Norbert Wiener

THE CLAIM

Norbert Wiener gave regulation a name and a mathematics

In 1948 Norbert Wiener published Cybernetics: Or Control and Communication in the Animal and the Machine (Wiener 1948), and self-regulation stopped being a figure of speech. Before that book, saying that a body governs itself was a description. After it, it was a set of equations with terms you could measure. Wiener built the name from the Greek kybernetes, the steersman, and in the introduction he explained the choice. He wanted the word to acknowledge that the first significant paper on feedback mechanisms was Clerk Maxwell on governors, published in 1868 (Maxwell 1868), and that the English word governor descends through a Latin corruption from the same Greek root. The name carries its own family tree.

Ask what the steersman image commits you to. If steering is the model, the interesting quantity in a living system is not the output. It is the error. A steersman does not hold the wheel at a fixed angle and hope. He holds a heading, reads how far he has drifted off it, and moves the wheel by whatever amount the sea demands this second. The heading is fixed. The correction is never fixed. That distinction is the whole of cybernetics, and it is why a mathematician belongs in a history of tone. Wiener supplied a formal grammar for something the healing arts had been describing in ordinary language for two thousand years. A living body does not simply run. It keeps correcting itself while it runs, and the quality of that correcting is the thing worth measuring.

1941 TO 1943

An anti-aircraft gun taught Wiener what a nervous system does

Cybernetics began as a gunnery problem, not a biological one. From 1941 to 1943 Wiener worked for the National Defense Research Committee on the automatic aiming of anti-aircraft guns, alongside the engineer Julian Bigelow. The problem is brutal and simple. An aircraft moves. A shell takes seconds to arrive. To hit anything you must aim not where the plane is but where it will be, and you must do it from a noisy radar trace. Wiener stated the task in one line: to predict the future of a curve is to carry out a certain operation on its past. The mathematics he built there became the Wiener filter, written up as a classified report in 1942 and declassified in 1949.

The war work matters here for what it forced him to notice. A gun director that corrects its own aim is a servomechanism, and Wiener and Bigelow found that theirs failed in a very particular way when they turned its correction strength too high. It did not go dead. It did not go slack. It began to hunt, swinging past the target and back again, each swing wider than the last, until it was useless. The failure sat in neither the motor nor the sensor. It sat in the relationship between them. Wiener carried that observation across the river to a physiologist, and the way we describe the nervous system changed permanently.

SEVEN PAGES, 1943

Purpose was redefined as negative feedback in seven printed pages

The founding document of cybernetics is not the 1948 book. It is a paper of seven pages: Arturo Rosenblueth, Norbert Wiener and Julian Bigelow, Behavior, Purpose and Teleology, Philosophy of Science, volume 10, issue 1, January 1943, pages 18 to 24 (Rosenblueth et al. 1943). The authors sign off from Harvard Medical School and the Massachusetts Institute of Technology. Rosenblueth was Walter Cannon's collaborator and a physiologist of the heart. Bigelow was an engineer. Wiener was a mathematician. Three fields, one page limit, one argument.

Their move was to split the word feedback in two. In the broad engineering sense, feedback returns some of the output energy to the input, and when the returned fraction carries the same sign as the original signal it is positive: it adds to the input, it does not correct it. In the restricted sense, the behavior of an object is controlled by the margin of error at which it stands with reference to a specific goal, and the returned signals restrict outputs that would otherwise overshoot. That second sense is negative feedback. There is a second line on page 19 worth carrying out of the paper as well. When we perform a voluntary action, the authors write, what we select voluntarily is a specific purpose, not a specific movement. We do not command muscles. We trip a purpose and the reaction follows.

THE CEREBELLUM

A shaking hand and an oscillating machine are the same equation

On page 20 of the 1943 paper the three authors looked at a machine hunting around its target and said in print that it looked like a patient. They had described a machine built to home on a moving light whose feedback is inadequately damped, so that each correction overshoots and provokes a still larger correction in the other direction, until a series of increasingly larger oscillations makes the machine miss the goal entirely. Then they turned to the clinic. A patient with cerebellar disease shows no obvious motor disturbance at rest. Ask that same patient to carry a glass of water from a table to his mouth and the hand begins to oscillate, wider and wider, precisely as the glass approaches the target.

From that resemblance they proposed something no anatomist had proposed: that the main function of the cerebellum is the control of the feedback nervous mechanisms involved in purposeful motor activity. Notice the form of the argument, because it is the form this whole section of the history runs on. It is not a lesion study. It is a signature match. Underdamped control has a fingerprint, which is oscillation that grows as the target is approached, and the authors found that fingerprint on a human hand. When two systems built of entirely different material fail in the same shape, the failure belongs to the organization and not to the material. That single principle is why tone can be discussed across tissues, across species and across scales without the argument going soft.

THE LOOP CLOSES

The nervous system stopped being a one way street in 1948

The most consequential sentence Wiener wrote for this history sits in the introduction to Cybernetics, and it is an act of demolition. The teaching diagram of the nervous system in 1948 ran left to right: receptor, afferent nerve, cord, brain, efferent nerve, muscle. Input at one end, output at the other, the brain in the middle as a switchboard. Wiener took that arrow and bent it into a ring. Muscle is not the end of the chain. Muscle is a sense organ that also moves, and every contraction it makes is reported straight back into the system that ordered it, through proprioceptors and through the special senses alike.

Ask what that does to the idea of a reflex. If the loop is closed, there is no such thing as a stimulus arriving at a neutral nervous system. Every incoming signal lands on a system already carrying the trace of what it just did. State comes first. The response is a function of the input and of the standing condition of the loop receiving it, which is why identical inputs produce different outputs in the same person on different days, and why two people with matching images can present with completely different bodies. This is the formal skeleton under everything the tone model asserts. Tone is that standing condition, expressed as tissue tension, readable at every scale from a single motor unit to a whole posture.

PROPRIOCEPTION

Ataxia is what a body does when the return signal goes quiet

Wiener's chosen test case for feedback in a person was the smallest act he could name: picking up a pencil. He worked through what the act actually requires. There must be a report to the nervous system, conscious or unconscious, of the amount by which we have failed to pick up the pencil at each instant. Take that report away and the act collapses. If the proprioceptive sensations are wanting, Wiener wrote, and we do not replace them by a visual or other substitute, we are unable to perform the act and find ourselves in the state known as ataxia.

That sentence holds a clinical fact worth keeping. The muscle in an ataxic limb can be strong. The nerve to it can conduct. The joint can be free. What is missing is information, and the loss of information alone is enough to destroy the movement. A patient who walks well while watching his own feet and falls the moment the lights go out is not demonstrating a weak leg. He is demonstrating a loop running on a borrowed channel. Strength is not the variable under test. Regulation is the variable. Any approach to a human body that measures only force and range of motion is reading one side of an equation and calling it the answer.

TONIC

Wiener's own word for the state he was chasing was tonic

Wiener spent years on heart muscle, and the word he reached for was tone. In the introduction to Cybernetics he describes the joint program with Rosenblueth as concerning the relation between the tonic, clonic and phasic contractions in epilepsy on one side, and the tonic spasm, beat and fibrillation of the heart on the other. They treated heart muscle as an irritable tissue conducting impulses through a network of connected excitable elements. The mathematics appeared as Wiener and Rosenblueth, The Mathematical Formulation of the Problem of Conduction of Impulses in a Network of Connected Excitable Elements, Specifically in Cardiac Muscle, in the Archivos del Instituto de Cardiologia de Mexico, volume 16, pages 205 to 265, 1946 (Wiener and Rosenblueth 1946).

Fibrillation is the cleanest demonstration of tone that medicine owns. A fibrillating heart is not a weak heart. Every fiber in it is contracting, and it is burning fuel at a furious rate while moving almost no blood. What has failed is the phase relationship between the fibers. Same tissue, same fuel, same contractile protein, opposite outcome, and the only variable that changed is the timing of the conversation between the parts. Wiener modeled that failure as a property of the network rather than of the cell. Hold it next to a spine, a diaphragm or a pelvic floor and the point transfers without strain. Coordination is a state. States can be well tuned or badly tuned. A badly tuned state can look like tremendous effort while producing nothing at all.

TEN CONFERENCES

Ten Macy conferences built one vocabulary across a dozen fields

Between 21 March 1946 and 24 April 1953 the Josiah Macy Jr. Foundation hosted ten conferences on feedback, chaired by the neurophysiologist Warren McCulloch (Heims 1991). From the seventh meeting in 1950 the series carried the official title Cybernetics: Circular Causal and Feedback Mechanisms in Biological and Social Systems. The room held McCulloch and Walter Pitts, who had modeled the neuron as a logical element (McCulloch and Pitts 1943); John von Neumann and Claude Shannon; the anthropologists Margaret Mead and Gregory Bateson; the psychiatrist W. Ross Ashby; Rosenblueth; and Wiener himself. Wiener records in his introduction that McCulloch made the arrangements with the Macy Foundation in the spring of 1946 for the first of the series.

What came out of those rooms was not a theory owned by one discipline. It was a shared vocabulary that survives translation. Loop, gain, delay, damping, error, set point, oscillation. A physiologist and an electrical engineer could finally disagree with each other using the same nouns, which is the precondition for making progress rather than noise. That is the example this section keeps returning to. The healing arts have spent a long time defending vocabularies from one another. The Macy rooms show what happens when a common language is built on purpose instead. Nobody had to surrender their subject matter, and every subject came out sharper than it went in.

THE NO MAN'S LAND

Wiener worked on purpose in the gaps between established fields

Wiener stated his working principle plainly. For many years he and Rosenblueth had shared the conviction that the most fruitful areas for the growth of the sciences were those which had been neglected as a no man's land between the various established fields. He did not treat that as a poetic sentiment. He treated it as a research strategy, and he ran it out of Rosenblueth's monthly discussion meetings on scientific method at Harvard Medical School, in the circle of Walter B. Cannon, the physiologist who had named homeostasis in the previous generation. Rosenblueth later moved to the Instituto Nacional de Cardiologia in Mexico, and the collaboration continued across a border.

He was equally plain about why the strategy is hard. If a physiologist who knows no mathematics works together with a mathematician who knows no physiology, Wiener wrote, the first will be unable to state his problem in terms the second can manipulate, and the second will be unable to put the answer in any form the first can understand. The remedy is not a summary and not a translator. The remedy is people genuinely at home in more than one territory, working together long enough to build a common idiom. That is expensive, slow and unfashionable, and it produced one of the most portable ideas in twentieth century biology.

CORRECTIONS

Wiener and Rosenblueth settled the technical sense of cybernetics in 1947, and Rosenblueth led the 1943 paper

First, Wiener did not invent the word cybernetics. The Greek kybernetike appears in Plato as a metaphor for governing, and the French physicist Andre-Marie Ampere used cybernetique in 1834 for the sciences of government in his classification of human knowledge (Ampere 1834). Wiener and Rosenblueth settled on the modern technical sense in the summer of 1947, more than a century later. Coining the technical meaning and supplying its mathematics is a large enough achievement without inflating it into inventing the word. Second, he did not coin the word feedback either. The 1943 paper says outright that engineers already used the expression in two different senses. What the three authors contributed was the argument that the restricted, negative sense of feedback is what purpose is made of.

Third, the most quoted line associated with Wiener is usually pinned to the wrong paper and the wrong number of authors. The familiar sentence about the best material model of a cat comes from Rosenblueth and Wiener, The Role of Models in Science, in Philosophy of Science, 1945 (Rosenblueth and Wiener 1945), and it has two authors, not one. The 1943 paper carries a plainer relative of it on page 23: the ultimate model of a cat is of course another cat, whether it be born of still another cat or synthesized in a laboratory. A fourth correction belongs with those. Behavior, Purpose and Teleology is not Wiener working alone. Rosenblueth is the first author, and the clinical observation that anchors the argument came from the physiologist, not the mathematician.

WIENER AND THE MODEL

What Wiener built into the Unified Model of Tone

The Unified Model of Tone credits Wiener in a single line: Norbert Wiener gave regulation its formal grammar of feedback. That grammar is not decoration in the model. It is the machinery. The central claim of the model is a claim about a loop, and Wiener is the figure who wrote down what a loop is and what can go wrong inside one.

Take the feedback thread first. In the model a regulatory reflex is not a fixed arc but a loop with a responsiveness of its own, which can be sharp or sluggish. Much of disease is that responsiveness going slack. Wiener reached the same shape from the machine side between 1941 and 1943, when a gun director with its correction strength turned too high began to hunt instead of settling. The failure sat in the tuning of the loop. It sat in neither the motor nor the sensor. Read a guarding pattern or a tremor that way and the question becomes how well the loop answers changing demand, not how strong the tissue is.

Then the information thread, which the model builds from the ground up. Information is difference. A signal means something only insofar as it could have been otherwise, and a light that is always on tells you nothing. DNA stores its message as an ordered sequence of four molecules, and that pattern is written in space. A processor stores its message as two voltage levels read against the edge of a clock, and that pattern is written in time. Same principle in two media. A pattern of differences is a code, and waves are simply the most convenient way to make a difference that travels. The error signal Wiener formalized is exactly such a difference: the margin between where a system stands and where it is aimed.

The model then takes one step past him. Wiener established that purposeful behavior requires negative feedback and that the quantity worth measuring is the error. The model names what the body's control system is holding steady. It is not a value but an organization, and that organization is tone. Fibrillation is the demonstration Wiener himself worked on in 1946. Every fiber contracts, fuel burns fast, almost no blood moves, and the only variable that changed is the phase relationship between the parts. Strength was never the variable under test.

The boundary is worth stating plainly. Wiener never wrote about tone in the sense used on this site, never examined a spine, and made no clinical claim about manual care of any kind. The mathematics and the measurement are his. The claim about what the living loop is regulating is the model's own. Clinicians who described the facilitated segment were feeling that same loop from the other end.

WHAT THE RECORD SHOWS

Wiener and negative feedback in dated facts, from Ampere in 1834 to the last Macy conference in 1953

• 1943. Behavior, Purpose and Teleology ran seven pages in Philosophy of Science, volume 10, issue 1, pages 18 to 24 (Rosenblueth et al. 1943). Rosenblueth signed from Harvard Medical School, Wiener and Bigelow from the Massachusetts Institute of Technology.
• Page 19. The 1943 paper states that all purposeful behavior may be considered to require negative feedback, and that what we select voluntarily is a specific purpose rather than a specific movement.
• Page 20. The same paper matched a cerebellar patient carrying a glass of water to an underdamped machine, and proposed that the main function of the cerebellum is the control of motor feedback (Rosenblueth et al. 1943).
• 1948. Cybernetics ran 8 chapters and 212 pages. Its introduction replaced the one way reflex chain with a circular process, running out to the muscles and re-entering through the proprioceptors (Wiener 1948).
• 1946. Wiener and Rosenblueth modeled cardiac conduction in Archivos del Instituto de Cardiologia de Mexico, volume 16, pages 205 to 265, treating fibrillation as a failure of the network rather than of the cell (Wiener and Rosenblueth 1946).
• 1946 to 1953. Ten Macy conferences ran from 21 March 1946 to 24 April 1953 under Warren McCulloch (Heims 1991). From the seventh meeting in 1950 the series was titled Cybernetics: Circular Causal and Feedback Mechanisms in Biological and Social Systems.
• 1834. Andre-Marie Ampere used cybernetique for the sciences of government (Ampere 1834) more than a century before Wiener and Rosenblueth settled the modern technical sense in the summer of 1947.