2.11 Early Computing Influence

Early computing influence refers to the profound effect that the first generation of digital computers—developed between the late 1930s and the 1950s—had on the formation of cybernetic communication theory. The emergence of programmable digital computation provided cybernetics with its most powerful analogy, its most important intellectual ally in John von Neumann, and its most consequential long-term institutional connection: the convergence of computing and communication research that would reshape both fields over the following decades.

The Digital Computer as Conceptual Model

The digital computer made its conceptual contribution to cybernetics at a specific, historically early moment: before computers were common, when their logical architecture was still being designed and their philosophical significance was still being worked out. The question "What is a computer, fundamentally?" was being answered simultaneously by Turing, von Neumann, and Wiener in the late 1930s and 1940s, and the answers they gave shaped the conceptual vocabulary of cybernetics.

Alan Turing's 1936 paper "On Computable Numbers" had shown that any computation that could be precisely specified could in principle be performed by a simple abstract machine—the Turing machine—operating on a tape according to a finite set of rules. This result established that computation is substrate-independent: the same computation can be realized in many different physical implementations. Turing's universal machine—a Turing machine that could simulate any other Turing machine when given the latter's description as input—provided the conceptual model of a general-purpose device that could process any information according to any specifiable procedure.

For cybernetics, the Turing machine suggested that information processing—including the information processing of the brain—might be analyzable in terms of formal computation. If brains process information and computers process information, and if all information processing can be reduced to computation in Turing's sense, then the formal analysis of computation might illuminate the formal analysis of cognition. This reasoning motivated the computational theory of mind and, in the communication domain, motivated models of communication as information processing.

Von Neumann Architecture and the Brain

The most influential conceptual bridge between early computers and cybernetic communication theory was the work of John von Neumann, who was the central intellectual figure connecting the two traditions. Von Neumann contributed to both the mathematical foundations of cybernetics (he was a regular participant in the Macy Conferences) and the practical design of early computers (his 1945 "First Draft of a Report on the EDVAC" established the stored-program computer architecture that became standard).

Von Neumann was intensely interested in the analogy between computers and brains. His Yale lectures, delivered in 1956 and published posthumously as The Computer and the Brain (1958), systematically compared the logical and statistical properties of digital computers and neural networks:

• Digital vs. analog processing: Computers process information digitally (discrete binary values); neurons appeared to process information both digitally (action potential: fire or don't fire) and through graded analog signals. Von Neumann analyzed both modes and their implications for reliability and capacity.
• Reliability through redundancy: Both computers and brains must achieve reliable computation from unreliable components. Computer designers achieve reliability through careful error control; brains achieve reliability through massive redundancy—many parallel neurons doing similar computations, with the outputs averaged or voted.
• Memory architecture: Von Neumann distinguished between the computer's active processing units and its memory stores, noting that the brain's organization of memory—distributed, associative, and highly parallel—differed radically from the serial, addressable memory of early computers.
• Processing speed vs. parallel organization: Early computers were fast but sequential; brains were slow (neurons fire at most a few hundred times per second) but massively parallel. Von Neumann computed that the brain's parallel architecture gave it overall computational capacity comparable to or exceeding contemporary computers despite the neuronal speed disadvantage.

These comparisons were not merely analogical: they motivated specific theoretical investigations and empirical research programs. If brains and computers are both information-processing systems, then insights from either domain illuminate the other. The computer provided a working existence proof that information processing could be implemented in a physical substrate—and suggested that the brain's implementation might be analyzable in similar terms.

McCulloch and Pitts: Neural Networks as Computing Devices

The most direct connection between early computing concepts and communication neuroscience was the 1943 paper by Warren McCulloch and Walter Pitts, "A Logical Calculus of the Ideas Immanent in Nervous Activity." McCulloch was a neurophysiologist and Pitts a mathematical prodigy; their collaboration produced a formal model of neural computation that was simultaneously a model of logical computation.

McCulloch and Pitts modeled neurons as threshold logic units: each neuron sums its inputs and fires (produces an output of 1) if and only if the sum exceeds a threshold. They proved that networks of such threshold units could compute any logical function—any Boolean combination of inputs. This established that neural networks, understood as logical computing devices, were in principle computationally universal: they could implement any computation that could be specified in logical terms.

The McCulloch-Pitts model was influential for early computing in both directions:

• It suggested that the brain could be understood as a computing device, modeled using the logic of Boolean algebra.
• It suggested that computing devices organized as networks of threshold elements might exhibit cognitive capabilities, motivating early AI and neural network research.

For cybernetic communication theory, the McCulloch-Pitts model established that information processing in neural systems—including the processing that underlies perception, decision-making, and communication—could be analyzed using formal logical and mathematical methods continuous with those used to analyze engineered computing systems.

Stored-Program Computers and Memory in Communication

The stored-program computer—in which the program (instructions) is stored in the same memory as the data, allowing programs to be modified by the computation itself—was the central architectural innovation of early computing. This architecture, outlined by von Neumann and implemented in machines like the ENIAC (1945), EDSAC (1949), and IAS computer (1951), established several concepts that fed directly into cybernetic communication theory:

Memory as an active communication resource: In stored-program computers, memory is not merely a passive storage medium but an active participant in computation—programs are read from memory, modified, and written back. This dynamic view of memory as part of the communication process influenced cybernetic models of the relationship between memory and ongoing communication.

The separation of hardware and software: The fact that the same physical computer could run different programs—and that programs could be written, debugged, and modified independently of the hardware—established the distinction between the physical substrate of communication and the informational structure carried by that substrate. This distinction became fundamental in information theory (the distinction between the channel and the message) and in cybernetic communication theory more broadly.

Recursive computation: Stored-program computers can execute programs that generate other programs, that modify themselves, or that simulate other computers. This recursive capacity suggested that communication systems capable of representing their own structure and function—self-referential systems—were not a logical impossibility but a realizable engineering objective. This influenced second-order cybernetics and its interest in self-referential and reflexive communication.

Automata Theory and Communication Complexity

The theoretical study of automata—abstract computing devices with finite or infinite memory—developed in close connection with early computing and fed directly into cybernetic models of communication:

Finite automata are computing devices with a finite number of internal states, which read inputs and transition between states according to transition rules. They can be understood as models of simple communication systems in which the system's response to each input depends on its current internal state (context). The study of what languages finite automata can recognize—and what they cannot—defined the boundaries of context-free communication versus context-dependent communication.

Turing machines as universal computing devices established the limits of what any computational communication system can do: there exist well-defined computations that no Turing machine can perform (the halting problem), establishing fundamental limits on self-knowledge and prediction within formal systems.

Probabilistic automata: Extensions of automata theory to probabilistic transition rules connected formal computing theory directly to Shannon's probabilistic information theory, enabling the analysis of how memory and computational structure interact with information content and channel capacity.

Early AI and Communication Modeling

The earliest artificial intelligence research, emerging in the mid-1950s from figures like John McCarthy, Marvin Minsky, Allen Newell, and Herbert Simon, drew on the confluence of cybernetics and early computing to model cognitive and communicative processes:

Language processing: Early natural language processing programs attempted to model human communication by implementing formal grammatical rules in computer programs. These efforts revealed the complexity of syntactic parsing and semantic interpretation, driving theoretical linguistics (Chomsky's formal language theory) and computational linguistics.

Problem-solving and planning: Newell and Simon's General Problem Solver modeled cognitive problem-solving as a search process through a problem space—a computational model of purposive communication in which the agent uses feedback (current state vs. goal state) to guide sequential problem-solving actions.

Simulation of human communication: Some early AI researchers attempted to model specific aspects of human communication—question answering, dialogue, story understanding—as computational processes, creating formal models that connected engineering computation directly to the analysis of human communication.

Long-Term Convergence: Computing and Communication Theory

The influence of early computing on cybernetic communication theory was not just conceptual but institutional: the convergence of computing and communication as fields created shared research programs, shared funding sources, and shared academic institutions that continue to shape both fields.

The ARPANET (1969)—precursor to the Internet—was the product of research that explicitly combined communication theory with computing: distributed computing systems communicating over packet-switched networks required both Shannon's information theory (for channel capacity analysis) and computer science (for protocol design and network routing). The Internet as a communication system is a direct institutional descendant of the convergence of early computing and communication theory in the cybernetic tradition.

The influence of early computing on cybernetic communication theory established a template for thinking about communication that persists: communication as information processing, communicating systems as computing devices, and the formal analysis of communication as continuous with the formal analysis of computation. This template has been enormously productive but has also been criticized for reducing the richness of human communicative experience—meaning, relationship, power, culture—to the formal operations of a computing machine.