Negative feedback
Negative feedback is the principle behind how a thermostat keeps your house warm, how your body maintains its temperature, and how engineers stopped electronic amplifiers from tearing themselves apart. It is one of the most quietly pervasive ideas in science and technology. The concept seems simple: when a system drifts away from where it should be, something pushes it back. But inside that simplicity lies a rich history stretching back to water clocks in Alexandria around 200 BCE, a contested patent battle at Bell Laboratories in the 1920s and 1930s, and a scientific debate that swept from engineering into biology, economics, and the study of the mind itself.
What makes a system stable rather than runaway? Why do some feedback loops spin out of control even when they are designed to correct themselves? And how did an idea rooted in furnaces and millstones become the organizing principle of cybernetics?
Ktesibios of Alexandria is the earliest name attached to negative feedback as a deliberate technique. Working in the 3rd century BCE, he refined the water clock by adding mechanisms that maintained a constant level in the reservoir, and self-regulating mechanisms of this kind were in use as early as 200 BCE.
The idea did not develop much further until the 17th century, when Cornelius Drebbel built thermostatically controlled incubators and ovens in the early 1600s. Around the same period, centrifugal governors were already regulating the distance and pressure between millstones in windmills. These devices held a working balance not because any human adjusted them, but because the system itself detected drift and corrected it.
James Watt patented his own form of governor in 1788 to control the speed of his steam engine, and that machine became famous enough that the centrifugal governor entered engineering textbooks as a defining example. James Clerk Maxwell then brought mathematical rigor to the subject in 1868, describing what he called "component motions" that lead to a decrease in a disturbance or the amplitude of an oscillation. Mercury thermostats dating from around 1600 used the expansion and contraction of mercury columns to control furnace vents, and a steering engine patented in 1866 applied power assistance to a ship's rudder through a feedback loop to hold the heading set by the steersman.
Harold Stephen Black came up with the idea of using negative feedback in electronic amplifiers in 1927, while working at Bell Laboratories. He submitted a patent application in 1928 and followed that with a detailed paper in 1934, in which he defined negative feedback as a type of coupling that reduced the gain of the amplifier while greatly increasing its stability and bandwidth.
The patent itself, US Patent 2,102,671, was not granted until 1937, and it ran to 52 pages plus 35 pages of figures. The first 43 pages were described at the time as amounting to a small treatise on feedback amplifiers.
The advantages Black identified were considerable. Negative voltage feedback reduces non-linear distortion, producing higher fidelity; it increases circuit stability so that gains remain stable across variations in temperature, frequency, and signal amplitude; it slightly increases bandwidth; and it considerably reduces harmonic, phase, amplitude, and frequency distortions as well as noise.
Karl Kupfmuller had published papers on a negative-feedback-based automatic gain control system and a feedback system stability criterion in 1928, developing the theory in parallel. Harry Nyquist, also of Bell Laboratories, then proposed a stability criterion and a graphical plot to identify which feedback systems would remain stable, a tool that became essential to both amplifier and control system design. Nyquist and Hendrik Bode built on Black's foundational work to produce a general theory of amplifier stability.
A negative feedback loop can still oscillate, and understanding why took years of careful analysis. The culprit is phase shift: as a signal travels around a loop, its timing shifts. At certain frequencies, that phase shift can reach 180 degrees, at which point the corrective signal arrives back in phase with the disturbance it was supposed to cancel. What was negative feedback becomes positive feedback, and the system runs away.
Even before the phase shift reaches that critical 180-degree threshold, stability degrades. Under- and overshoot following a disturbance grow larger and larger. Engineers address this through a design step called compensation, which attenuates or reshapes the phase of the problematic frequencies. Many negative feedback systems also incorporate low-pass filters or dampers to suppress the frequencies most prone to instability.
In amplifier design, the closed-loop gain of a negative feedback amplifier is set by the formula involving the open-loop gain A and the feedback fraction beta, and when beta multiplied by A is much greater than one, the closed-loop gain becomes approximately equal to one divided by beta. This near-independence from the open-loop gain is why Black called the feedback arrangement a way to "desensitize" the circuit. The factor by which feedback reduces disturbances is called the improvement factor, or equivalently the desensitivity factor, and equals one plus beta multiplied by A. A higher improvement factor reduces both the error signal at the amplifier input and the effect of internal noise or distortion on the output.
In biological systems, negative feedback is the mechanism behind homeostasis, the word biologists use for the body's ability to regulate its internal environment. Blood pressure regulation involves the baroreflex, a reflex arc that senses pressure changes and acts to counteract them. Erythropoiesis, the production of red blood cells, is similarly regulated by feedback.
The endocrine system offers one of the clearest documented examples. The hypothalamus secretes corticotropin-releasing hormone, which prompts the anterior pituitary gland to release adrenocorticotropic hormone, known as ACTH. ACTH in turn directs the adrenal cortex to secrete glucocorticoids such as cortisol. Once sufficient glucocorticoids are circulating, they act back on both the hypothalamus and the pituitary gland to suppress further stimulation, reducing the output of glucocorticoids. The loop is self-limiting by design.
When these loops fail, the consequences can be severe. If the negative feedback regulating blood glucose levels breaks down, glucose concentrations in the blood can rise dramatically, a condition that results in diabetes. The chemistry of reversible reactions also exhibits the same principle: when nitrogen gas is added to a sealed system containing nitrogen, hydrogen, and ammonia at equilibrium, the equilibrium shifts toward producing more ammonia in response, partially offsetting the added nitrogen.
Norbert Wiener helped pull negative feedback out of engineering and into a much broader intellectual framework. Writing in 1948, he defined feedback in general as "the chain of the transmission and return of information", and negative feedback specifically as the case when "the information fed back to the control center tends to oppose the departure of the controlled from the controlling quantity."
Early cybernetics researchers generalized the concept to cover any goal-seeking or purposeful behavior, going so far as to state that all purposeful behavior may be considered to require negative feedback. W. Ross Ashby, however, pointed out a limit to this framing. He observed that while the idea of feedback is simple and natural in elementary cases, it "becomes artificial and of little use when the interconnections between the parts become more complex," and that complex, richly cross-connected systems can only be understood as a whole, not as an interlaced set of independent feedback circuits.
In 1924, Friis and Jensen had described what they called "positive feedback" and made passing mention of "negative feed-back action" as a contrast. By the time Wiener wrote in 1948, the term negative was already carrying multiple, colliding meanings: the mathematical sign of a multiplier in control theory, the reduction of a gap between desired and actual behavior in business modeling, and the valence of criticism versus praise in psychology. Later authors proposed alternatives, including degenerative, self-correcting, balancing, and discrepancy-reducing, in an effort to reduce that confusion.
Wiener was notably skeptical of the claim that free markets operate as a self-correcting feedback system. In 1948 he wrote that there is "a belief current in many countries and elevated to the rank of an official article of faith in the United States that free competition is itself a homeostatic process" but added that "unfortunately the evidence, such as it is, is against this simple-minded theory."
Hetrodox economists including financier George Soros and Herman Daly, who was with the World Bank from 1988 to 1994, have similarly questioned whether market forces reliably maintain economic equilibrium. Government automatic stabilizers, by contrast, are programs explicitly designed to act as negative feedback to dampen fluctuations in real GDP.
In environmental science, negative feedback operates across atmospheric and ecological systems. When incoming solar radiation increases, planet temperature rises, which allows more plant life to grow. That plant growth produces compounds such as sulfur, which generate more cloud cover. Greater cloud cover raises the Earth's albedo, the fraction of solar radiation that is reflected away, and the increased albedo then reduces the amount of radiation reaching the surface, pulling temperature back down. A parallel loop runs through the hydrological cycle: rising temperatures produce more water vapor, which forms more clouds, which block incoming solar radiation, which lowers temperature, which reduces water vapor production. Both loops act to stabilize climate conditions rather than amplify them. The phase-locked loop, first described in 1932, applies the same correction principle in electronics to keep a generated waveform locked in constant phase with a reference signal.
Up Next
Common questions
What is negative feedback and how does it work?
Negative feedback occurs when some function of a system's output is fed back in a way that reduces fluctuations, whether caused by changes in input or by other disturbances. It works by measuring the difference between a desired value and the actual value, then applying a correction that narrows that gap, promoting stability and equilibrium.
Who invented the negative feedback amplifier?
Harold Stephen Black invented the negative feedback amplifier at Bell Laboratories in 1927. He submitted a patent application in 1928 and published a detailed paper in 1934. The patent, US Patent 2,102,671, was granted in 1937 and ran to 52 pages plus 35 pages of figures.
What are examples of negative feedback in biology?
Negative feedback in biology includes the baroreflex that regulates blood pressure, the hormonal cascade controlling glucocorticoid secretion through the hypothalamus, pituitary gland, and adrenal cortex, and the regulation of blood glucose levels. When the glucose feedback loop fails, it can result in diabetes.
How far back does the history of negative feedback go?
Self-regulating mechanisms based on negative feedback were in use as early as 200 BCE, when Ktesibios of Alexandria refined water clocks to maintain a constant reservoir level. Cornelius Drebbel built thermostatically controlled incubators and ovens in the early 1600s, and James Watt patented a centrifugal governor for his steam engine in 1788.
Why can negative feedback loops still oscillate?
Negative feedback loops can oscillate because of phase shifts around the loop. At certain frequencies, the phase shift reaches 180 degrees, causing the corrective signal to arrive back in phase with the original disturbance and turning negative feedback into positive feedback. Engineers address this through compensation, low-pass filters, or dampers.
How does negative feedback apply to environmental and climate systems?
In climate systems, negative feedback stabilizes temperature through loops involving solar radiation, cloud cover, and albedo. When solar radiation increases and raises planet temperature, plant growth and water vapor production increase, which generates more cloud cover, which reflects more solar radiation and reduces temperature. These loops counteract disturbances rather than amplify them.
All sources
50 references cited across the entry
- 1bookIntroduction to cyberneticsW. Ross Ashby — Chapman & Hall Ltd.; Internet (1999) — 1957
- 2bookEcologyRobert E. Ricklefs et al. — Macmillan — 2000
- 3journalOn The Definition of FeedbackArkalgud Ramaprasad — 1983
- 4journalResearch Notes. Feedback: The Definition of a ConstructDavid M. Herold et al. — 1977
- 5bookProcess Dynamics and ControlSudheer S Bhagade et al. — PHI Learning Pvt. Ltd — 2011
- 6bookPrinciples of Feedback ControlCharles H. Wilts — Addison-Wesley Pub. Co — 1960
- 7bookProcess Control: Concepts Dynamics And ApplicationsSK Singh — PHI Learning Pvt. Ltd — 2010
- 8bookA Real-Time Approach to Process ControlWilliam Y. Svrcek et al. — John Wiley & Sons — 2013
- 9webTypes of feedback controlCharles D H Williams — University of Exeter: Physics and astronomy
- 10journalA climate model-based review of drought in the Sahel: Desertification, the re-greening and climate changeAlessandra Giannini et al. — 2008-12-01
- 11journalFeedback for Physicists: A Tutorial Essay On ControlJohn Bechhoefer — 2005
- 12webU.S. Patent 2,102,671: Wave Translation SystemHarold Black — 1937-12-21
- 13journalElectrical engineering hall of fame: Harold S BlackJames E Brittain — February 2011
- 14journalIn Memoriam: Harold Stephen BlackCA Desoer — August 1984
- 15bookBasic electronics: Devices, circuits and its fundamentalsSantiram Kal — PHI Learning Pvt. Ltd — 2009
- 16bookIntuitive Analog Circuit DesignMarc Thomson — Newnes — 2006
- 17bookBasic Electronics: Devices, Circuits, and IT FundamentalsSantiram Kal — PHI Learning Pvt. Ltd — 2009
- 18inlineMarc T Thompson, p. 309
- 19bookThe Design of CMOS Radio Frequency CircuitsThomas H Lee — Cambridge University Press — 2004
- 20bookElectronic Circuits: Analysis simulation and designNorbert A Malik — Prentice Hall — 1995
- 21bookBasic Electronics: Devices, Circuits and IT fundamentalsSantiram Kal — PHI Learning Pvt. — 14 January 2009
- 22bookLinear Control Systems: For Punjab Technical UniversitySK Bhattacharya — Pearson Education India
- 23bookMicroelectronic Circuits: Analysis & DesignMuhammad Rashid — Cengage Learning — 2010
- 24bookCircuit Analysis and Feedback Amplifier TheoryWai-Kai Chen — CRC Press — 2005
- 25bookDesign with operational amplifiers and analog integrated circuitsSergio Franco — McGraw-Hill — 2002
- 26bookThe Design of High Performance MechatronicsG. Schitter et al. — IOS Press — 2014
- 27bookOp Amp Applications HandbookWalter G Jung — Newnes — 2005
- 28bookPsychomythics: Sources of Artifacts and Misconceptions in Scientific PsychologyWilliam R. Uttal — Psychology Press — 2014
- 29bookSelf-organization in biological systemsScott Camazine et al. — Princeton University Press — 2003
- 32webThe Study of Earth as an Integrated SystemNASA — 2016
- 34journalOceanic phytoplankton, atmospheric sulphur, cloud albedo and climateRobert J. Charlson et al. — 1987
- 35journalAmplified Arctic climate change: What does surface albedo feedback have to do with it?Michael Winton — 2006
- 36journalCloud Feedbacks in the Climate System: A Critical ReviewGraeme L. Stephens — 2005
- 37journalGlobal Iron Connections Between Desert Dust, Ocean Biogeochemistry, and ClimateT. D. Jickells et al. — 2005
- 38journalA climate model-based review of drought in the Sahel: Desertification, the re-greening and climate changeAlessandra Giannini et al. — 2008
- 39journalPort-based modeling of mechatronic systemsPeter C Breedveld — 2004
- 41bookPower From the WindRichard L Hills — Cambridge University Press — 1996
- 43bookBetween Human and Machine : Feedback, Control, and Computing before CyberneticsDavid A. Mindell — Johns Hopkins University Press — 2002
- 44journalHigh Frequency AmplifiersH. T. Friis et al. — 1924
- 45journalStabilized Feedback AmplifiersH.S. Black — January 1934
- 46bookA history of control engineering 1930-1955Stuart Bennett — Institution of Electrical Engineers — 1993
- 47journalKarl Kupfmuller, 1928: an early time-domain, closed-loop, stability criterionC. Bissell — 2006
- 49bookThe Fifth Discipline: The Art and Practice of the Learning OrganizationPeter M. Senge — Doubleday — 1990
- 50bookScience and Policy in Natural Resource Management: Understanding System ComplexityHelen E. Allison et al. — Cambridge University Press — 2006
- 51bookOn the Self-Regulation of BehaviorCharles S. Carver et al. — Cambridge University Press — 2001-05-07