Does Torture Work?

by Brian Dunning
July 21, 2026

Sponsor an episode!

Today we’re going to have an objective and dispassionate conversation about a topic that is extremely distasteful: the use of torture to extract intelligence from a prisoner. The technique paints a permanent stain on its perpetrator, on the supervising agency, and on the government that authorized it; but some might argue that it is no stain at all, that these are people who deserved to be treated poorly, and who deserve to be put through whatever it takes to obtain what we need from them. Which is true? I argue that we can answer that with a simple test: Does torture actually work? Does it produce good intelligence, bad intelligence, or nothing at all? Fortunately, we have whole history books of answers to draw from.

First, a definition is in order. This discussion is limited to the use of torture as a way to get information from a prisoner. It is the intentional infliction of physical or mental pain or suffering with the express purpose of extracting information, though the term is also used when that pain or suffering is inflicted for other purposes, such as punishment or intimidation — or even just deranged jollies for the perpetrator. Those other purposes are outside the scope of this episode. Within certain legal frameworks, the term torture only applies when the perpetrator is a public official or is acting in a legal capacity; however, for our purpose, we don’t necessarily care about this. If a mob boss is torturing someone to find out where they hid the cash, that also falls within the scope of our question. Is torturing an effective way to get information from an unwilling person?

A familiar example that often comes to mind is the infamous Abu Ghraib prison scandal of 2003-2004, in which certain American soldiers and intelligence personnel committed numerous human rights violations against Iraqi detainees. It was inhumane, degrading, humiliating, and brutal; but the best known of the abuses did not take place within the context of interrogation. Many of the actions were, at best, done in advance of anticipated interrogation to “soften them up.” Some of the victims were even people with no military intelligence value. So legally, the majority of Abu Ghraib fell under definitions of torture other than the one we’re interested in, and it would not be included in our discussion today.

So you’ve got your terrorist in custody, and you need to know where the bomb is and when it’s going to go off. How do you get him to talk, because he is (obviously) highly motivated not to give you any helpful information at all. Let’s take a quick look at the most common of your basic options:

• Building a rapport. This was made famous in World War II by the Luftwaffe interrogator Hanns Scharff, who used genuine kindness, gave genuine gifts, and never lied or tricked his subjects. The amount of valid intelligence he collected from Allied prisoners was legendary. His go-to method was to have long, friendly, honest conversations about life, the war, and everything; and occasionally he would drop a bit of misinformation which the prisoner would often correct. Today this has been formalized around the world as the PEACE method (Prepare, Engage, Account, Closure, and Evaluate).
  
  
• Structured confrontation. Hollywood has shown this in virtually every cop movie in every interrogation room. The rewards of talking and the consequences of silence are both explained, and the good cop/bad cop technique is a staple. Today it’s called the Reid Technique, and involves everything you see in movies: asking the same question many times, repeating the prisoner’s story making it sound really bad, asking loaded and leading questions, all designed to drive the prisoner to seek refuge in the good cop’s promised rewards.
  
  
• Truth drugs. Mostly tried during World War II, a variety of compounds were tried to get people to talk, including sodium amytal, sodium pentothal, scopolamine, and even an injected marijuana extract. The idea was to put the subject into a deeply relaxed state in which they would lose their inhibitions; and this part of it worked. However the subjects would give all kinds of mixed-up combinations of fact, fiction, and fantasy. At their best, they impair the subject’s ability to lie effectively. No truth drug was ever found to be reliable enough.
  
  
• Torture. Also known by friendly names such as “enhanced interrogation” and currently in use all around the world. Basically, it’s beating up your guy until he talks and can include all imaginable varieties of physical and psychological torment. Legally, torture violates all kinds of laws everywhere, including the UN Convention Against Torture, the Geneva Conventions, and also United States federal law. But if the bomb’s going to go off and this guy is responsible, some might argue torture is justified if it works. Today’s question is whether it does.

(Incidentally, I consider happywashed terminology like “enhanced interrogation” to be an insult to the intelligence, so I’m not going to use such terms today. Torture is torture. It’s not a pretty thing, and I see no justification for making it palatable through happywashing.)

So let’s have a look at the evidence in favor of torture being an effective way to get good, truthful information from a prisoner. This comes mainly from the officials in charge of the clandestine services who have employed it, and who favor its continued usage. Not a very impartial source, to be sure; however, this is also something you can’t do controlled testing on as it would clearly be unethical. Research on torture effectiveness has to rely on observational studies, meaning studies that observe actual things in the real world that are happening anyway, regardless of the research.

The principal advocate for torture in recent decades in the United States was Jose Rodriguez, Director of the CIA’s National Clandestine Service from 2004 to 2007. He stated that the most valuable information came from the highest-value prisoners such as Khalid Sheikh Mohammed and Abu Zubaydah, but only after they were tortured. This intelligence, he said, fed into the hunt for Osama bin Laden. Based on a number of such reports, President George W. Bush approved the continued use of torture (though Obama did later put a stop to it with Executive Order 13491 in 2009, two days after taking office). However the CIA continued to be a fan, with Director John Brennan echoing Rodriguez’s assessment in 2014, saying that torture “did produce intelligence that helped thwart attack plans, capture terrorists, and save lives.”

However, the evidence against torture being effective ends up being a much larger body of work, and it’s backed up by plenty of research. Torture produces information, yes; but information of completely unreliable quality. Prisoners will say anything to stop the torture. They’ll say whatever they think you want them to say. But even if they wish to be truthful, their ability to do so is severely impaired. Professor of experimental brain research Shane O’Mara published the book Why Torture Doesn’t Work: The Neuroscience of Interrogation in 2015, which examined the biomedical literature to see what torture actually does to the brain. Among its effects are physiological changes that blur and break down memories, and often replace them with false ones. The more “broken” the prisoner, the more broken (literally) his memories.

But the evidence extends far beyond published studies. Perhaps the most glaring single example was false information gathered through the torture of detainee Ibn al-Shaykh al-Libi indicating a link between Iraq and al-Qaeda, which was then infamously used as a large part of the justification for the 2003 invasion of Iraq. CIA director George Tenet later explained that the “weapons of mass destruction” story was fabricated in order to get the American public to support an invasion that was actually based on this intelligence. It is likely history’s worst example of torture producing false intelligence which led to the loss of many human lives.

The Senate Intelligence Committee conducted a five-year investigation into the use of torture and, in 2014, presented their findings that it is “not an effective means of obtaining accurate information.” In reaching this conclusion, one of their lines of investigation examined each of the 20 specific cases that the CIA had cited as its best successes. It turned out to have been a house of cards. In each and every one of the 20 cases, the Committee determined that the information obtained actually came from one of three sources:

1. It was obtained before torture was applied;
2. The information was unrelated to the CIA’s claimed outcome; or
3. The information had already been known from other sources.

And they put a nice cherry on top of their report, finding that the CIA had given Congress, the President, and the public “extensive inaccurate information” about their torture program having produced accurate and actionable results.

But just to be clear, by no means is the CIA a monolithic pro-torture machine. Multiple CIA directors, including Leon Panetta (2009-2011), Mike Pompeo (2017-2018), and William Burns (2021-2025) vowed not to reinstate torture. These included Gina Haspel (2018-2021), even though her own record as a CIA station chief in the 2000s included personally supervising torture and the destruction of videotapes documenting it. And of course we have others like George Tenet (1996-2004) who initiated the CIA’s use of torture, and John Brennan (2013-2017) who stopped short of reinstating it but always insisted it was a valuable tool.

Obviously the CIA has the same information everyone else has, which is that the conclusion is pretty clear that torture is one of the worst ways to get good information. So we’re forced to wonder: Why, then, have so many within the CIA been its primary advocates?

As with so many questions, this one has multiple complicated answers that are hard to boil down. The Senate Intelligence Committee’s report looked into it, and found that one of the reasons was simple political self-protection. Many CIA officials engaged in torture, and so have both legal and reputational reasons to justify their actions with claims of success. For some there have been financial incentives; for example, James Mitchell and Bruce Jessen were the psychologists who designed the “enhanced interrogation” methods and were paid some $80 million for it through a company they set up. They were called upon to testify as to the program’s effectiveness, so answering “yes” was the best way to navigate the obvious conflict of interest — which also cut the same way for the CIA officials who had hired them.

An element of practicality also comes into play. Training people how to torture someone is relatively easy, and the bar for success is very low: Yes, you did harm; nice job. Training someone to become an effective expert in rapport-based methods, and to have them actually become good at it, is much more involved, and the majority of trainees are never going to be successful at it as it requires some pretty rare talents. So perpetrators tend to do what’s easiest rather than what’s hardest.

There are also interesting psychological aspects at play that lead a perpetrator of torture to honestly overestimate its effectiveness; we cannot automatically assume that any CIA official defending it is being dishonest. For one thing, there is a natural tendency for the perpetrator to believe the prisoner is deserving of punishment, creating an inherent righteous justification for concluding the procedure was effective and then reporting it as such. Confirmation bias also comes into play: everyone gets excited when one case provides good information and the bomb is found in time, but they forget the ten cases where they were sent on a wild goose chase by misinformation. Confirmation bias is truly a phenomenon that cuts through just about every institution and process worldwide.

Life offers us lots of complicated problems, and for the most part, they have complicated solutions — like learning to be as good as Hanns Scharff was at rapport-based interrogation. But as we all do so often, some seek what appear to be magically easy solutions to these complicated problems (in this case, torture) and we learn the hard way that they simply don’t work. Torture is a way to get information, but whether it’s true is a roll of the dice.

By Brian Dunning
Follow @BrianDunning