A bloke at a barbecue once spent eleven minutes explaining why I was wrong about inflation. He had numbers. He had a chart on his phone, which he turned towards me at intervals, like a policeman with a warrant. By the end I believed exactly what I had believed at the start, only harder, and I had also formed a settled view about him.

Two experiments out of Stanford suggest that what would have worked was one sentence and a bit of actual curiosity.

What the experiment involved

In 2010, Frances Chen, Julia Minson and Zakary Tormala published a short paper in the Journal of Experimental Social Psychology reporting two studies on what they called expressed interest. Chen and Minson were in Stanford’s psychology department at the time; Tormala, who studies attitudes and persuasion, is at the Stanford Graduate School of Business.

The first study ran on a small deception. Fifty-six undergraduates who opposed a proposal to make comprehensive exams a graduation requirement were told they had been paired for an online chat with a student who backed it. No such student existed; every reply was scripted.

Half the participants received a block of pro-exam arguments. The other half got the identical block plus one line, in which the imaginary partner said he was interested and asked them to say more about why they thought that.

That was the entire intervention.

Participants who got the question rated their invisible opponent as considerably more open-minded, rated themselves as more receptive, and thought better of exam supporters as a group. Two judges who read the participants’ written replies, knowing nothing about who had received what, scored those replies as more receptive as well. The question changed how people behaved, not only how they felt.

Their opinion of the exam policy did not move at all.

So who actually changes

The second study turned the experiment around and pointed it at the asker. Forty-nine students who disagreed with the case for universal veganism watched a two-minute video of a campus animal rights advocate making it. One group was told to jot down comments comparing the speaker’s view with their own, which is more or less what people do anyway. The others had to prepare three open-ended questions aimed at understanding why he felt the way he did.

The questioners never asked them. No answers came back, because there was nobody to answer.

They still finished the video rating the speaker’s conclusions as more valid, thinking better of vegans generally, and warmer to the idea of a real conversation. Composing the question was sufficient. On the measure that matters most, movement on the issue itself, the askers shifted and the asked did not.

Which is a strange and faintly deflating result if you were hoping to use curiosity as a crowbar on somebody else.

Why a question does something an argument cannot

An elaboration question is a request to expand on what you just said. The authors were careful to distinguish it from a rhetorical question or a trap laid to catch you in a contradiction, which is what most questions in an argument actually are.

As for the mechanism, the paper offers candidates and admits it tested none of them. Writing a question may force you to process what the other person said on its own terms instead of scanning it for soft spots. It may switch on perspective-taking. Or it may be plain self-perception: you observe yourself behaving open-mindedly, conclude that you are an open-minded person, and then go on to act like one.

I find that last one uncomfortably convincing, because I recognise the reverse. Spend fifteen years in meetings assembling your rebuttal while somebody else is still talking and you quietly become a man who does not listen. I only noticed once I stopped having the meetings.

The same group has produced other findings that cut against the standard advice. In a 2013 paper in Psychological Science, Chen, Minson and two colleagues in Germany found that more direct gaze at a speaker’s eyes predicted less attitude change, and that people told to watch the speaker’s mouth instead were more persuaded. So much for looking them straight in the eye.

Later work by Michael Yeomans, Minson and colleagues, published in Organizational Behavior and Human Decision Processes, built an algorithm that detects what receptiveness sounds like on the page. In their field data, Wikipedia editors who opened a disagreement receptively were less likely to be personally attacked. That team went on to publish a tool that scores a draft for how receptive it reads, which I ran an old email through and would rather not discuss.

How much weight this can carry

One paper is one paper, and this one is small. Two samples of undergraduates, fewer than fifty usable participants each, arguing about graduation exams and veganism rather than abortion or immigration. Chen and her co-authors said so themselves, calling for the interventions to be tried in genuinely contentious settings, ethnic and geopolitical conflicts included.

A test on harder ground arrived in 2025, and the news was mixed. A preregistered field experiment by Erik Santoro, David Broockman, Joshua Kalla and Roni Porat, written up by Berkeley’s College of Letters and Science, tried receptive listening on American attitudes towards undocumented immigrants. Participants who were listened to attentively before hearing a personal story about an immigrant were no more persuaded than those handed the story cold. Broockman offered two ways to read it. Either listening feels good and does nothing, or people are less manipulable than assumed and are weighing arguments on their merits, which he called the cheerier reading.

So the technique is oversold as a lever on other people. As a way of stopping yourself from becoming the bloke with the chart, the evidence is sturdier, and it points inward.

I have started asking at barbecues. It has never once won me an argument, and it is the only move in my repertoire that has ever taught me anything.