Almost every love story contains a piece of luck that nobody bothers to credit. A seat allocation. A flatshare that fell through. A transfer to the wrong department. People tell these stories as though the connection was the engine and the circumstance was scenery, and the order feels obvious from the inside.

Social psychology has spent about seventy years testing that assumption. The results are old, well replicated, and mildly deflating: sheer nearness, with nothing else going for it, is enough on its own to move people toward liking each other. None of the studies below set out to rank proximity against looks or shared interests. What they show instead is that nearness alone, stripped of everything else, still does real work.

The classic demonstration came out of postwar housing. Leon Festinger, Stanley Schachter and Kurt Back surveyed married graduate students living in the Westgate West complex near MIT and published the results in 1950 as Social Pressures in Informal Groups. Residents had been assigned units more or less arbitrarily. Asked to name their closest friends in the complex, roughly two thirds named people living within a few doors of them, and the residents who happened to live near stairwells and mailboxes accumulated more friends than those tucked at the ends of corridors.

Traffic patterns, not temperament, were doing much of the sorting.

What the seating studies show

The pattern held up when researchers found cleaner ways to test it. Mady Segal, writing in the Journal of Personality and Social Psychology in 1974, studied recruits at a Maryland state police training academy where dormitory rooms and classroom seats were allocated alphabetically by surname.

Friendship choices tracked the alphabet.

Mitja Back, Stefan Schmukle and Boris Egloff pushed it further in a short report in Psychological Science in 2008, finding that randomly assigned seating in a single introductory university session still predicted who liked whom a year later. And a large randomised trial published in PLOS One in 2021 shuffled the seating charts of 182 primary and middle school classrooms, covering 2,966 students, and found that being seated next to someone raised the chance of a mutual friendship from about 15 per cent to 22 per cent over one semester.

These are studies of friendship among schoolchildren and undergraduates, measured over a semester or a year, in settings where somebody else controlled the seating plan. None of them followed anyone to a wedding.

Familiarity does more of the work than chemistry

The most quietly unsettling piece of this literature is Richard Moreland and Scott Beach’s 1992 experiment, published in the Journal of Experimental Social Psychology. Four women of broadly similar appearance sat in on a large college course, attending zero, five, ten or fifteen sessions. They were instructed not to speak to anyone. At the end of term, 130 students were shown slides of the four and asked to rate them.

Attractiveness ratings climbed with attendance, from roughly 3.6 for the woman nobody had seen to about 4.4 for the one who had sat there fifteen times. The authors also found that repeated exposure raised perceived similarity, and that the similarity judgement appeared to follow the attraction rather than produce it.

That last detail is the one most likely to be skipped. Shared passions are the reason people usually give for a connection, and in this study the sense of having something in common arrived after the liking did, on the basis of nothing but presence in a lecture hall. Which suggests that some of what we experience as recognising a kindred spirit may be a story assembled after the fact, from materials that repetition supplied.

Where the effect runs out

Repetition is not a lever that keeps working. R. Matthew Montoya and colleagues pulled together 268 curve estimates from 81 articles for a 2017 meta-analysis in Psychological Bulletin, and reported that liking rises with exposure, peaks, and then falls away. The relationship is an inverted U, not a slope. Familiarity buys goodwill up to a point and then begins spending it.

None of it forecasts a single relationship, either. All of this describes averages across groups, not individuals, and most of the evidence concerns liking rather than the thing people mean when they say falling for someone. Proximity is better understood as what decides the pool than as what decides the choice. Sitting near somebody for a semester does not make them your future spouse. It makes them eligible, which is not a small thing, because almost nobody outside the pool ever gets considered at all.

Proximity has changed shape

The obvious objection is that none of this survives the smartphone. Michael Rosenfeld, Reuben Thomas and Sonia Hausen worked from a nationally representative 2017 survey of American adults that asked respondents how and when they had met their partners, reaching back across earlier decades. Analysing those retrospective answers, they reported in PNAS that meeting online had overtaken meeting through friends as the most common way American heterosexual couples met by around 2013, while meeting through neighbourhood, family and church had been sliding since the Second World War. The data is self-reported and specific to the United States.

Physical nearness clearly matters less than it did in 1950. But the apps did not retire the underlying mechanism so much as rehouse it. A feed is repeated exposure with a filter attached, and whose face reappears this week is now decided partly by ranking code and algorithm rather than by a housing officer with a clipboard. The corridor is still there. It just belongs to someone else.

Most people can describe the sort of person they are looking for with some confidence. Rather fewer could say who chose the rooms they spent a year sitting in.