How 'Confirmation Bias' Can Lead to War

A possible reason for why an erroneous report about Iran and the killing of Israeli tourists in Bulgaria was repeated in American and Israeli media.

Last week Commentary reported that Iranian President Ahmadinejad had been "bragging about the slaughter of five Israeli tourists" in Bulgaria and that this bragging "contradicted" the Iranian government's denials of involvement in the Bulgarian bus bombing.

Commentary had gotten this information from The Times of Israel, which reported that Ahmadinejad had "gloated publicly on Thursday over the deaths of Israelis in a terror bombing in Bulgaria, and hinted that Iran was responsible for the attack." The Times of Israel in turn attributed this information to a report in Hebrew on Israel's Channel 2.

Somewhere in this Persian-to-Hebrew-to-English translation, something got lost--or added. Iran may or may not be behind the Bulgarian bombing, but there's no reference to the bombing in Ahmadinejad's speech, and a close appraisal of the speech makes it highly unlikely that Ahmadinejad meant to allude to the bombing.

Nima Shirazi, the blogger who first raised doubts about the Israeli interpretation of Ahmadinejad's remarks, calls the distortion "propaganda." But what seems to me more likely--and, in a way, more unsettling--is that the distortion wasn't intentional, but rather was the result of an essentially unconscious warping that comes naturally to humans.

Specifically, I'm betting that the culprit was "confirmation bias," the tendency of people to see evidence consistent with their pre-existing beliefs (sometimes when it isn't even there) and to ignore or minimize evidence inconsistent with their beliefs. Confirmation bias is at work every day, in Israel and Iran and the United States, often in ways that make war more likely. What follows is the dissection of a single, cautionary case of natural self-deception.

Let's start with the report in the Times of Israel, an English-language website based in Jerusalem. Gabe Fisher of the Times wrote:

Speaking hours after Israel's Prime Minister Benjamin Netanyahu had publicly blamed the bombing Wednesday at Bulgaria's Burgas airport on "Hezbollah, directed by Iran," Ahmadinejad described the attack as "a response" to Israeli "blows against Iran."

That's inaccurate, since Ahmadinejad didn't "describe" the attack--or mention it at all--in his speech. But Fisher may not have known that, since he seems to have never seen a complete transcript of Ahmadinejad's speech. The only source he cited was a report on Israel's Channel 2, whose on-air correspondent said that Ahmadinejad "couldn't have provided a clearer announcement than this" of Iranian involvement in the bombing. Maybe Fisher mistook this confident surmise about Ahmadinejad's subtext for a description of Ahmadinejad's text.

A more subtle but perhaps more consequential distortion was a shift in tense that took place in the course of translating Ahmadinejad's remarks. Here is the way Fisher rendered, in English, Channel 2's Hebrew version of the remarks:

"The bitter enemies of the Iranian people and the Islamic Revolution have recruited most of their forces in order to harm us," he said in a speech reported by Israel's Channel 2 TV (Hebrew). "They have indeed succeeded in inflicting blows upon us more than once, but have been rewarded with a far stronger response."

The phrase "have been rewarded with a far stronger response" makes it sound like Ahmadinejad is referring to something in the past--like, say, a bus bombing in Bulgaria that happened the day before. But it turns out that, in the original Persian, this part of the speech didn't refer to the past. Here is the direct Persian-to-English translation provided by Shirazi, which renders the remarks in the future tense: "And if they deliver a blow they will receive a blow [and] usually the blow they receive will be heavier than the blow they have delivered." And, for good measure, here is a second Persian-to-English translation, provided (via email) by Bahman Kalbasi, a BBC correspondent who is a native Persian speaker: "And if anytime they harm us, we will harm them back. And usually the harm we cause is greater than the harm they cause."

It's conceivable that Ahmadinejad's actual remarks, though ostensibly focused on the future, could be a cryptic reference to the bus bombing. But the shift in grammatical focus that happened in translation--the shift from future to past--rendered such an interpretation much more plausible. So how did this happen? Was the change made by Channel 2, in the Persian-to-Hebrew phase, or by the Times of Israel, in the Hebrew-to-English phase?

I consulted with Kevin Osterloh, an historian at the Miami University (Ohio) who is an expert on Hebrew. He said that Channel 2's on-air correspondent, in rendering Ahmadinejad's remarks in Hebrew, had put them in the present tense, saying, roughly, "We are sustaining specific blows from the enemy, but... we are also successfully landing blows of our own; sometimes the blows which we land are even more forceful than their blows." Osterloh said this needn't be considered a dramatic departure from the future tense; you might think of it as "present progressive-ish," he said, because "one can use the present tense in a general way to mean 'the way that we are wont to do things,' etc., and this nuance in fact makes more sense within the context of a threat."

OK, so Channel 2's on-air correspondent didn't do the lion's share of the tense shifting. However, the Channel 2 website also ran a printed excerpt from Ahmadinejad's speech, also in Hebrew, which did clearly change the tense of the remarks, moving the action into the past. It's this version of the remarks that the Times of Israel apparently translated from Hebrew to English, and that appeared in Gabe Fisher's story. (Osterloh independently translated the Hebrew to English and came up with a translation closely corresponding to Fisher's. Thus where Fisher wrote that Iran's enemies "have been rewarded with a far stronger response," Osterloh wrote that Iran's enemies "have earned in response a much more forceful counter-blow.")

Did Fisher listen to the Channel 2 broadcast, in which the translation is closer to accurate? Or did he just take the text from the Channel 2 website without clicking on the video player? I don't know. (I've emailed The Times of Israel, asking that this question be relayed to Fisher, whose coordinates I can't find. No reply yet.) But it would certainly be consistent with human nature--and an example of confirmation bias--to encounter both translations and, rather than wonder at the discrepancy and investigate it, go with the version more consistent with your pre-existing beliefs. To facilitate this choice, a rationale for doubting the accuracy of the less convenient translation might bubble up unbidden from your unconscious mind.

If this was indeed a case of confirmation bias, that doesn't mean that Fisher himself harbors dark suspicions about Ahmadinejad. In his capacity as a journalist he has a vested interest in Ahmadinejad's remarks being newsworthy. So it would be natural for him to enter the "research" process with a working hypothesis that he's emotionally attached to. Hence a form of confirmation bias that confronts all journalists, and I don't think any of us can honestly claim we always surmount it. Certainly I can't.

Note how this journalistic tendency can pave the way for war. When you've got Israeli readers who will click on stories about Ahmadinejad's "gloating," and American readers who will do the same, and Iranian readers who will click on stories about the malicious intent of America and Israel, then the natural workings of journalism will reinforce and amplify preexisting incendiary beliefs (though in Iran, of course, the press is less free and more subject to government influence, which brings problems of its own). So confirmation bias enters the system at two points. It motivates some readers to click on certain kinds of stories, and it encourages journalists to produce those kinds of stories even if they're misleading--and journalists don't even have to be bothered by conscious awareness of how they've abetted untruth!

Tense isn't the only thing that got garbled in translation. In Israel, as the Times reported, Ahmadinejad's reference to "harm" from its enemies was taken to refer to the killing of Iranian nuclear scientists, which has been widely attributed to Israel. This conveniently reinforced the belief that when Ahmadinejad alluded to Iranian retaliation, he was talking about the Bulgarian bombing. After all, there had already been speculation that, if the bombing was sponsored by Iran, this was probably retaliation for the murder of Iranian scientists.

But in fact Ahmadinejad doesn't mention the assassination of nuclear scientists. And Kalbasi says that if you look at the context of the quote--including an immediately preceding reference to the economic power of Iran's enemies--it seems very likely that Ahmadinejad was referring not to the assassination of Iranian scientists, but to economic sanctions. "It was very much a standard Ahmadinejad speech," Kalbasi said.

Kalbasi engaged in a twitter exchange with the Times of Israel, and, after I spoke with him, the Times, to its credit, ran a brief article quoting him and raising the possibility that it got things wrong (though not admitting error). But presumably that article wasn't ballyhooed on the home page, and it isn't linked to from the original, misleading article--and, anyway, it isn't quite as enticing as the article about Ahmadinejad's "gloating" was. At last check, the ratio of Facebook "recommends" between the original article and the quasi-correction was more than 5-to-1. What Mark Twain said in a much slower age is still the case: untruth can go "round the world while truth is pulling its boots on." And one reason is confirmation bias.

Kalbasi doesn't think the Israeli media's apparent distortion of Ahmadinejad's remarks was intentional. He agreed with me that this was probably confirmation bias at work. And he said he sees the same thing in the Iranian press--distortions of remarks by President Obama and Israeli leaders that may well be "innocent" in the sense that the person doing the distorting isn't conscious of it. Certainly in Iran and nearby countries you can find earnestly held but wildly implausible beliefs about Zionist machinations. In Saudi Arabia an obviously intelligent university professor once explained to me his theory of how 9/11 happened, and, though I forget the details, the subject of Israel definitely came up.

A striking thing about human self-deception is how diverse and subtle its sources can be. The classic form of confirmation bias is to choose the most convenient among competing pieces of evidence--as may have been the case if Fisher indeed was exposed to both the printed and audio versions of Channel 2's translations and chose the printed one. But look at some other elements of self-deception that seem to have been at play here:

1) Unreflectively narrowing the meaning of vague or ambiguous words. The word "harm" (or, depending on the translation, "blow") in Ahmadinejad's remarks had no clear, specific referent. But pegging the word to the assassination of Iranian scientists, and pegging its repetition to the murder of Israeli tourists, can happen pretty automatically if your mind is so inclined.

2) Accepting evidence uncritically. Commentary is the one player in this game of telephone that seems not to have distorted anything. Still, if Ahmadinejad's remarks, as misleadingly conveyed by Israeli media, hadn't fit so nicely into Commentary's world view, might they have been treated more skeptically? Certainly, in the case of Commentary--and in the case of all of us--you can imagine a translated quote from a world leader that would so violate our world view that we'd demand corroboration of the translation before trusting it. But evidence that is welcome gets a pass.

3) Making slight and essentially unconscious fudges. Translation is a complicated thing, and even something as straightforward-sounding as grammatical tense can be fuzzy. If you go into the process of translation thinking you already know what the speaker was saying, the fuzziness can grow.

Obviously, I don't know exactly what was going on in the heads of the players in this story, and this reconstruction has involved some speculation. But the main point is that the incendiary reports about Ahmadinejad's remarks in Israeli and American media were very likely wrong, yet it's highly plausible that at no point did anyone consciously choose to perpetrate distortion. That's the way human psychology is.

I want to emphasize that I think it's entirely possible that Iran is, in fact, behind the Bulgarian bus bombing. And, if so, retaliation for the assassination of Iranian scientists would indeed be a plausible motivation.

Still, two points are worth making:

First, given the way human psychology works, the thought that Ahmadinejad "gloated publicly... over the deaths of Israelis" may well be more conducive to war than the belief that Iran is responsible for the deaths. And this "gloating," which many Israelis now believe happened, apparently didn't.

Second, as we listen to Bibi Netanyahu and others assure us that they have overwhelming evidence of Iranian involvement, it's useful to keep in mind that confirmation bias is at work in politicians and intelligence analysts as well as in the rest of us. If we learned nothing else from the runup to the Iraq War, we should have learned that.

I'm sure that Colin Powell believed the things he told the UN Security Council about evidence of Saddam Hussein's possession of weapons of mass destruction, and I'm sure George Bush believed them, and I assume the people who briefed Powell and Bush believed them. But they weren't true. And now 100,000 people are dead, and Iraq is still a cauldron of violence.

After the Times of Israel story came out, Colin Kahl, a Washington national security analyst and former Obama administration official, tweeted a link to it along with the comment, "Iranian leaders are going to gloat their way right into a war." All told, a more accurate rendering of the moral of the Times of Israel's story is: "Israelis and Iranians and Americans may deceive themselves right into a war."

[Note, 7/26, 10:30 p.m.: The paragraph above that begins with "OK," has been changed to correct an inaccuracy. It originally read, "OK, so Channel 2's on-air correspondent didn't do the lion's share of the tense shifting. However, the Channel 2 website also ran a printed excerpt from Ahmadinejad's speech, also in Hebrew, which did clearly change the tense of the remarks (to the "present perfect" tense, which is kind of a misnomer, since it places the action clearly in the past--in this case: "have been rewarded with a far stronger response".) It's this version of the remarks that the Times of Israel apparently translated from Hebrew to English, and that appeared in Gabe Fisher's story."

The reason for the change is this: Although both the Times of Israel and Osterloh (the Hebrew expert I consulted) translated the Hebrew into the present-perfect tense of English, it turns out that there is, strictly speaking, no present-perfect tense in Hebrew. The Hebrew they translated was in the past tense--although, as Osterloh has now explained to me, it was "past tense with a present-perfect sense." In any event, the original point of this paragraph remains true: Channel 2's translation of the Persian into Hebrew took action that in Ahmadinejad's original remarks hadn't been placed the past (i.e. Iranian retaliation) and, by changing the tense, placed the action in the past. (Thanks to the commenter FuzzyFace, who pointed out that Hebrew has no present perfect tense even before Osterloh emailed me with the same point.)]