You Don't Fall for Someone Like You. You Fall for Someone You Think Is Like You.

Last week we put numbers under one of dating's oldest questions: opposites don't attract — across a century of couple research, people who end up together are alike, overwhelmingly. This week, the twist that a careful reader might already have spotted hiding in that piece. Because “couples are alike” is a fact about who ends up paired. It says nothing about what actually pulls two people toward each other in the first place — and when researchers isolated that question directly, the answer came back stranger and more human than the census: what predicts attraction is not being similar. It is believing you are.

The Study That Separated the Two

The anchor is a 2008 meta-analysis by R. Matthew Montoya, Robert Horton, and Jeffrey Kirchner in the Journal of Social and Personal Relationships, built to pull apart two things that casual conversation treats as one. Actual similarity: how alike two people measurably are — same attitudes, same traits, on paper. Perceived similarity: how alike one of them believes they are, whatever the paper says. The analysis aggregated 460 effect sizes from 313 laboratory and field investigations — decades of experiments where researchers manipulated or measured each kind of similarity and watched what it did to attraction.

Both mattered, overall — this is not a debunking of similarity. The surprise was in the pattern. Actual similarity predicted attraction most strongly among strangers who had never interacted — the classic lab setup where you read a profile of someone's attitudes and rate how much you'd like them. Add real interaction, and its effect shrank. In existing relationships, actual similarity was no longer a significant predictor of attraction at all. Perceived similarity showed the opposite constitution: it predicted attraction at every level — no interaction, short interaction, and inside established relationships. The measurable match fades as a force the moment two real people start talking. The felt match never stops working.

How This Fits Last Week Without Breaking It

Read quickly, the two articles look like a contradiction: last week said couples are alike; this week says being alike stops predicting attraction once you know each other. They are not in conflict, and the distinction between them is the most useful idea on this page. Last week's research measured the census; this week's measured the mechanism. Horwitz and colleagues counted who durably pairs — and yes, the people who end up together really are similar, for reasons that piece was careful to leave open: similar people choose each other, but they also meet each other (sorted schools, workplaces, circles) and become each other (decades of shared meals and elections). Montoya and colleagues asked a different question: at the moment attraction happens, which similarity is doing the pulling? And the answer is the one in your head. The census and the mechanism are two photographs of the same river — one of where the water ends up, one of what makes it move. Real similarity shapes the pool you draw from and the life you converge into. Felt similarity is what falls in love.

Why the Feeling Outruns the Fact

Sit with the strangeness of it: two people can be measurably alike and feel like strangers, or measurably different and feel like twins — and it's the feeling that moves. The likely reasons are ordinary once named. Perceived similarity is really a bundle of experiences: being understood, having your jokes land, the conversation that doesn't require translation. Those experiences are attraction's raw material whether or not a questionnaire would certify the match. And perception is partly a product of liking as well as a cause — we meet someone, something clicks, and we start noticing everything shared and forgiving everything not. The arrow runs both ways, which is exactly why the feeling can hold strong in long relationships after the paper match has stopped mattering: at that point, the sense of being fundamentally alike is the bond, continuously renewed by how two people read each other.

The Dishonest Conclusion, Declined

There is a cynical takeaway available here, and we want to name it in order to refuse it: if attraction runs on perceived similarity, manufacture some. Mirror their interests, echo their opinions, agree with the take you don't hold. Set ethics aside for a moment and it fails on pure engineering: perceived similarity built on performance has a shelf life measured in dates, because the perception eventually meets the person. We've already written about what happens when a match falls for lines that weren't yours — the borrowed-brilliance problem — and fabricated common ground is the same loan at the same interest rate. The feeling of alikeness is powerful precisely when it's tracking something real; counterfeit it and you've built attraction to someone who doesn't exist, which is a strange prize to win.

The honest takeaway points somewhere better: real common ground often goes unperceived, and that's fixable. If attraction needs similarity to be felt, then the practical work of a profile, a first conversation, a good question is making the genuine overlaps visible — legible — instead of leaving them to luck. This is what prompts are actually for: not performance, but surface area — giving your real particulars enough daylight that the right person can recognize themselves in them. And it's why values beat interests as the thing to make legible: a shared taste in music produces a pleasant hour of felt similarity; a shared sense of what a life is for produces the durable kind.

Who This Describes — and Who It Might Not

The standing caveat, in its own section as always. This meta-analysis aggregates predominantly laboratory studies and Western field samples, and most of what it measures is initial attraction and liking — how drawn people feel — rather than long-run relationship outcomes. “Perceived similarity predicts attraction in existing relationships” is a claim about the feeling of attraction within them, not a demonstrated recipe for lasting satisfaction. Strong evidence for the mechanism where it was measured; appropriate humility about cultures, contexts, and decades it wasn't.

The Bottom Line

Two weeks, two findings, one picture. The people who end up together are alike — that's the census, and it's robust. But the force that pulls two particular people toward each other runs on recognition, not resemblance: the sense of being alike, which real interaction can build even where checklists disagree and which checklists can't supply where the feeling is absent. So the practical posture isn't hunting your statistical twin, and it isn't performing one. It's making yourself legible enough — in your profile, your prompts, your first conversations — that when someone genuinely overlaps with you, they can feel it. The click, it turns out, is the mechanism. The trick is making sure it's clicking on something true.