The 36 questions: what the study actually asks you to do

You have probably seen the list. Thirty-six questions, in three escalating sets, passed around since 2015 as a formula for falling in love with a stranger. Someone always brings it up at a dinner party. Someone always says they tried it and it worked, or tried it and it was excruciating.

The list is real and it comes from real research. Most of what gets said about it does not.

What the study was

The source is a 1997 paper by Arthur Aron, Edward Melinat, Elaine Aron, Robert Darrin Vallone and Renee Bator, published in Personality and Social Psychology Bulletin. They were not trying to manufacture romance. They were trying to solve a practical problem for researchers who study close relationships: closeness normally takes months to build, which makes it almost impossible to study in a lab. They wanted a procedure that could generate it in an afternoon.

What they built is straightforward. Two people who do not know each other work through three sets of twelve questions, roughly fifteen minutes per set, forty-five minutes in total. They take turns, alternating who answers first. The questions escalate. The early ones are easy, the kind of thing you would answer at a bus stop. By the third set they are asking about your relationship with your mother and what you would regret not having told someone if you died tonight.

The control group got thirty-six questions too, matched for length and structure, but small talk throughout. That comparison is what makes the finding mean anything.

It worked. Pairs who did the escalating questions reported substantially more closeness afterward than the small-talk pairs, and the levels they reported were comparable to relationships they already had in their lives.

What it did not find

Here the popular version and the paper part company, and the gap is worth knowing.

The study measured closeness, generated quickly and measured immediately. It did not measure love, and Aron has said plainly that the procedure seems unlikely to produce loyalty, dependence or commitment, which take longer to develop. Closeness after forty-five minutes is a real thing, and it is not the same thing as a relationship.

The participants were college students recruited from psychology classes. One pair from the original research later married, which is a lovely story and is repeated as though it were a result. Out of everyone run through the procedure, one marriage is not a rate that demonstrates anything.

And the four minutes of silent eye contact that the popular version treats as the final step is not in the published procedure. It is not in the version the Greater Good Science Center at Berkeley publishes either. It came along with the retelling.

The study measured closeness generated in forty-five minutes. That is a real thing, and it is not the same thing as a relationship.

None of this makes the research less interesting. It makes it a different and better piece of information than the one that went viral.

The result nobody quotes

The 1997 paper reports three studies, and the two that get ignored are the ones worth your attention.

Having established that the procedure works, the researchers went looking for what would make it work better. They tested whether it helped to match pairs who agreed on important attitudes. They tested whether it helped to tell people in advance that the other person would probably like them. They tested whether it helped to name getting close as the explicit goal of the session.

None of the three produced a significant effect on closeness, and the authors report they had the power to detect effects of the size they were looking for.

Read that precisely. It is a null result for those three manipulations, in that setting, at the effect sizes the studies were built to catch. It is not a finding that shared attitudes or mutual liking never matter anywhere, and the picture is not uniformly flat: Study 3 turned up an interaction involving introversion and extraversion.

Bounded that way it is still striking, because it runs against how dating is usually organized. Matching the pair on attitudes did not raise closeness. Priming them to expect mutual liking did not. Naming closeness as the goal did not. What the procedure supplied instead was structure: sustained, escalating disclosure, taken in turns.

That is a result about method rather than chemistry. Two people who match on paper and want it to work did not get there faster by matching harder.

Why this is bad news for the feed

Most dating products are built almost entirely around variables that resemble the ones the study manipulated.

A filter resembles attitude matching. A compatibility percentage resembles a promise of mutual liking. An app announcing your best matches today resembles naming closeness as the goal. Aron and colleagues tested the lab versions of those three conditions and none of them raised closeness. They did not test filters, match scores, or any product feature, so the resemblance is ours to argue rather than theirs to have shown.

What the research says actually generates closeness is the one thing a feed of profiles is structurally incapable of providing: forty-five uninterrupted minutes of two people escalating on each other. Reciprocity cannot happen when one person is reading and the other is a photograph. We wrote about what that format does to your judgment in why a feed of strangers makes you feel like inventory, and about the exhaustion it produces in swipe fatigue.

To be careful about the inference: Aron and colleagues studied strangers in a lab in the nineties, not app users, and nobody has run this procedure as a test of dating-app design. What the study supports is narrower and still useful. In that setting, the form of the conversation did more for closeness than any of the matching done beforehand.

How to actually use the questions

If you want to try them, a few things follow directly from how the study was run.

Both people have to want to. The procedure was volunteers who signed up, sat down and agreed to the exercise. Producing a list mid-date and announcing that science says this will make us fall in love is a different activity with a different outcome.

Take turns. Aron's procedure had participants alternate, but it never compared alternating against the alternative. That comparison came later, from Susan Sprecher and colleagues in the Journal of Experimental Social Psychology in 2013. They put unacquainted pairs through structured disclosure either taking turns or with one person disclosing at length first, and the turn-taking pairs reported more liking, closeness and enjoyment. The advantage held even after the one-sided pairs swapped roles and got their turn. So one person disclosing while the other collects is not merely an untested variant, it is the arrangement that came off worse when someone checked.

Do not skip to the end. The sets escalate, and the third set arrives after the first two have done their work. Opening with the deepest question is a departure from the procedure that was tested rather than a shortcut through it.

Expect closeness, not commitment. You are running a procedure that reliably produces a real feeling in forty-five minutes. Treat that as a beginning worth having rather than a verdict. The feeling is genuine and it is early.

The part Bloom is built around

We think the mechanism generalizes past the first conversation, and that is most of what Bloom is.

Profiles on Bloom are built to be read rather than glanced past, because the first disclosure has to land on a person rather than a frame. The free plan gives you a set number of likes a day, since a bounded day is the condition under which anyone has attention left for a real conversation. Every preference filter is free and so is seeing who liked you, which is the industry's favorite paywall and one we refuse to use. We explain why on our pricing page.

And the questions do not stop mattering once you are together. Prompts in Relationship Mode are the same mechanism aimed at the relationship you already have: reciprocal, escalating, taken in turns, on an ordinary Tuesday rather than in a lab. John Gottman's research on couples puts weight on this kind of small repeated moment, though it studied established couples rather than new pairs, so it is a reason to think the mechanism keeps earning its place rather than proof that it does. We went through it in bids for connection and what makes a relationship last.

Forty-five minutes

The honest summary of the 1997 research is less romantic than the version you have heard and more useful. Two strangers can get meaningfully closer in forty-five minutes. Whether they had been matched on important attitudes, whether they expected to like each other, and whether they were trying to get close all mattered less than whether they answered escalating questions in turn.

You do not need the list to do that. The list is scaffolding for something most people already know how to do and rarely get the conditions for.

Bloom is pre-launch, starting in Utah along the Wasatch Front. The only thing to do today is join the waitlist and grow with us.