The 36 questions to fall in love: what the study actually says

In January 2015, a New York Times essay (“To Fall in Love With Anyone, Do This”) made the “36 questions to fall in love” famous. Since then, the protocol resurfaces at dinner parties, on dating apps and in magazine listicles, usually distorted. Yet the original study, published in 1997 by Arthur Aron and his team, is more interesting than its legend: it shows, with numbers, that you can manufacture closeness between two strangers in forty-five minutes, provided you respect one precise mechanism.

The real protocol: 3 sets, 45 minutes, an escalation

In the experiment, pairs of strangers receive three sets of questions on cards: 36 questions in total, 15 minutes per set. The first set stays accessible (“Would you like to be famous?”), the second digs deeper (“What is your most treasured memory?”), the third exposes (“What, if anything, is too serious to be joked about?”). Each person answers each question, taking turns. The control group spends the same 45 minutes on small talk.

The result: on the closeness scale used (IOS, two circles overlapping more or less), the escalating-disclosure condition averaged 4.06, against 3.25 for small talk, a gap considered large by psychology's standards (d = 0.88). More striking: the closeness felt after those 45 minutes exceeded the closeness that 30% of comparable students attributed to the closest relationship in their life.

What the study does not say

Media success grafted three elements onto the study that aren't in it. Setting the record straight doesn't weaken it. It shows what actually matters.

• “Falling in love”: the phrase appears nowhere in the paper, whose subtitle reads “Some Preliminary Findings”. The protocol creates felt closeness, not a couple.
• The “4 minutes of eye contact”: absent from the 1997 protocol. That was an addition in the New York Times essay.
• The couple who got married: an anecdote mentioned about a 1991 pilot version, not a measured result of the published studies.

Why it works: the mechanism, not the questions

Studies 2 and 3 of the paper are the most instructive, and almost never cited. The researchers tested what changes the outcome: matching pairs by attitudes? No effect. Leading participants to expect mutual liking? No effect. Making “getting close” an explicit goal? No effect on final closeness. The only thing that matters is the nature of the task: reciprocal, escalating, structured disclosure.

The mechanism is confirmed across the literature. Collins and Miller's meta-analysis (1994) shows we like people who confide in us, and like even more the people we have confided in. Sprecher's experiments (2013) add that turn-taking is essential: an intimate monologue creates no bond; alternating exchange does. And Reis and Shaver showed that intimacy arises when the other person's response makes you feel understood and accepted, hence the importance of the setting.

A delightful detail from study 3: when closeness is set as the goal of the moment, introverts end up just as close as extroverts. The format levels what personality separates.

The obstacle isn't awkwardness: it's predicting awkwardness

If disclosure works so well, why does nobody do it spontaneously? A series of twelve experiments published in 2022 (Kardas, Kumar and Epley) answers precisely: because we mispredict. Participants massively overestimated how awkward a deep conversation with a stranger would be, and underestimated how connected they'd feel. The identified mechanism: we underestimate how interested the other person really is in what we have to say. Even finance executives at a conference, asked point blank to describe “the last time you cried in front of another person”, came out more connected and happier than they had predicted.

That's exactly why an external structure changes everything: when the questions are provided and everyone plays along, nobody has to cross the barrier of the first personal question alone.

How to use it at an event

The point is not to have your participants recite the 36 questions: research shows the exact questions hold no magic. What matters is reproducing the protocol's three ingredients.

• Escalation: start accessible, climb gradually. An exposing question asked too early repels; asked at the right moment, it bonds.
• Reciprocity: everyone answers every question, taking turns. No one-way interviews.
• Structure: written questions, provided by the event, that legitimize the disclosure. Collins and Miller's meta-analysis shows that confiding in strangers without a frame backfires.

That's the principle behind the questions affina hands to every group: they are calibrated across three phases (warm up, build, reflect), reproducing the escalation of Aron's protocol, with depth dosed question by question. The 2E model that structures the product makes progressive disclosure (the expressive E) one of the two ingredients of connection; the 36 questions study is its best-known experimental validation.