Confirmation Bias

A person in uniform standing at a podium with a microphone in front of a flag Description automatically generated with medium confidence

Air Force Lt. Gen. Sami D. Said briefing the press on the August 9 drone strike against a Toyota Corolla in Kabul, Afghanistan, November 3, 2021. Photo: U.S. Defense Department.

The NYTimes’ recent series of articles exposing the civilian cost of the U.S. wars in Iraq, Syria and Afghanistan should prompt public fury. Hundreds of pages of previously unpublished Pentagon documents refute claims by military planners and politicians that air strikes against Taliban and Isis fighters have been conducted with concern for the protection of civilians. In fact, hundreds if not thousands more civilian deaths have occurred than previously acknowledged. Compounding these omissions and evasions is the fact that even in the rare cases when mistakes were acknowledged, little or no investigation followed. No U.S. civilian or military officials have been punished, reprimanded, or even re-trained in response to these myriad casualties, a large percentage of which were children. And almost no compensation has been paid to the families of victims, or to survivors forced to live with disabling injuries.

But essential as the reporting is, it nevertheless sits uneasily with me. In fact, I’m infuriated by it, in particular its focus on “confirmation bias”: The psychological tendency, as Times reporters summarize it, “to search for and interpret information in a way that confirms a pre-existing belief.” Examples highlighted in the articles include the mistaken belief that “people streaming toward a fresh bombing site…are ISIS fighters, not civilian rescuers,” or that “men on motorcycles” are fighters in formation, not just random men on motorcycles. In another example cited by the Times, intelligence reports of an alleged car bomber driving a “darkly-colored heavily armored vehicle” were used by an “air support coordinator” to justify destroying an unarmored blue car and a white van following it. Seven people were killed, all of them civilians instructed by Isis to flee the area. One of the dead was an infant child on his mother’s lap. That’s not “confirmation bias,” that’s manslaughter.

In Florida (where I live), manslaughter (stat. §782.07) is defined as: “The killing of a human being by the act, procurement, or culpable negligence of another, without lawful justification.”  There are usually under 100 cases per year of negligent manslaughter in Florida. They involve vehicular negligence, negligent handling of a weapon by an adult, children playing with guns and “34 other negligent killings.”

Under federal law (62 stat.756):

“Manslaughter is the unlawful killing of a human being without malice. It is of two kinds:

Voluntary—Upon a sudden quarrel or heat of passion. Involuntary—In the commission of an unlawful act not amounting to a felony, or in the commission in an unlawful manner, or without due caution and circumspection, of a lawful act which might produce death.”

Murder and manslaughter in the U.S. are usually charged in state, not federal courts. The exceptions include cases in which a federal official is killed, when the crime occurs on federal land, or during the commission of a bank robbery.

Under the Uniform Code of Military Justice, the crime (article 119) of manslaughter is defined in a similar way:

1. Any person subject to this chapter who, with an intent to kill or inflict great bodily harm, unlawfully kills a human being in the heat of sudden passion caused by adequate provocation is guilty of voluntary manslaughter.

2. Any person subject to this chapter who, without an intent to kill or inflict great bodily harm, unlawfully kills a human being—

3. by culpable negligence; or (2) while perpetrating or attempting to perpetrate an offense…is guilty of involuntary manslaughter and shall be punished as a court-martial may direct.

Manslaughter is not a specifically designated offense under international law. But there are clearly defined “war crimes” that correspond to manslaughter as defined above. They are described under Article Eight of the Rome Statute of the International Criminal Court and include:

(iv)  Intentionally launching an attack in the knowledge that such attack will cause incidental loss of life or injury to civilians or damage to civilian objects.

(v)  Attacking or bombarding, by whatever means, towns, villages, dwellings or buildings which are undefended, and which are not military objectives.

(The U.S., along with with China, Iraq, Israel, Libya, India, Qatar, and Yemen, is not a party to the Rome statute, though it does sometimes cooperate with the Court.)

The recent NYTimes articles reported case after case of the U.S. military launching strikes certain to cause “loss of life or injury to civilians”; attacks that arose from “culpable negligence” or lack of “caution and circumspection”; and which lacked “lawful justification.” Under state, federal, military, and international law, the planners and perpetrators of these bombings are guilty of involuntary manslaughter or war crimes. Admittedly, I’m no attorney, and the applicability of the plain language found in these statutes would need to be tested in court. But the New York Times’ writers and editors are certainly not psychologists and neither is U.S. Central Command spokesman Capt. Bill Urban, who agreed with reporters that “confirmation bias is a real concern.”

The term confirmation bias was coined by British cognitive psychologist Paul Cathcart Wason in 1960 following a series of experiments in deductive logic. He asked participants to propose the simplest rule underlying the triple, 2-4-6. Most of them, according to his subsequent publication, failed to do so, supposing for example, that the rule concerned even numbers, or increases by two. In fact, the underlying principle was simply ascending numbers (for example, 1, 14, 32) demonstrating that the participants were trying to confirm their initial impression – thus, “confirmation bias.” And from there, the concept took off. It has been tested against theories of heuristics, Bayesian probability, information processing, cost-benefit analysis, and even evolutionary psychology. It is used today to explain fake news, vaccine hesitancy, stock market behavior, and now in the NYTimes, manslaughter or war crimes.

In fact, Mason’s original data showed that a significant number of his experimental subjects quickly deduced the rule underlying his triplet, and most of the rest figured it out on a second try. Follow-up research has tended to prove that most people are fully capable of challenging their initial beliefs; indeed, many are eager to do so, especially when they are rewarded for it. In other words, if a colonel tells a sergeant he will be commended for stopping drone strikes against innocent civilians, such attacks will likely diminish, possibly even end. Confirmation bias as such – a supposed in-born psychological tendency to follow first impressions – simply doesn’t exist. Decision making is always context specific and contingent on social, historical, ideological, and institutional factors.

In some cases, confirmation bias is a valuable heuristic. As cognitive psychologists Klayman and Ha argued, “positive” bias can be a useful and indeed necessary methodology to quickly arrive at a decision, especially in an emergency. If a middle-aged man suddenly grabs his left arm or clutches his chest and collapses to the floor, it’s a good idea for somebody to check his breathing, hand him an aspirin, and call 911. In the absence of other information, what else can we do except draw upon our previous experience and understanding? But positive bias can’t justify the actions of U.S. military and civilian leaders in Washington or drone operators at the Creech Air Force Base in Nevada. If they were to deploy positive bias in decision making, they would refrain from using drone strikes at all! As the NYTimes report indicated, there were literally hundreds of Pentagon records of deadly airstrikes on civilian populations in Iraq, Afghanistan, Syria and elsewhere. Confirmation or positive bias, properly applied by planners and drone operators, would warn them they were about to commit a war crime.

In fact, the evidence of war crimes and manslaughter committed by the U.S. military in the Middle East, Central America and Southeast Asia is voluminous. In Vietnam, some 600,000 civilians were killed by American forces, mostly from air bombing. U.S. sanctions against Iraq from 1990-1999 killed about 500,000 children. The current estimate of people in Afghanistan malnourished as a result of U.S. sanctions there is about 60%. According to the Cost of War Project of the Watson Institute at Brown University, total civilian deaths from all the post-9/11 wars is about 300,000, not including deaths from associated displacement and disease. In other words, the killing of civilians during the last two decades by drone strikes, bombings and other acts of war is an entirely predictable and indeed expected consequence of the policies pursued. They are not the product of confirmation bias – quite the contrary. They are the result of imperialism, indifference, and racist aggression. They are conscious acts of manslaughter, that is, war crimes deserving of the most severe sanctions our justice systems — state, national, military, and international — can deliver.

Stephen F. Eisenman is Professor Emeritus of Art History at Northwestern University and the author of Gauguin’s Skirt (Thames and Hudson, 1997), The Abu Ghraib Effect (Reaktion, 2007), The Cry of Nature: Art and the Making of Animal Rights (Reaktion, 2015) and many other books. He is also co-founder of the environmental justice non-profit,  Anthropocene Alliance. He and the artist Sue Coe and now preparing for publication part two of their series for Rotland Press, American Fascism Now.