Research figure: B. F. Skinner · Written by Artificial Sapiens
What Was Project Pigeon?
Project Pigeon was B. F. Skinner's World War II research program to investigate whether trained pigeons could function as part of a guided weapon's target-tracking system. Developed from experiments that Skinner began around 1940 and pursued with General Mills and American defense authorities, it combined operant training, optics, mechanical control, and a proposal for military guidance. The birds were expected to peck at an image of a target projected onto a translucent surface. In a functioning guidance assembly, the location of their pecks or the resulting movement of a sensitive screen would provide information for correcting a weapon's course. Skinner later described the work and its postwar continuation in his primary historical account, “Pigeons in a Pelican” (1960).
The historical result is precise. Skinner and collaborators demonstrated substantial control over the pigeons' trained visual responses and constructed experimental guidance apparatuses. They did not establish a combat-ready pigeon-guided weapon, demonstrate validated operational accuracy against real targets, or deploy such a system in war. The U.S. defense program ended without field adoption, while an independently organized Navy continuation, known as ORCON, explored related principles after the war. This distinction between behavioral feasibility, integrated engineering performance, and military acceptance is essential to understanding Project Pigeon. Skinner (1960); Capshew (1993).
The project matters to the history of psychology because it moved operant research outside the usual laboratory question of how an organism acquires a response. Skinner attempted to make measured, trained behavior into a working component of a technical system. The resulting history illuminates the opportunities and constraints of behavioral engineering, the relationship between experimental success and technological reliability, and the ethics of placing animals inside weapons. A surviving experimental nose-cone apparatus, donated by Skinner, is cataloged by the Smithsonian National Museum of American History.
The Wartime Problem and the Intellectual Origin
Precision guidance before dependable miniature electronics
During World War II, armed forces pursued ways to improve the accuracy of weapons directed toward moving or distant targets. Engineers worked on optical tracking, radio control, radar, servomechanisms, and other guidance technologies under severe constraints of weight, expense, vibration, reliability, and deployment time. Skinner's proposal entered that environment as a biological solution to a sensing problem: birds already possessed visual discrimination and could be trained to report visual location through a repeatable motor response. Its potential practical attraction was the combination of a highly developed natural visual system with a response that could be coupled to machinery. The period and program are reconstructed in the historical study “Engineering Behavior” by James H. Capshew (1993).
The word guidance requires care. A bird could learn to track a visible image, while an entire gliding weapon had to maintain stable flight, translate a behavioral signal into the appropriate correction, cope with uncertainty, and arrive accurately. The pigeon was proposed as one element in that complete control chain. Skinner did not demonstrate that the animal alone could perform the entire guidance function.
Skinner's experimental background
Burrhus Frederic Skinner (1904–1990) had developed an experimental method centered on the consequences of behavior, response measurement, and deliberately arranged contingencies. His 1938 book The Behavior of Organisms described a program of studying individual organisms under controlled conditions, including response differentiation and reinforcement. Earlier traditions supplied important groundwork: Edward Thorndike had formulated the law of effect, while John B. Watson had advanced an influential early program of behaviorism. Skinner's contribution was his distinctive operant experimental system and, subsequently, the philosophical articulation of radical behaviorism. These intellectual foundations should be studied in his main research chapter, rather than retroactively treating the wartime project as the origin of all operant psychology. Skinner Foundation bibliography.
In conventional laboratory experiments the researcher could simplify the environment, control the reinforcement schedule, and document response rates. Project Pigeon imposed different demands: rapid discrimination, persistence despite interrupted reinforcement, visual variation, mechanical coupling, and survival of a trained response under unfamiliar physical circumstances. Applied reliability had to extend beyond the fact that an animal could peck a key in a chamber. Skinner (1960); Capshew (1993).
Historical Development, 1940–1953
Early demonstrations, beginning around 1940
In his retrospective account, Skinner located the beginning of the idea in the early wartime period and described initial investigations with pigeon tracking and moving visual targets. He imagined training birds to respond in ways that could report target location to a device. These were exploratory demonstrations conducted before the full defense-contract program. Initial difficulty in persuading government and engineering authorities was part of the history, and the retrospective character of Skinner's 1960 account should remain visible: it is a firsthand explanation written after the events, informed by the author's investment in the proposal. Skinner (1960).
General Mills and the experimental engineering effort
The project developed more substantially when General Mills became involved. Work in Minneapolis, including an improvised research setting in a flour mill, brought together behavioral training and efforts to create a functioning optical-mechanical interface. General Mills' engineering participation was indispensable: the relevant device needed optical projection, a reliable signal from the pecking surface, and mechanical systems capable of converting that signal into corrections. Calling the whole enterprise solely Skinner's invention obscures this collaborative technical labor. Capshew (1993); Peterson (2004).
The work was also a setting for Skinner's own methodological development. As Gail Peterson later reconstructed, an episode in the flour mill in 1943 helped Skinner recognize the power of manually reinforcing successive approximations to a desired performance. Skinner had already described response differentiation in 1938; the particular event involved a new, vividly demonstrated practical understanding of hand shaping. It is therefore historically inaccurate to say either that the entire method of shaping began from nothing during Project Pigeon or that wartime work added nothing to Skinner's techniques. Peterson (2004).
Defense evaluation in 1943 and the termination of support
Skinner and the General Mills team obtained a formal opportunity to develop and demonstrate the system for the wartime defense research organization. The National Defense Research Committee, operating within the Office of Scientific Research and Development, evaluated whether pigeon-based guidance deserved further support. Historians distinguish the supported experimental period in 1943, the decisions not to continue the contract in early 1944, and the broader administrative record of program termination. Popular histories sometimes compress these stages into a single termination date in late 1944. The best-supported account avoids treating those different decisions as if they were identical events. Capshew (1993); Skinner (1960).
Skinner believed the behavior of the pigeons had been impressive and that unfamiliarity with an animal-based guidance system contributed to the rejection. Historical interpretations also identify the unresolved engineering connection between animal response and steering control, limited comparative evidence, competing technologies, and the military's need for a deployable system. These explanations are compatible in part, yet their relative causal importance cannot be inferred from successful training demonstrations alone. A later conference study by Eleuterio-da-Rocha and Strapasson proposed that the system's military limitations deserve more attention than some popular accounts allow; that interpretation remains an additional historiographical argument, rather than a controlled experimental verdict. Skinner (1960); Capshew (1993).
The postwar ORCON program, 1948–1953
The line of research had a second institutional life. Beginning in 1948, a Naval Research Laboratory group investigated organic control, usually abbreviated ORCON, and continued work into 1953. ORCON was related to Skinner's conception, while constituting a distinct later program with its own investigators, tests, constraints, and decisions. It should not be presented as uninterrupted wartime funding for Project Pigeon or as evidence that Skinner personally directed all postwar work. Skinner discussed both projects in his 1960 retrospective. Skinner (1960); Capshew (1993).
The later program made the problem of quantitative performance especially important. It is one thing to show consistent pecking in a managed demonstration and another to establish discrimination accuracy, tracking precision, reliability across target types, resistance to interference, and effective command of a complete weapon system. Neither the continuation of research nor the end of ORCON by itself proves Skinner's initial idea was impossible. Both underscore that a promising behavioral component required stronger comparative operational evidence than the historical program provided.
How the Pigeon-Based Guidance System Was Intended to Work
Visual projection and the target image
A proposed guidance head placed a bird in an enclosed compartment facing a small translucent surface, with an optical system displaying a view of the target. As the projected target shifted relative to the screen, the bird's task was to keep contacting the target image. This employed visual tracking and discrimination, not a pigeon understanding military strategy or the consequences of a strike. The animal needed an acquired response pattern that could be captured by a sensor. Skinner (1960); Smithsonian collection record.
The historical designs changed during development. Accounts describe experimental targets on moving fields, a target projected onto a surface, peck location as a behavioral signal, and screens whose mechanical displacement could close electrical contacts. For this reason a single simplified diagram cannot represent every stage of the project. Popular descriptions also use the terms bomb, gliding bomb, rocket, and missile with varying precision. Skinner's 1960 title refers to the Pelican, a guided-weapon development context; experimental nose-cone prototypes and the imagined complete weapon should be distinguished.
Pecking, sensing, and course correction
The proposed chain can be stated as five linked operations. First, optics produced a visual target image. Second, a trained pigeon oriented toward and pecked the image. Third, the mechanical or electrical interface registered where the bird contacted or displaced the target surface. Fourth, the control mechanism interpreted deviation from a desired center. Fifth, actuators in a complete system would change the vehicle's orientation so that the image returned toward the target position. The arrangement created a form of closed-loop feedback: the next sensory image would depend partly on the previous correction. Skinner (1960).
The chain exposed multiple independent opportunities for error. Optical projection might fail to preserve a discriminable target, the bird might respond to an irrelevant feature, peck locations might be noisy, the sensor could register incorrectly, the servo could lag or oscillate, and the airframe might respond poorly. Stable training controlled only some of these factors. A functional demonstration of a bird maintaining contact with a moving image could not measure the full closed-loop system's effectiveness under operational conditions without integration and realistic trials.
Why three pigeons appeared in the design
The best-known prototype had three compartments. Multiple trained birds offered the prospect of cross-checking responses and maintaining a usable signal if an individual was inconsistent. This was an early form of biological redundancy, although historical accounts do not establish a universally validated voting algorithm or a combat-qualified error tolerance. The Smithsonian artifact, accession 1981.0997.02, preserves a three-compartment nose-cone assembly. Its museum description emphasizes the potential for the birds' pecks to serve as a release signal, while Skinner's broader account discusses directional tracking and control. Those descriptions capture related aspects of historical proposals, and should not be flattened into a claim that the surviving artifact proves a completed integrated steering system.
Reinforcement and behavior maintained during a task
A pigeon learns a pecking response when its occurrence in a relevant situation has consequences that increase future responding. During training, Skinner arranged food reinforcement and adjusted contingencies until contact with a visual target became reliable. He also investigated how behavior could continue when reinforcement was intermittent or temporarily unavailable. These principles were essential because a bird performing in a guidance device could not depend on immediate food reward after every contact. Reinforcement history could maintain a learned pattern for limited periods, yet generalization to real flight conditions required independent evidence. Skinner (1960); Peterson (2004).
Research Design and Experimental Evidence
What was experimentally manipulated?
The controlled behavioral components involved variations in visual targets, target position, training contingencies, reinforcement delivery, and the mechanical circumstances surrounding the birds. The procedures included arranging discriminations among visible figures, bringing the response progressively under stimulus control, establishing extended responding, and using moving displays to test tracking. In these settings the bird was the experimental organism, the visual display provided discriminative information, reinforcement was the manipulated learning consequence, and pecking location, frequency, or sustained tracking served as observable behavior. Skinner (1960).
These experiments were not a single modern preregistered trial with a common participant count and a single published outcome measure. The historical source describes an evolving engineering research program with demonstrations and apparatus development. Its claims therefore need to be evaluated separately: acquisition of a response, maintenance of target contact, instrument operation, and prospective effectiveness of a military vehicle are different endpoints.
What did the demonstrations establish?
The project produced evidence that pigeons could be trained to track visual targets and to persist in responding under conditions created by Skinner's team. Demonstrations showed that trained animal behavior could interact with apparatus and might be usable as a signal in a technical control system. The studies also informed subsequent analyses of shaping and stimulus control. This was a serious behavioral feasibility result, not merely a fanciful proposal. Skinner (1960); Peterson (2004).
The surviving record does not supply the kind of publicly documented, controlled operational validation that would establish accuracy against representative military targets, performance under full flight dynamics, reliability across deployment conditions, error rates against competing guidance systems, and a demonstrably acceptable cost and failure profile. Those omissions limit retrospective conclusions. It would be misleading both to dismiss all the animal training as a failure and to characterize the weapon as successfully proved in combat-ready form. Capshew (1993).
The problem of demonstration films
The team used films to communicate the work to institutional decision-makers. This matters because footage can show the coordination of an animal and apparatus under selected, controlled circumstances while leaving unobserved the distribution of failures, the number of attempts, the selection of cases, and how the system would perform outside the filmed setting. Film scholar Benjamin Schultz-Figueroa analyzed Project Pigeon's visual record as a historical construction of technologically usable animal vision. His work demonstrates that the films were themselves instruments of persuasion within military science. Schultz-Figueroa (2019).
A demonstration film can be authentic and informative without meeting the standards of a blinded or representative performance evaluation. For scientific reconstruction, archival film, Skinner's personal narrative, contract decisions, surviving equipment, and later independent historical analysis should be treated as different evidence types rather than interchangeable proof of full system performance.
Why Was Project Pigeon Rejected?
Behavioral reliability was necessary but not sufficient
Skinner's recollection emphasizes officials' difficulty in accepting that pigeons could do important work in a guidance system. Social and institutional expectations about credible technology plausibly influenced judgments. In his own account, he was strongly impressed by the behavior he could obtain from the birds and frustrated by the proposal's reception. As a historical actor, Skinner provides invaluable testimony about his goals and experiments, but his frustration does not independently establish the decision-makers' motives. Skinner (1960).
Capshew's technological history situates the decision in the wartime organization of psychological expertise, engineering contracts, demonstration practices, and defense priorities. Military acceptance required confidence in a complete device and evidence of its comparative advantage. The interface between the behavioral signal and the control machinery posed problems separate from whether a pigeon could discriminate an image. A decision to discontinue support could reflect several interdependent factors, including project maturity, evaluation confidence, rival systems, schedule, and institutional risk. Capshew (1993).
Competing explanations and their evidentiary status
Three stories recur in secondary accounts: that officials simply found the idea ridiculous; that a technical servomechanism prevented adequate integration; and that electronic guidance made animal control unattractive. Each has a historical basis as a relevant concern, but no single slogan explains every decision made between the initial wartime evaluations and ORCON's later termination. Claims that the atomic bomb made accurate conventional guidance unnecessary also oversimplify the different tactical problems. The historiography supports a multicausal interpretation, with disagreement remaining about the relative weight of technical performance and institutional attitudes. Capshew (1993); Schultz-Figueroa (2019).
An important methodological point follows: a negative institutional decision is not an experiment demonstrating that animal-based visual discrimination is impossible. Equally, success in a small behavioral task does not oblige a defense agency to accept an integrated system with unresolved reliability questions. Project Pigeon became historically famous partly because both types of evidence could be misread as answering the other question.
Skinner's Collaborators, Later Researchers, and Attribution
Project Pigeon sits at the intersection of Skinner's experimental agenda, the General Mills engineering effort, defense-research administration, and other researchers working with trained animals. The primary conceptual proposal and retrospective account are Skinner's. Mechanical and organizational realization was collaborative. Government and Navy investigators made decisions and conducted continuation work that should be attributed to their respective institutions. Later historical studies by Capshew and Schultz-Figueroa are independent interpretations of the surviving evidence, not additional experiments conducted by Skinner.
The graduate researchers Keller and Marian Breland became influential in the commercial and applied training of animals, developing Animal Behavior Enterprises after their work in Skinner's environment. Their later landmark paper, “The Misbehavior of Organisms” (1961), showed that species-typical behavioral patterns could disrupt some apparently well-established trained performances. It does not retrospectively prove that Project Pigeon's specific task failed, but it demonstrates why applied behavior engineering must take biology and ecological predispositions seriously. Breland and Breland (1961); Smithsonian historical account.
Nor should Project Pigeon be described as the founding clinical trial of contemporary applied behavior analysis. Modern applied behavior analysis acquired its scientific and professional identity through later researchers, including Donald Baer, Montrose Wolf, Todd Risley, and a much broader subsequent literature. The present article concerns a wartime experimental engineering attempt; it establishes no clinical treatment effect and no therapeutic recommendation. The broader learning tradition is examined in Learning Theories in Education.
Behavioral Engineering: Scientific Meaning and Limits
The organism as a trained sensing-and-response component
The engineering ambition was to specify a visual condition, arrange a trained response, measure that response, and put its output into a larger mechanism. This makes Project Pigeon an early, unusually concrete example of an attempt to treat learned animal behavior as part of a control architecture. Its importance lies in the full chain from biological perception through reinforced action to machine response. A pigeon was expected to contribute discriminative visual behavior that the surrounding machinery could transform into guidance information. Skinner (1960); Capshew (1993).
The fact that the bird could track a target does not establish that behavior was mechanically interchangeable with an electronic sensor. Living organisms vary with motivation, fatigue, stress, learning history, and stimulus conditions. In a system whose output depends on those processes, reliability is an empirical property to be characterized, not simply assumed from an attractive training demonstration. The behavioral theory identified tools for changing response probabilities; engineering evaluation had to establish the physical and operational consequences of those probabilities.
Selection, contingency, and feedback
The project illustrates an interesting relationship between two feedback systems. Within learning, consequences over repeated trials selected patterns of response. During a proposed tracking operation, displacement of the target image would help generate moment-to-moment corrective action through a mechanical feedback loop. These are analytically different timescales: a learned disposition acquired through reinforcement history versus online correction enabled by visual and mechanical signals. The distinction prevents careless descriptions of each peck as a new act of learning or each guidance adjustment as a food reinforcement event.
Skinner later developed broader writings about behavioral selection and the use of behavioral knowledge outside laboratories, including his 1981 article “Selection by Consequences.” That later philosophical and scientific program provides context, but it is not empirical evidence that Project Pigeon's specific military engineering system was reliable. Skinner (1981).
Why a modern analogy needs restraint
One can compare Project Pigeon conceptually with an engineered loop that uses a biological image detector instead of an electronic pattern recognizer. The comparison clarifies the system's functional architecture. It does not establish that pigeons operated as digital classifiers, that Skinner invented machine vision, or that the project directly caused contemporary artificial intelligence research. Modern perception systems, control algorithms, robotics, and machine learning each have their own independent histories and evidence. Project Pigeon is best understood as a historically situated attempt at organic control and a case study in the translation of laboratory behavior into technology.
Ethics of Animal-Based Military Technology
The project aimed to place live birds inside the guidance section of a weapon intended for destructive military use. In an actual expendable weapon, the animals would face grave and predictable risk. Their participation raises animal-welfare questions that differ from ordinary laboratory learning studies: the proposed use was integral to a lethal device, the animals could not consent, and military effectiveness was the ultimate objective of the technology. Historical description therefore requires moral as well as mechanical analysis, without projecting today's formal laboratory standards as if they were identical to those of the 1940s.
A careful ethical assessment considers necessity, proportionality, the severity and duration of animal harm, whether equivalent objectives could be pursued without animals, and the welfare conditions imposed during training and testing. Project Pigeon was not adopted as a combat weapon, so descriptions of the fate of hypothetical deployed birds should remain conditional. The preserved prototype is evidence of a genuine proposal and development effort, rather than evidence of wartime animal deployment. Smithsonian collection record; Capshew (1993).
A second ethical issue concerns the uses of behavioral control. An ability to shape an animal's movements can serve basic science, enrichment, beneficial training, or destructive purposes. The scientific description of the contingency does not determine the moral legitimacy of the institution applying it. The Project Pigeon case makes that responsibility visible because the trained behavior was intended to participate in weapons guidance. Its history invites scrutiny of who defines an engineered objective, which organisms bear its costs, and what evidence justifies deployment.
Historical Significance and Contemporary Assessment
What survives as an established finding?
There is strong historical documentation that Skinner undertook the program, that trained pigeons performed target-related visual responses in controlled settings, that apparatuses were built, and that defense evaluation failed to produce operational adoption. The 1960 firsthand narrative, 1993 peer-reviewed historical study, 2019 media historiography, and Smithsonian museum object support complementary parts of that account. The role of flour-mill training in Skinner's appreciation of practical shaping is documented through Peterson's later analysis. Skinner (1960); Capshew (1993); Peterson (2004); Smithsonian.
What remains limited or contested?
The extent to which military skepticism, equipment limitations, alternative guidance programs, institutional politics, and operational performance each determined termination remains a question for documentary historiography. The behavior could be observed; the counterfactual claim that a fully funded system would have succeeded cannot be tested from surviving demonstrations. Reports and films do not support modern numerical claims about hit probability, cross-condition generalization, or comparative weapon reliability unless those values are accompanied by identifiable underlying tests. For that reason this article assigns no invented success rate to the system.
Relevance to psychology and technology
The project remains a powerful example of how scientific knowledge travels through institutions. Skinner entered with expertise in experimental control of learned behavior and confronted a multidisciplinary engineering task that required optics, mechanics, military evaluation, and organizational legitimacy. Important learning principles survived the weapon program's termination; those principles can be investigated in their own right, without claiming the war project validated every later behavioral intervention. Related modern discussion about intervention components and evidence belongs to Behavior Change Techniques: What Research Supports.
A further consequence was historiographical. Skinner's narrative presented the episode as a story about the difficulty of recognizing an unconventional scientific opportunity; Capshew exposed its institutional and technological setting; Schultz-Figueroa examined how its filmed demonstrations produced an image of reliable animal machinery. Together these perspectives provide a richer account than the popular anecdote of pigeons piloting a bomb. Skinner (1960); Capshew (1993); Schultz-Figueroa (2019).
Frequently Asked Questions
Did B. F. Skinner actually train pigeons to guide a missile?
Skinner and collaborators trained pigeons to track and peck visual targets as part of experimental guidance apparatuses. They proposed connecting these responses to a guided weapon. A military-ready pigeon-guided missile was not demonstrated or deployed. Skinner (1960); Smithsonian.
Was Project Pigeon a success or a failure?
It was a behavioral training and experimental apparatus achievement with an unsuccessful military-development outcome. Its partial success was establishing usable learned responses under controlled experimental conditions; its limiting result was the absence of a validated and accepted full weapon system. Treating these as separate endpoints resolves the apparent contradiction. Capshew (1993).
How did a pigeon know which target to hit?
The birds learned to peck a visual target shown on a screen. Reinforcement during training increased appropriate contact with the relevant image. That did not require the pigeon to represent the military identity of a target or intend a weapon's outcome. Skinner (1960).
Why were there three pigeons?
The best-known proposed guidance head held three birds in separate compartments. Multiple birds could potentially provide redundant or corroborating response information. The surviving artifact documents this design, while the detailed performance of any proposed combined-signal mechanism must be evaluated separately. Smithsonian.
What were Project Pelican and Project ORCON?
Pelican referred to the guided-weapon development context associated with wartime work; Project Pigeon referred to Skinner's animal-based guidance research. ORCON, short for organic control, was a related, distinct Naval Research Laboratory continuation beginning in 1948 and ending in 1953. These terms identify related but historically different elements of the research. Skinner (1960).
Did this project create operant conditioning or modern artificial intelligence?
Operant research substantially preceded Project Pigeon, and the wartime work offered an important setting for applied shaping and experimental technology. It supplies an instructive historical comparison for feedback systems, not evidence of direct invention of contemporary machine learning or AI systems. Peterson (2004); Capshew (1993).
Related Articles
B. F. Skinner — The central scientific chapter on Skinner's life, operant research system, theoretical writings, and historical influence.
Learning Theories in Education: Behaviorism, Cognitivism, Constructivism, and Sociocultural Learning — Scientific context for behavioral learning and its relationship to other explanatory traditions.
Behavior Change Techniques: What Research Supports — Modern behavior-change evidence and its methodological boundaries, separate from wartime military engineering.
B. F. Skinner Box and the Operant Conditioning Chamber — The operant chamber and response measurement.
B. F. Skinner's Superstition in the Pigeon Experiment — The 1948 response-independent food experiment.
B. F. Skinner: Biography and Intellectual Development — Archival chronology and intellectual formation.
References
Breland, K., and Breland, M. (1961). The misbehavior of organisms. American Psychologist, 16(11), 681–684. https://doi.org/10.1037/h0040090.
Capshew, J. H. (1993). Engineering behavior: Project Pigeon, World War II, and the conditioning of B. F. Skinner. Technology and Culture, 34(4), 835–857. https://doi.org/10.2307/3106417.
National Museum of American History, Smithsonian Institution. (n.d.). B. F. Skinner's nose cone of a pigeon-guided missile (Object 1981.0997.02). https://americanhistory.si.edu/collections/object/nmah_690069.
Peterson, G. B. (2004). A day of great illumination: B. F. Skinner's discovery of shaping. Journal of the Experimental Analysis of Behavior, 82(3), 317–328. https://doi.org/10.1901/jeab.2004.82-317.
Schultz-Figueroa, B. (2019). Project Pigeon: Rendering the war animal through optical technology. JCMS: Journal of Cinema and Media Studies, 58(4), 92–111. https://doi.org/10.1353/cj.2019.0059.
Skinner, B. F. (1938). The behavior of organisms: An experimental analysis. D. Appleton-Century. B. F. Skinner Foundation full bibliography.
Skinner, B. F. (1960). Pigeons in a pelican. American Psychologist, 15(1), 28–37. https://doi.org/10.1037/h0045345.
Skinner, B. F. (1981). Selection by consequences. Science, 213(4507), 501–504. https://doi.org/10.1126/science.7244649.