus of the Nervous Net

In cybernetics, reason and time had long been related, especially in regard to prediction and probability. As is well documented, cybernetics emerged from work at the radiation lab at the Massachusetts Institute of Technology (MIT) on anti-aircraft defense and servo-mechanisms during the Second World War. The MIT mathematician Norbert Wiener, working with the MIT-trained electrical engineer Julian Bigelow, and the physiologist Arturo Rosenblueth, reformulated the problem of shooting down planes in the terms of communication—between an airplane pilot and the anti-aircraft gun. These researchers postulated that under stress airplane pilots would act repetitively and, therefore, display algorithmic behaviors analogous to servo-mechanisms, and amenable to mathematical modeling and analysis. This understanding allowed for all entities to be “black boxed” so as to be studied behaviorally (Rosenblueth, Wiener, & Bigelow, 1943, pp. 18-24).

In 1943, inspired by this idea that machines and minds might be thought together through the language of logic and mathematics, the psychiatrist Warren McCulloch and the logician Walter Pitts at the University of Illinois at Urbana–Champaign, decided to take quite literally the machine-like nature of human beings. The pair would later go to MIT at 1952 at Norbert Wiener’s behest (Kay, 2001). Their article “A Logical Calculus of Ideas Immanent in Nervous Activity,” appearing in the Bulletin of Mathematical Biophysics, has now come to be one of the most commonly referenced pieces in cognitive science, philosophy, and computer science. There is a series of moves, both technical and conceptual, by which neurons could be made equivalent to logic gates, and therefore “thought” made materially realizable from the physiological actions of the brain. Not only did these moves reformulate psychology, but they also demonstrated a broader transformation in the constitution of evidence and truth in science.

The model of the neural net put forth in this article has two characteristics of note that are critical in producing our contemporary ideas of rationality (Abraham, 2002; Kay, 2001). The first claim is that every neuron firing has a “semiotic character”; that is, it may be mathematically rendered as a proposition. ³ To support this claim, Pitts and McCulloch imagined each neuron as operating on an “all or nothing” principle when firing electrical impulses over synaptic separations; that is, the pair interpreted the fact that neurons possess action potentials and delays, as equivalent to the ability to effect a discrete decision. This effect affirms or denies a fact (or activation). From this follows the claim that neurons can be thought of as signs (true/false), and nets as semiotic situations, or communication structures (just like the structured scenarios of communication theory; McCulloch & Pitts, 1943/1970) (Figure 1).

OPEN IN VIEWER

Figure 1.

Mathematical theory of communication.

This discrete decision (true or false, activate or not) also made neurons equivalent to logical propositions and compatible with being though in the terms that define a Turing machine

The second element of the model that is important here is the adoption of a strictly probabilistic and predictive temporality. Neuronal nets are determinate in terms of the future (they are predictive), but indeterminate in terms of the past.

In other words, within this model, given a net in a particular time state (T), one can predict the future action of the net (T + 1), but not the past action (Figure 3). From within the net, one cannot determine which neuron fired to excite the current situation.

McCulloch offered as an example the model (Figure 2) of a circular memory neuron activating itself with its own electrical impulses.

OPEN IN VIEWER

Figure 2.

Images of different states of nets.

OPEN IN VIEWER

Figure 3.

Neural net structures expressing the indeterminacy of the past, as signals may come through a number of different routes.

In such a case, at every moment, what results as a conscious experience of memory is not the recollection of the activation of the neuron, but merely an awareness that it was activated in the past, at an indeterminant time. The firing of a signal, or the suppression of firing, can only be known as declarations of “true” or “false”—true there was an impulse, or false, there was no firing—not an interpretative statement of context or meaning that might motivate such firing.

Within neural nets, at any moment, one cannot know which neuron sent the message, when the message was sent, or whether the message is the result of a new stimulus or merely a misfire. In this model (Figure 4), the net cannot determine with any certitude whether a stimulus comes from outside or from within the circuit; whether it is a fresh input or simply a recycled “memory.” Put another way, from within a net (or network), the boundary between perception and cognition, the separation between interiority and exteriority, and the organization of causal time are in-differentiable. But rather than being a disadvantage for the capacity of a neural net, McCulloch and Pitt’s brilliance was to see this as an advantage.

OPEN IN VIEWER

Figure 4.

Memory neuron.

They end on a triumphant note, announcing an aspiration for a subjective science. “Thus our knowledge,” they wrote, “of the world, including ourselves, is incomplete as to space and indefinite as to time. This ignorance, implicit in all our brains, is the counterpart of the abstraction which renders our knowledge useful” (McCulloch & Pitts, 1943/1970, p. 35). If subjectivity had long been the site of inquiry for the human sciences, now, perhaps, it might—in its very lack of transparency to itself, its incompleteness—become an explicit technology.

McCulloch and Pitts were explicit that their work was a Gedankenexperiment, a thought experiment that produces a way of doing things, a methodological machine. Cheerily, McCulloch admitted that this was an enormous “reduction” of the actual operations of the neurons. ⁴ “But one point must be made clear: neither of us conceives the formal equivalence to be a factual explanation. Per contra!” At no point should anyone assume that neural nets were an exact description of a “real” brain (McCulloch & Pitts, 1943/1970, p. 22). In fact, nets are not representations, they are methodological models, and processes. McCulloch and Pitts discussed this logical reasoning as an experiment, a machine perhaps like the ones described by Deleuze and Guattari in Thousand Plateaus (1987), that does not describe a reality, but rather helps scientists and engineers envision new types of brains and machines, and challenge what scientists thought they knew about how mental processes work.

McCulloch went even further to propose that logic was not reasonable, but actually, psychotic. He proudly announced that the nature of computing is analogous to a psychotic mind:

What we thought we were doing (and I think we succeeded fairly well) was treating the brain as a Turing machine; that is, as a device which could perform the kind of functions which a brain must perform if it is only to go wrong and have a psychosis . . . (von Neumann, 1948/1986, p. 422)

These statements should not, however, be thought in terms of human subjectivity or psychology. Rather, McCulloch was responding to a famous paper delivered by the mathematician John von Neumann on logical automata, where Von Neumann laid out an argument for why computers and brains were not the same, but also the necessary conditions that would need to be created to make them commensurable (von Neumann, 1948/1986).

These circuits now labeled by McCulloch as “psychotic” because they challenged scientific perspective, and reformulated the boundaries of interiority and exteriority, between knowledge and practice. At stake in this statement was the set of methodologies and practices in the human sciences, statistics, and engineering, the epistemology, if we will, that might build new machines—whether organic or mechanical.

It is perhaps no surprise that psychosis might offer the possibility of producing a logic “spoken” directly by nerves, or that it should be related to computational machines and digital mediums. Friedrich Kittler has already suggested that the initial effect of psychoanalysis was the externalization of the psyche and its incorporation into larger discursive networks. In delineating the “discourse network of 1800” from the “discourse network of 1900,” Kittler specifies the latter as being concerned with an obsession with the minute, unimportant, and indiscriminately recorded, which characterized the nascent media technologies of the time (Kittler, 1999, p. 3). Therefore, Freud’s obsessive concern not with the obviously scripted “events,” but with slips of the tongue, minute details, and so forth, advances a larger technical assemblage obsessed with delivering recorded and stored events, detached from any clear referential relation to an external and meaningful “reality” (Kittler, 1990).

But cybernetic invocations of psychoanalysis, and psychosis, complicate the seamless extension of the 1900 discourse network into the present. For Freud found psychotics threatening. The classical definitions of psychosis rest on the subject’s inability to organize time and space. To cite Roger Caillois (1935), there is a mimetic excess to these states; the subject is “consumed” by the environment, unable to delineate the boundaries between the Self and Others. Such proximities—between the doctor and the patient, mind and body, and desire and knowledge—Freud found troubling. For the psychoanalysis of the early 20th century, struggling for credibility under the terms of objectivity offered at the time, psychosis presented a danger to the practice of transference in psychoanalysis, and a threat to the clear-cut separation between the analyst and the analysand. Cyberneticians, however, displaced the relationship between the analyst and analysand entirely. The autonomous circuit can directly speak, thus providing the material (organic or electronic) substrate to language initially sought for in theories of telepathy, occultism, and psychosis. This emerging performativity and materiality of the net was what defined cybernetic methods as separating from earlier histories of psychoanalysis and psychology. However, this new epistemology was only made possible by deferring any encounter with historicity.

Irrespective, therefore, of this models’ relationship to actual neurons, or human memory and intelligence, the pair had established that a capacity for logic and very sophisticated problem solving might emerge from small physiological units such as neurons linked up in circuits. In doing so, and by way of exploiting the amnesia of these circuits, McCulloch and Pitts were able to make neural nets analogous to communication channels, and shift the dominant terms for dealing with human psychology and consciousness to communication, cognition and capacities. Their conception of the neural net informs a change in attitudes to psychological processes that makes visible an epistemological transformation in what constituted truth, reason, and evidence in science.

Let me outline this new epistemology briefly here. It rests on three important points that are seemingly unimportant alone, but are significant when recognized as joining a history of logic, engineering practices, and the human sciences in a new assemblage. The first is that logic is now both material and behavioral, and agents can be an- or non-ontologically defined or “black boxed.” Ontology here is being reductively defined as a concern with being, for cybernetician “what” something is was replaced by the question of what work does an entity or agent do? In this sense, cybernetics, as understood by McCulloch (there are of course many cybernetics), lacked concern with taxonomy, classification, or static well-defined categories. The second is that cybernetic attitudes to mind rest upon a repression of all questions of documentation, indexicality, archiving, learning, and historical temporality. And third, the temporality of the net is preemptive, it always operates in the future perfect tense, but without necessarily defined endpoints or contexts (Halpern, 2005). Nets are about T + 1, the past is indeterminate: McCulloch regularly argued against caring about the actual context, or specific stimulus that incited trauma in patients, or systems. ⁵ Together, these points meant that rationality could be redefined as both embodied and affective, and good science was not the production of certitude but rather the account of chance and indeterminacy.

Control and Computing

Having inserted the logic of the machine into the brain, this model was then fed back into the design of machines. The model of the cycling memory neuron in fact directly refracts an earlier concept of control in the Turing machine (and would later become the model for memory in von Neumann’s architecture for EDVAC (( E lectronic D iscrete V ariable A utomatic C omputer); von Neumann, 1945). Control in the Turing machine is the head that “reads” the program from memory and then begins the process of executing it according to the directions in the memory. Control, therefore, directs the next operation of the machine. Control is directed by the program. The control unit, or the reading head in a Turing machine, is directed by the tape it is reading from memory, not reverse. Control is that function that will read and act upon these retrieved data, inserting the retrieved program or data into the run of the machine. Such machines do not operate top-down, but rather in feedback loops between storage, processing segments, and the interface for input and output. In his 1946 report on building a computing machine, the ACE Report, Turing reiterated that only the possession of memory

give[s] the machine the possibility of constructing its own orders; i.e. there is always the possibility of taking a particular minor cycle out of storage and treating it as an order to be carried out. This can be very powerful. (A. M. Turing, 1986, p. 21)

If there is a feature that allows computers to act in uncanny and unexpected ways, it is this surprising capacity to change the pattern of action by way of insertion of a program from storage. Control in computers is like reverberating circuits in brains, and both are classically defined, in psychoanalysis, as psychotic insofar as they involve receiving memories without history.

The historical redefinition of rationality demanded therefore a reconsideration of what “control” might mean. In most contemporary scholarship, control was correlated with prediction, knowing the future, and command (often military; Edwards, 1997; Galison, 1994). In their famous book, Planning and Coding, von Neumann and Goldstine introduced flow charts and circuits for stored program computers. However, in describing their circuits, they wrote, “[w]e propose to indicate these portions of the flow diagram of C by a symbolism of lines oriented by arrows [see Figure 5] . . . Second, it is clear that this notation is incomplete and unsatisfactory . . .”(Goldstine & Neumann, 1948, p. 157). In other words, control is not definable; its operable imagining and its explicit definition are incommensurate. But rather than treat this failure in representation as a problem, this threshold became a technological opportunity; this emergent space between the definable and the infinite provided the contours of the engineering problem—an opportunity to turn from logic to technology.

OPEN IN VIEWER

Figure 5.

Diagrams of “c.”

Significantly, for us, McCulloch and Pitts inverted the problem posed by the original negative proof of the entscheidungsproblem that is the Turing machine. If throughout the 19th and earlier 20th centuries an army of mathematicians and philosophers struggled to infinitely extend the limits of logical representation to which the Turing machine is a negative proof demonstrating the impossibility of fully representing all statements in first-order logic, then McCulloch and Pitts had a different epistemology and frame (Gödel, 1962; Goldstine & Neumann, 1948; Reck, 2002; Russell, 2009; A. Turing, 1936). Accepting that there were many things that could not be known or computed, McCulloch inverted the question Turing had posed. If instead of seeking an absolute reasonable foundation for mathematical thought that an army of other logicians and mathematicians, including Gottlob Frege, Kurt Gödel, David Hilbert, Bertrand Russell, Alfred North Whitehead, and Alan Turing, had attempted, McCulloch and Pitts chose to ask instead: What if mental functioning could now be demonstrated to emanate from the physiological actions of multitudes of logic gates? What could be built? Not: What could be proven? The problem could be inverted from seeking the limits of calculation to examining the possibilities for logical nets. What had been an absolute limit to mathematical logic became an extendable threshold for engineering. McCulloch implied we should turn instead to accepting our partial and incomplete perspectives, our inability to know ourselves and make this “psychosis” in his words an “experimental epistemology” (McCulloch, 1965, p. 359).

Affective Logics

What the cybernetic reformulation of logic as “psychotic” permitted was an abandonment of ontological concern with the past and the present in the interest of focusing on future interactions. These models measured not what is happening, but prepare us for what will happen as a result of finding patterns of past data, which ironically is devoid of historical temporalities. While such a-historicity had perhaps long been a problem for statisticians, in cybernetics, problems of causality were repressed by creating models that did not engage with historicity or linearity; this repression of historicity, however, as I hope to show, will come back to create new challenges for the sciences of communication and society now framed in terms of storage and information. The transformation in truth claims and epistemology opened a new frontier for study—subjective interactions in environments with incomplete information.

These nervous networks and logical rationalities proliferated in the social and human sciences. Cybernetic and communicative concepts of mind were part of a broader shift at the time in concepts of reason, psychology, and consciousness; informing everything from finance and options trading equations, to environmental psychology and urban planning programs of individuals such as Kevin Lynch, and later MIT’s Architecture Machine Group and the Media Lab headed by Nicholas Negroponte, to the political science models of Karl Deutsch at Harvard, and the “bounded rationality” introduced by Herbert Simon and widely considered the foundation of contemporary finance. The post-war social sciences were repositories of these techniques that transformed what had once been a question of political economy, value production, and the organization of human desire and social relations to problems of circulation and communication by way of a new approach to modeling intelligence and agency (Crowther-Heyck, 2005; Dumit, Circuits in the mind, personal communication/lecture, May, 20, 2007; Halpern, 2015; H. Simon, 1955; H. A. Simon, Egidi, Viale, & Marris, 1992).

This rationality is also sensible, perhaps affective—a situation that puts in considerable revision dominant understandings of digital and computational mediums as distancing, disembodied, or abstract. And if it is one of the dominant assumptions in the study of modern history and governance that liberal subjectivity and economic agency are defined as a logic guided by a reason separate from sense, then these discourses mark a clear contrast. The historian of science Lorraine Daston reminds us that we would do well to recall that those things today considered virtuous and intelligent, such as speed, logic, and definitiveness in action, were not always so. She is explicit: Rationality in its Cold War formulation, despite the insistence of technocrats, policy makers, and free-market advocating economists, is not reason as understood by Enlightenment thinkers, liberals, or even modern logicians (Daston, 2011; MacKenzie, 2006; Mirowski, 2002).

If this is true, then our financial instruments, markets, governments, organizations, and machines are rational, affective, sensible, and preemptive, but not reasonable. To recognize the significance of this thinking in our present, it might help to contemplate Brian Massumi’s definition of “preemption.” Preemption, he argues, is not prevention; it is a different way of knowing the world. Prevention, he claims, “assumes an ability to assess threats empirically and identify their causes.” Preemption, however, is affective; it lacks representation and is a constant nervous anticipation, at a literally neural if not molecular level, for a never fully articulated threat or future (Massumi, 2007, p. 4).

Cyberneticians, within 10 years from the war, moved from working on anti-aircraft prediction to building systems without clear endpoints or goals, and embracing an epistemology without final objectives, or perhaps objectivity (even if many practitioners denied this). Nets, taken as systems, are probabilistic scenarios, with multiple states and indefinite run times even if each separate neuron can act definitively. In cognitive and early neuro-science, the forms of knowledge being espoused were always framed in terms of experiment, never definitive conclusions. “Experimental epistemologies,” as McCulloch put it, came to mean that there are never final facts, only ongoing experiments.

These human and social scientists made operative the unknowable space between legibility and emergence, and turned it into a technological impulse to proliferate new tools of measurement, diagrams, and interfaces. At the limits of this analysis is the possibility that emergence itself has been automated. As the theorist Luciana Parisi puts it, cybernetics takes hold of the space between infinity and logic, and makes it the very site of technical intervention, the very site to proliferate algorithms into life (Parisi, 2013). If cybernetics initially sought to control the future, now control itself became the unclear site of emergence, an indefinable state that was part of networks operating in the future without full definition or information either about endpoints or pasts. The problem of how to act under conditions of uncertainty, or how to define a man or a machine, became instead a pragmatic mandate and a focus on process. Instead of asking what is a circuit, a neuron, or a market, human scientists turned to asking what do circuits do? How do agents act? Creating an ongoing opportunity to entangle calculation and life at the level of nervous networks, by correlating the nervous system with the financial and political system.

Memory as a Cyclical Machine

Having supposedly exorcised the ghosts of historicity, cyberneticians, however, continued to struggle with memory and signification. In a 1952 letter to the cybernetician Norbert Wiener, Gregory Bateson spelled out the problem of memory, time, repetition, and rationality:

What applications of the theory of games do, is to reinforce the players’ acceptance of the rules and competitive premises, and therefore make it more and more difficult for the players to conceive that there might be other ways of meeting and dealing with each other. . . . I question the wisdom of the static theory as a basis for action in a human world. The theory may be “static” within itself, but its use propagates changes, and I suspect that the long-term changes so propagated are in a paranoidal direction and odious. (Bateson, 1952, p. 2)

Discussing the premier private consulting group to the U.S. government and military on national security and public policy—the RAND Corporation—Bateson makes explicit a new dilemma: violence. In this formulation, players no longer create violence because of a misdirected desire resulting in a loathing for an imagined Other, but instead, they are led to produce violence through a self-referential performance within the game. Bateson correlates “static” games with paranoid schizophrenics, as a perceptual problem resulting in repetitive cycles culminating in potentially genocidal violence (nuclear war in this case)—in his language, a “paranoidal direction.” Authority emerging from the pure self-reference of the data field is psychotic and comes at the expense of futurity. Bateson fears that the performance of past data paraded as prophecy will produce only repetition without difference. In a stunning inversion of psychoanalytic concerns, Bateson recognizes that the ubiquity of computational logics makes distance impossible to achieve, and induces violence, not as a result of any misdirected object choices or imagined enemy Others—game theories have no such formulations within them—but as the result of performing and repeating commands without interpretation. In fact, it is precisely the lack of imagination that defines this condition. Bateson foresees a total war without desire.

Having displaced older terms of consciousness, reason, and desire from the algorithmic rationality of the network, these terms would return in cybernetics under the guise of visualization, time, and memory. At the famous Sixth Macy Conference on Circular and Causal Feedback Mechanisms in Biological and Social Systems in 1949 in New York City, memory was increasingly problematized in terms of the relations between its dynamic and stable elements and storage. In this instance, the immediacy and temporality of the televisual came to replace the older conceptions of tapes, photographs, and films. McCulloch opened the conference with a beacon and a warning. He offered the example of a new type of tube, in development at Princeton, similar to a cathode ray tube, that beams onto a screen on which items are stored. The persistence of the “memory” of the beam is temporary and must be refreshed. This idea of a cycling, or scanning memory, McCulloch viewed as offering the possibility of miniaturizing and expanding machine memory (Pias, 2003).

His second example was a warning from John von Neumann: Even the entire number of neurons in the brain, according to calculation, could not account for the complexity of human behavior and ability. McCulloch reported the finding that “the performance of the army ant . . . is far more complicated than can be computed by 300 yes or no devices” (Pias, 2003, p. 31). But this was not to say that these capacities need be understood as illogical or analog. Rather, McCulloch turned to another model that might retain the logical nature of the neurons, but still account for the capacity to learn, and behave at scales beyond the comprehension of computation.

The answer, coming through a range of discussions about protein structure and memory within cells, involved refreshing information in time. Wiener argued, “this variability in time here postulated will do in fact the sort of thing that von Neumann wants, that is, the variability need not be fixed as variability in space, but may actually be a variability in time.” (Pias, 2003, p. 35). The psychologist John Stroud offered the example of a “very large macro-organism called a destroyer.” This military ship has endless “metabolic” changes of small chores throughout the day, but still retains the function of a destroyer. This systemic stability, but internal differentiation and cycling, became the ideal of agency and action in memory

McCulloch and Stroud went on to present understandings of memory in terms of an opposition between perfect retention of all information with retroactive selection or memory as a constant active site of processing of information for further action, based on internal “reflectors,” or “internal eyes.” “We may,” Stroud stated,

need only very tiny little reflectors which somehow or other can become a stimulus pattern which is available for this particular mode of operation of our very ordinary thinking, seeing, and hearing machinery. This particular pattern of reflectors is what I see as it were with my internal eyes just as what I see when I look at a store window, is a pattern on the retinal mosaic. (Pias, 2003, p. 121)

Mental processes are equated here with processing data, and pattern seeking, but it is internal “eyes” from within the psychic apparatus that allow a self-reflexive apparatus for deferring decisions and agency. Mind emerges from multiple time systems operating between the real-time present of reception and circulating data, and memory in time, a cyclical “refreshing” as in a television screen system, where change, and differentiation—between the organism and the environment, between networks—becomes possible through the delay and reorganization of circuits from within the organism. The problems of computational representation, the initial problems that were faced in mathematically and logically representing intelligence, were reorganized away from a language of conscious and unconscious, discrete and infinite, reason and psychosis to the new terms of vacillating temporalities between immediacy and reflexivity.

Bateson, also an attendant at the aforementioned conferences and one of the founders of family therapy and addiction treatment programs, offered one of the more compelling models and practices for rethinking mind in his use of a model of the “double bind” to explain psychic suffering, addiction and other maladjusted and compulsive behaviors. In a conference in 1969, at the National Institute of Health, he offered an example to demonstrate his ideas of both psychology and treatment. He discussed a research project conducted with porpoises trained at Navy research facilities to perform tricks and other trained acts in return for fish. One day, he recounted, one of the porpoises was introduced to a new regimen. Her trainers deprived her of food if she repeated the same trick. Starved if she repeated the same act, but also if she did not perform, the porpoise was caught in a double bind. This experiment was repeated with numerous porpoises, usually culminating in extreme aggression, and a descent into what from an anthropomorphic perspective might be labeled disaffection, confusion, and anti-social and violent behavior. Bateson with his usual lack of reservation was ready to label these dolphins as suffering a paranoid form of schizophrenia. The anthropologist was at pains, however, to remind his audience that these psychotic animals were acting rationally. In fact, they were doing exactly what their training as animals in a navy laboratory would lead them to do. Their problem was that they had two conflicting signals. The poor animals, having no perspective on their situation as laboratory experiments were naturally breaking apart, fissuring their personalities (and Bateson thought they had them) in efforts to be both rebellious and compliant, but above all to act as they had been taught. Bateson argued this was the standard condition for humans in contemporary societies.

Having established the mechanisms that led to a decentered and multiple subject, Bateson commenced to articulate the dangers and possibilities of this condition. He recalled how, between the 14th and 15th time of demonstration, one of the porpoises “appeared much excited” and for her final performance gave an “elaborate” display, including multiple pieces of behavior of which four were “entirely new—never before observed in this species of animal” (Bateson, 2000, p. 278). These were not solely genetically endowed abilities, then, but were learned, the result of an experiment in time. This process in which the subject, whether a patient or a dolphin, uses the memories of other interactions and other situations to transform their actions within the immediate scenario was represented as the site of innovation. The dolphin’s ego (insofar as we decide she has one) was sufficiently weakened to develop new attachments to objects in its environment through the memories of its past and of other types of encounters. This rewired network of relations was what was held to lead to emergence through the re-contextualization of the situation within which the confused and conflicted animal found itself.

Despite the suffering of this poor animal, Bateson ends on a positive note, having now successfully made the psyche inter-subjective and simultaneously amenable to technical appropriation via family therapy (Bateson, 2000, p. 278). The productivity of a schizoid situation rested for Bateson on the discovery made by both communication theory and physics that different times could not communicate directly to one another. Only temporal differences resist circulation from within the definition of communication that was being put forward here. Bateson applied this understanding liberally to animals. In cybernetic models, the ability of an entity to differentiate itself from its environment and make autonomous choices is contingent on its ability to simultaneously engage in dangerous spatial proximities, for example, between the porpoise and her trainer, with other entities and the ability to achieve distance from them in time.

At stake in the negotiation over the nature of networks and the time-scale of analysis was nothing less than how to encounter difference—whether between individuals, value in markets, or between vast states during the Cold War. A question that perhaps started in psychoanalytic concerns over psychosis found technical realization in cybernetics. For cyberneticians, the problem of analog or digital, otherwise understood as the limits between discrete logic and infinity, the separation between the calculable and the incalculable, the representable and the non-representable, and the differences between subjects and objects, was transformed into a reconfiguration of memory and storage—a transformation that continues to inform our multiplying fantasies of real-time analytics while massive data storage infrastructures are erected to insure the permanence, and recyclability, of data.

While the time of neural nets and communication theories is always preemptive as invoked by Massumi, the shadow archive haunting the speculative network is one of an endless data repository whose arrangement and visualization might return imagination and agency to subjects. These wavering interactions—between the networked individual and the fetish of data—preoccupy us in the present, speaking through our contemporary concerns with data mining, search engines, and connectivity. The relationship between rationality and control drives the ongoing penetration and application of media technologies as the result of an imperative to seek consciousness through better visualization and collective intelligence through the collaboration of many logical, but hardly reasonable, agents. Architecturally these dual desires incarnate themselves in a proliferation of interfaces and a fetish for visualization and interactivity, merged with an obsession to amass and store data in huge systems of data centers and server farms. What had first been articulated as a problem of memory and time has now become a compulsion for analytics.

Theorizing the Nervous Network

I opened this essay arguing that cybernetics and its affiliated communication and human sciences aspired to the elimination of difference in the name of rationality, a dream of self-organizing systems and autopoietic intelligences produced from the minute actions of small, stupid, logic gates, a dream of a world of networks without limit, focused eternally on an indefinite, and extendable, future state.

Earlier in this essay, I also invoked Bateson’s concerns about self-referential violence. What Bateson articulated was the worry that in the real-time obsession to entangle life with calculative logics, learning, and by extension thought, would itself be automated in such a way as to lead to violent harm, and perhaps, the destruction of the world.

This condition only becomes inevitable, however, if we ourselves descend into the logic of immediate and real-time analytics. We must avoid this conclusion, and this condition. Like Bateson’s porpoise, torn between reactionary return, and self-referentiality, we are forced to ask about the other possibilities that still lie inside our machines and our histories. The cycles of the porpoise reenact the telling of cybernetic history where ideas of control and rationality are often over-determined in their negative valence, and the inevitability of the past to determine the future is regularly assumed. Perhaps the hope is in the very machinery that was imagined into being through the cybernetic circuits—systems that can both recognize and disavow their history, for which memory and archiving remain at tense and productively incommensurable distances, incarnated in a drive to accelerate the speed of speculation while intensifying the infrastructure for data gathering and storage, all supported by a nervous rationality that is constantly multiplying from within.