In 1764, a young Milanese aristocrat named Cesare Beccaria published a slim treatise called On Crimes and Punishments. His argument was straightforward: state torture is unreliable, unjust, and barbaric. It should stop. Within a generation, nearly every European state agreed. Torture was formally abolished across the continent. The Enlightenment had won.
And then it came back.
Not as a medieval relic, but as a modern bureaucratic program, complete with manuals, training curricula, psychological consultants, and legal memoranda explaining why it wasn’t really torture. The states that brought it back were not autocracies clinging to old habits. They were democracies: France, Britain, the United States. The nations most loudly committed to human rights became the ones most creatively dedicated to circumventing them.
This is the central paradox of state torture in the modern era. The evidence that it doesn’t work is overwhelming. The neuroscience is clear. The historical record is unambiguous. Governments keep doing it anyway. Understanding why state torture persists requires looking not at what it produces (bad intelligence, false confessions, international condemnation) but at what it actually does for the people and institutions that practice it.
What the Enlightenment Actually Abolished
Beccaria’s On Crimes and Punishments didn’t emerge from nowhere. By the mid-eighteenth century, judicial tortureThe legally sanctioned use of physical pain by courts to compel confessions or testimony, practiced in medieval and early modern Europe before Enlightenment reforms., the legally sanctioned infliction of pain to extract confessions, was already falling out of favor among European jurists. Frederick the Great of Prussia had restricted its use in 1740. What Beccaria did was provide the philosophical framework: torture punishes before guilt is established, it rewards physical endurance over truthfulness, and it degrades the state that employs it.
The reforms were real. Sweden abolished judicial torture in 1734, Prussia in 1754, France in 1788. By the early nineteenth century, the practice had been formally eliminated across most of Europe. But what was abolished was a specific legal institution: the public, court-ordered application of pain as part of criminal proceedings. What was not abolished, because it had never been formally acknowledged, was the quieter use of coercion by police, colonial administrators, and military officers operating outside the courtroom.
This distinction matters enormously. The Enlightenment abolished torture as spectacle. It did not abolish torture as practice. It just pushed it underground, where it became harder to monitor and easier to deny.
The Colonial Laboratory
The gap between prohibition and practice became most visible in colonial settings. European powers that had abolished torture at home practiced it routinely in their colonies, where the legal protections extended to citizens did not apply to colonial subjects. The British in India and Kenya, the French in Algeria and Indochina, the Dutch in Indonesia: colonial administration ran on coercion, and coercion frequently meant state torture.
The French experience in Algeria between 1954 and 1962 is the defining case study. When the National Liberation Front (FLN) launched an urban insurgency in Algiers in 1957, the French government granted General Jacques Massu extraordinary powers. The French authorities had already sanctioned “extended questioning” and “coercive measures.” There were no written orders to torture, but multiple army officers later confirmed they had received verbal permission to use “whatever means necessary.”
Massu’s paratroopers dismantled the FLN network in Algiers within months. They also tortured thousands of Algerians, many of whom had no connection to the insurgency. Massu himself later admitted that “institutionalized torture” had become routine. A government commission reported in September 1957 that torture was a frequent practice.
The military victory was real. So was the strategic catastrophe. The systematic use of torture became an international scandal, generated global sympathy for Algerian independence, and contributed directly to France’s eventual withdrawal. Torture “worked” in the narrowest tactical sense (some of the intelligence gathered was accurate) while failing completely at the strategic level. France won the Battle of Algiers and lost the war.
The CIA’s Long Education
The United States followed a similar trajectory, just more slowly and with better paperwork. In 1963, the CIA produced the KUBARK Counterintelligence Interrogation manual, a classified guide to extracting information from “resistant sources.” KUBARK was a CIA cryptonym for the agency itself, and the manual was remarkably candid about the limitations of coercion. It acknowledged that “the threat of pain is often more effective than pain itself” and that coercive techniques “can backfire.”
The manual drew on research from the CIA’s MKUltra program and from studies of Soviet and Chinese interrogation methods. It recommended sensory deprivation, sleep disruption, isolation, and psychological manipulation. It also noted, almost as an aside, that these methods frequently produced false information. The agency knew this in 1963. It used the techniques anyway.
KUBARK’s methods spread through the Americas via the CIA’s training programs for Latin American military and police forces during the Cold War. The 1983 Human Resource Exploitation Training Manual, a direct descendant of KUBARK, was used to train interrogators in Honduras, Guatemala, El Salvador, and elsewhere. When both manuals were declassified in 1997 following a Freedom of Information Act request by the Baltimore Sun, the institutional lineage was unmistakable: the same techniques, the same justifications, the same acknowledgment buried in the footnotes that coercion produces unreliable results.
After September 11: The Machinery Restarts
The attacks of September 11, 2001, did not create the American torture apparatus from nothing. They reactivated an institutional memory that had never fully disappeared. Within months, the CIA had contracted psychologists James Mitchell and Bruce Jessen, who designed an “enhanced interrogation” program drawing on the agency’s Cold War research. Mitchell and Jessen’s firm eventually earned more than $80 million from the CIA.
The program held at least 119 men in secret “black siteA secret detention facility operated by an intelligence agency outside normal legal oversight, where detainees are held without official acknowledgment.” prisons across multiple countries between 2002 and 2008. The techniques included waterboarding, prolonged stress positions, sleep deprivation lasting up to 180 hours, confinement in coffin-sized boxes, and “rectal feeding,” a euphemism for sexual assault.
The 2014 Senate Select Committee on Intelligence report, based on a review of more than six million pages of CIA documents, concluded unambiguously that the program was ineffective. The CIA’s justifications rested on “inaccurate claims of their effectiveness.” Detainees provided false information under duress. Accurate intelligence came from detainees before they were subjected to enhanced interrogation, not after. The program never produced “imminent threat” intelligence, the ticking-time-bomb scenario that its defenders invoked constantly.
The Senate report also documented something more troubling than ineffectiveness: the CIA had systematically lied about the program’s results to Congress, to the President, to the Justice Department, and to itself. The agency described in the report was not one that tortured reluctantly and stopped when it realized the methods didn’t work. It was one that tortured, discovered the methods didn’t work, and then fabricated evidence that they had.
Why It Doesn’t Work: The Neuroscience
The question of whether torture “works” for intelligence-gathering has been answered definitively by neuroscience. Shane O’Mara, professor of experimental brain research at Trinity College Dublin, laid out the case in Why Torture Doesn’t Work: The Neuroscience of Interrogation (2015). The argument is not moral (though O’Mara is not indifferent to morality). It is mechanical.
Extreme stress damages the hippocampus and prefrontal cortex, the brain regions responsible for memory encoding, retrieval, and executive function. A person being tortured is not choosing between telling the truth and lying. They are in a neurological state where the distinction between true memories and confabulated ones collapses. The stress hormones flooding their brain actively degrade their ability to recall accurate information. They will say whatever they believe will make the pain stop, and under sufficient duress, they will genuinely lose the ability to distinguish what they know from what they are inventing.
This is not a theoretical objection. It is what happened, repeatedly, in the CIA’s program. Detainees provided false leads that the agency chased as high priorities, wasting resources and personnel on fabricated intelligence. As a review in the journal Peace and Conflict summarized: the stressors inherent in torture “make our recollections less accurate and make us more susceptible to confabulating entirely false ones.”
Professional interrogators, the people who actually extract information for a living, have been saying this for decades. Rapport-based interviewing techniques consistently outperform coercion. As one former military interrogator put it, in a line that should be carved above every intelligence agency’s door: “Torture is for amateurs.”
Why It Persists: The Psychology of the Torturer
If torture doesn’t produce reliable intelligence, why do governments keep using it? The answer lies not in what torture does to the victim, but in what it does for the torturer and the institution behind them.
Stanley Milgram’s obedience experiments at Yale in the early 1960s demonstrated that ordinary people will inflict what they believe to be severe pain on strangers when instructed to do so by an authority figure. In Milgram’s studies, 65 percent of participants administered what they believed were potentially lethal electric shocks when told to continue by an experimenter in a lab coat. The experiment was designed in the shadow of the Holocaust; Milgram wanted to understand how ordinary Germans had participated in mass atrocity. His answer was unsettling: they were not uniquely evil. Almost anyone would do it, given the right institutional framework.
Philip Zimbardo’s Stanford Prison Experiment in 1971 extended this insight, showing how quickly people assigned to positions of authority over others would begin to abuse that authority. (The experiment was deeply methodologically flawed, as later investigations revealed, but the broader pattern it gestured at, that situational power corrupts, has been confirmed by subsequent research.) Zimbardo himself later drew explicit parallels between his experiment and the abuses at Abu Ghraib, arguing in The Lucifer Effect that the soldiers involved were not “bad apples” but products of a “bad barrel“: an institutional environment that normalized, encouraged, and rewarded cruelty.
Darius Rejali, the political scientist whose Torture and Democracy (2007) remains the most comprehensive study of the subject, identified several mechanisms that sustain state torture programs:
- Institutional momentum: once a torture program begins, it tends to escalate. Interrogators who have used moderate coercion and obtained ambiguous results escalate to more extreme methods, reasoning that the problem is insufficient intensity rather than a flawed approach.
- Bureaucratic self-justification: agencies that have invested resources in a torture program have powerful incentives to claim it is working. The CIA’s fabrication of success stories was not an anomaly; it was a predictable institutional response.
- Competitive masculinity and group bonding: torture units develop internal cultures where the willingness to inflict pain becomes a marker of toughness and loyalty. Refusing to participate marks you as unreliable.
- Diffusion of responsibilityA social psychology phenomenon where individuals in a group feel less personally accountable for an outcome because culpability is spread across multiple participants.: modern torture programs distribute culpability across so many participants, the psychologists who designed the program, the lawyers who authorized it, the guards who carried it out, the administrators who looked away, that no individual feels fully responsible.
The Democracy Paradox
Rejali’s most counterintuitive finding is that democracies did not merely tolerate state torture. They innovated it. As public monitoring, free press, and legal accountability increased in the twentieth century, democratic states developed techniques specifically designed to leave no marks: stress positions, waterboarding, sleep deprivation, temperature manipulation, sensory overload, and psychological manipulation. Rejali calls these “clean” torture methods, and he traces their proliferation directly to democratic accountability.
The logic is perverse but internally consistent. A dictatorship can torture openly; the purpose is partly theatrical, a demonstration of what happens to those who resist. A democracy cannot afford the spectacle. It needs deniability. So it develops methods that leave no physical evidence, methods that allow officials to say, with a straight face, “We don’t torture,” while doing exactly that.
This is why the legal memoranda produced by the Bush administration’s Office of Legal Counsel are so revealing. The “torture memos” did not argue that coercion was acceptable. They redefined torture so narrowly, requiring “severe physical suffering” equivalent to “organ failure, impairment of bodily function, or even death,” that virtually anything short of killing the subject could be classified as something else. The linguistic gymnastics were the point. The goal was not to legalize torture but to make it linguistically invisible.
The Real Function of State Torture
If torture is not primarily an intelligence-gathering tool, what is it? The historical record suggests several overlapping functions:
Social control through terror. Torture is often used not to extract information from the person being tortured, but to send a message to everyone else. The French use of torture in Algeria terrorized the broader population into compliance, at least temporarily. Latin American military dictatorships tortured dissidents not because they expected to learn anything, but because the knowledge that dissidents were being tortured deterred opposition. The target audience is not the victim; it is the victim’s community.
The CIA’s interventions during the Cold War followed a similar logic: the intelligence agencies Washington trained across Latin America used coercive techniques not primarily for intelligence, but for political suppression.
Performance of sovereignty. There is something primal in the act of one person exercising total physical control over another. For institutions and states experiencing a perceived threat, torture serves as a reassertion of power. After September 11, the United States was in genuine existential panic. The torture program was, in part, a way for the national security apparatus to demonstrate, to itself as much as to anyone else, that it was doing something. The “something” didn’t need to work. It needed to feel decisive.
Institutional survival. Once a torture program exists, the people running it have careers, budgets, and reputations invested in its continuation. The psychologists Mitchell and Jessen earned $80 million. The legal architects of the program received promotions. The CIA officers who ran the black sites advanced in their careers. Ending the program would mean admitting it was a mistake, and institutional bureaucracies almost never volunteer that admission.
The Professional Alternative
The irony is that effective interrogation methods exist. They are just boring. Rapport-based interviewing, where an interrogator builds a relationship with the subject, establishes trust, and uses conversational techniques to elicit information, consistently outperforms coercion in controlled studies and in practice. The FBI, which opposed the CIA’s enhanced interrogation program from the beginning, obtained more actionable intelligence from the same detainees using standard interview techniques than the CIA obtained through torture.
The problem is that rapport-based methods require skilled, patient, linguistically competent interrogators. They take time. They don’t look dramatic. They don’t satisfy the institutional demand for visible action. A senator who asks “What are we doing about terrorism?” wants to hear something more forceful than “We’re building trust with detainees over extended conversations.” Torture persists, in part, because it looks like the kind of thing that serious people do in serious situations, even though the serious people who actually study interrogation know it doesn’t work.
What the Pattern Tells Us
The history of state torture is not a story of progress interrupted by occasional regression. It is a story of permanent tension between the public commitments democracies make and the private practices they tolerate. The Enlightenment did not end torture; it drove it into the shadows. The Universal Declaration of Human Rights did not prevent the French from torturing in Algeria, the British from torturing in Northern Ireland, or the Americans from torturing in Afghanistan and Iraq. Legal prohibitions set a standard. They do not, by themselves, enforce it.
The pattern is consistent across centuries and continents: a perceived existential threat (revolution, terrorism, insurgency) creates political permission for coercion. Institutional actors develop or revive techniques that offer plausible deniabilityA condition in which a state or official can credibly deny involvement in a covert action because no formal evidence of their participation exists.. The techniques don’t produce reliable intelligence, but they do produce institutional momentum, career incentives, and a sense of decisive action. By the time the failure becomes undeniable, the program has generated enough false justifications, legal cover, and bureaucratic inertia to sustain itself for years.
Understanding this pattern is not defeatist. It is diagnostic. If torture persists not because it works but because of the institutional and psychological pressures that sustain it, then those pressures are the intervention points: transparency, accountability, professional standards for interrogators, and the political courage to accept that effective intelligence work looks nothing like what it does in the movies.
The Brain Under Duress
Start with the most basic question: does torture produce accurate information? Neuroscience answers this clearly, and the answer is no.
Shane O’Mara, professor of experimental brain research at Trinity College Dublin, examined the question in Why Torture Doesn’t Work: The Neuroscience of Interrogation (2015). His argument is mechanical, not moral. When a person is subjected to extreme stress, the body floods with cortisol and adrenaline. These stress hormones damage the hippocampus and prefrontal cortex, the brain regions responsible for memory retrieval and executive function. The person being tortured is not weighing whether to tell the truth or lie. Their brain is in a state where the distinction between real memories and fabricated ones breaks down.
Under sufficient pain and fear, people will say whatever they think will stop it. They will confess to things they didn’t do. They will provide information they don’t have, inventing plausible-sounding details because the brain, desperate to end the threat, generates content automatically. This is not weakness or moral failure. It is neurology. As a review in Peace and Conflict put it: the stressors inherent in coercive interrogation “make our recollections less accurate and make us more susceptible to confabulating entirely false ones.”
This is exactly what happened in the CIA’s post-9/11 enhanced interrogation program. The 2014 Senate Intelligence Committee report found that detainees under duress provided false leads that the CIA chased as high priorities, wasting resources on fabricated intelligence. Accurate information came from detainees before they were subjected to enhanced interrogation, not after. The program never produced “imminent threat” intelligence: the ticking-time-bomb scenario its defenders constantly invoked simply never materialized.
Why Ordinary People Torture
If torture doesn’t work, why do people keep doing it? The question assumes that torturers are unusual: uniquely sadistic, or uniquely evil. The psychological research suggests the opposite. The most disturbing finding in the literature on torture is how ordinary the people who do it tend to be.
Stanley Milgram’s obedience experiments at Yale in the early 1960s remain the foundational demonstration. Milgram recruited ordinary men and told them they were participating in a study on learning. Their role was to administer electric shocks to a “learner” (actually an actor) whenever the learner answered a question wrong. The shocks increased in intensity. The learner screamed, begged, mentioned a heart condition, and eventually went silent. An experimenter in a lab coat calmly told the participant to continue.
Sixty-five percent of participants went all the way to the maximum voltage: what they believed were 450-volt shocks to a person who had stopped responding. They were not enjoying it. Many were visibly distressed, sweating, trembling, laughing nervously. They did it anyway. The authority figure told them to, and the institutional setting (a prestigious university, a man in a lab coat, the language of scientific necessity) was enough to override their own moral judgment.
Milgram designed the experiment in the shadow of the Holocaust. He wanted to understand how ordinary Germans had participated in mass atrocity. His answer was uncomfortable: they were not uniquely evil. Given the right institutional scaffolding (authority, legitimacy, incremental escalation, diffusion of responsibilityA social psychology phenomenon where individuals in a group feel less personally accountable for an outcome because culpability is spread across multiple participants.) most people will inflict serious harm on others.
Philip Zimbardo’s 1971 Stanford Prison Experiment extended this logic, assigning students randomly to the roles of “guards” and “prisoners” in a simulated prison. Within days, guards were stripping prisoners, forcing them into degrading positions, and using sleep deprivation as punishment. The experiment was later revealed to have serious methodological problems, including evidence that Zimbardo actively coached the guards toward abusive behavior. But the core insight, that institutional roles and situational pressures can rapidly transform behavior, has been confirmed by subsequent research in less contrived settings.
When photographs from Abu Ghraib emerged in 2004 showing American soldiers forcing Iraqi prisoners into naked human pyramids, dragging them on leashes, and posing with them like trophies, Zimbardo saw his experiment playing out in real life. In The Lucifer Effect, he argued the Abu Ghraib guards were not “bad apples” but products of a “bad barrel”: an institutional environment where dehumanization was routine, oversight was absent, and abuse was tacitly encouraged by leadership.
The Institutional Machine
Darius Rejali’s Torture and Democracy (2007), the most exhaustive study of how torture operates in modern states, identifies the mechanisms that keep torture programs running once they start:
Escalation is built in. An interrogator who uses moderate coercion and gets ambiguous results doesn’t conclude the method is flawed. They conclude they haven’t gone far enough. The logic of escalation is internal to the practice: if a little pain didn’t produce information, more pain might. This is neurologically backwards (more stress means less reliable memory) but psychologically intuitive, which is why it happens over and over.
Bureaucracies protect themselves. The CIA’s enhanced interrogation program didn’t just fail; the agency fabricated evidence of its success. This wasn’t a cover-up in the conventional sense. It was institutional self-preservation. The people who designed, authorized, and implemented the program had careers and budgets dependent on its continuation. Admitting failure would mean admitting that they had tortured people for nothing. Bureaucracies almost never make that admission voluntarily.
Responsibility dissolves. Modern torture programs distribute culpability so widely that no individual bears the full weight. The psychologists who designed the techniques said they were just providing expertise. The lawyers who wrote the legal memos said they were just offering analysis. The guards who carried out the methods said they were following orders. The administrators who approved the program said they were relying on expert judgment. Everyone was responsible. No one was responsible.
Group identity hardens. Torture units develop internal cultures where willingness to inflict pain becomes a test of loyalty and toughness. Refusing to participate marks you as weak or unreliable. Rejali found this pattern across dozens of countries and time periods: elite military and police units, precisely because of their insularity and esprit de corps, are especially prone to normalizing abuse.
The Democracy Problem
Rejali’s most unsettling finding is that democracies didn’t just tolerate state torture in the twentieth century. They innovated it.
As public scrutiny, free press, and legal accountability increased, democratic states developed “clean” techniques: methods designed to leave no marks. Stress positions. Waterboarding. Sleep deprivation. Temperature extremes. Sensory deprivation. Psychological manipulation. These techniques proliferated specifically because democratic governments needed deniability. A dictatorship can torture openly; the spectacle is part of the point. A democracy needs methods that allow officials to stand at a podium and say “we don’t torture” without technically lying.
The legal memoranda produced by the Bush administration’s Office of Legal Counsel illustrate this perfectly. The so-called “torture memos” redefined torture so narrowly (requiring “severe physical suffering” equivalent to “organ failure” or “death”) that virtually anything short of killing the subject could be classified as something else. The point was not to legalize torture. It was to make it linguistically invisible. “Enhanced interrogation.” “Coercive measures.” “Special treatment.” The euphemisms have been remarkably consistent across decades and languages: the French in Algeria used strikingly similar terminology in the 1950s.
What Torture Actually Does
If torture doesn’t reliably extract information, what function does it serve? The historical record suggests three primary ones.
Terror as governance. Torture is frequently aimed not at the victim but at the victim’s community. The French tortured in Algeria to suppress a population, not to gather intelligence on specific operations. Latin American juntas tortured dissidents to deter opposition. The message is not “tell us what you know” but “this is what happens.” The CIA’s Cold War interventions trained security services across the developing world in these methods, and the purpose was political suppression, not intelligence collection.
Performance of power. After September 11, the United States was in a state of institutional panic. The torture program was partly a way for the national security apparatus to demonstrate, to itself as much as to anyone, that it was doing something. The “something” didn’t need to work. It needed to feel decisive. The CIA contracted Mitchell and Jessen for over $80 million, not because their methods had a track record of success (they didn’t) but because the program signaled seriousness in a moment when seriousness was the institutional currency.
Self-sustaining bureaucracy. Once established, torture programs generate their own justifications. Careers are built on them. Budgets depend on them. Ending the program means admitting it was wrong, and that admission has consequences: legal liability, professional disgrace, institutional reckoning. It is always easier to keep going and claim the program is working than to stop and face what you’ve done.
The Alternative That Works (and Why It’s Ignored)
Effective interrogation techniques exist. They are rapport-based: an interrogator builds a relationship with the subject, establishes trust, uses conversational techniques to elicit information. They require skilled, patient, linguistically competent professionals. They take time. They work.
The FBI, which opposed the CIA’s enhanced interrogation program from the start, obtained more actionable intelligence from the same detainees using standard interview methods. Professional interrogators have a saying that belongs above every intelligence agency’s door: “Torture is for amateurs.”
These methods are ignored not because they fail but because they don’t look like what people expect serious counterterrorism to look like. Building rapport with a suspected terrorist over weeks of patient conversation doesn’t satisfy the political demand for visible toughness. It doesn’t make for compelling testimony before a congressional committee. It doesn’t project the image of decisive action that institutional leaders crave in moments of crisis.
Torture persists not despite its failure but because of the institutional, psychological, and political incentives that have nothing to do with its stated purpose. Understanding this is the first step toward breaking the pattern: the problem is not that governments don’t know torture is ineffective. The problem is that effectiveness was never the point.
