What New Friendship Research Gets Wrong About Why We Bond
A 2026 study finds friends wildly overestimate how alike they are, and that the feeling of similarity, not the real thing, is what actually predicts satisfaction.
Ask someone why a particular friendship works and you'll usually get a version of the same answer: we're just alike. Same sense of humor, same taste, same way of seeing things. It's a tidy story, and it's mostly wrong, or at least wrong about which half of it is doing the work.
A 2026 study of friend groups found that people vastly overestimate how similar they actually are to their close friends, in personality, in values, in the small daily preferences that feel like they must be shared because the friendship feels so easy. The more striking finding wasn't the overestimation itself, it was what actually predicted satisfaction in the friendship: perceived similarity, not measured, real similarity. Two people who are, on paper, fairly different in temperament can feel deeply matched, and that feeling of being matched does most of the work that real overlap was assumed to do.
Friendship Runs on Perception, Not on Matching Traits
This is a genuinely uncomfortable result for the tidy version of the friendship story, because it means the thing people credit for a bond, the sense of being cut from similar cloth, is frequently a story the friendship itself produced rather than a fact that caused the friendship. You don't necessarily become close to someone because you're alike. You come to believe you're alike because you're close, and closeness is generous about filling in the resemblance afterward.
What the research points to instead is a cluster of partner qualities that matter more than trait-matching: whether someone is responsive, whether they show up reliably when it costs them something, whether they make you feel understood in the specific moment you needed to be, rather than in general. Those are qualities in how a person treats you, not facts about how alike your two personalities are on some inventory. A wildly extroverted person and a quietly introverted one can be best friends for thirty years, not despite the difference, but because responsiveness and reliability don't require matching temperaments to deliver. It's entirely possible to be nothing alike and completely each other's person.
Why the Old Exchange Model Undersells What Friendship Does
Psychology has leaned for decades on a transactional frame for close relationships: a friendship persists roughly as long as what each person puts in stays balanced against what they get out, favors traded, support given and returned, an implicit ledger both people keep without quite admitting to keeping it. It's not a crazy model. Plenty of relationships really do run aground on a felt imbalance, one person always giving, the other always taking.
But an exchange frame quietly assumes the point of a friendship is the goods being exchanged, favors, advice, company, and that the relationship is basically a delivery mechanism for those goods, replaceable in principle by anyone else who could deliver the same goods. The perception research suggests something closer to the opposite: a large part of what a friendship is doing is producing the feeling of being known and understood by this particular person, which isn't a good that transfers if you swap in someone else offering the same favors. Two friends can trade nothing of practical use for months and still be doing real relational work, because the exchange was never the actual mechanism. The ledger model explains why some friendships fail. It's much weaker at explaining why the good ones actually work.
What This Changes About "Finding Your People"
The common advice for building a social life leans hard on the matching frame: find people who share your interests, your values, your sense of humor, and the friendship will follow from the overlap. It's not bad advice exactly, shared interests are a genuinely efficient way to end up in the same room as someone repeatedly, which is a real precondition for closeness. But it points people at the wrong variable to optimize once they're already in the room.
Once you're talking to someone, the thing worth paying attention to isn't how much you have in common on paper. It's whether the interaction itself produces the feeling of being tracked, responded to, actually heard on the specific thing you just said rather than redirected back to the other person's story. That's a much harder thing to screen for on a profile or in a first conversation, and a much better predictor of whether the friendship will hold up. It also means someone who shares almost none of your stated interests can turn out to be exactly your person, and someone who checks every box on paper can turn out to produce none of the actual feeling that makes a friendship durable. The overlap gets you in the room. It was never the thing keeping you there.
A Short Reflection: Assumed Versus Confirmed
Pick one close friend. Down one side of a page, or just in your head, list three or four things you're fairly sure you share with them, a value, a taste, a way of seeing something. Then, honestly, mark which of those you've actually confirmed in a real conversation versus which ones you've simply assumed because the friendship feels close enough that shared ground seems obvious.
Most people find at least one item that's never actually been tested, a belief you'd have bet money on that turns out to rest entirely on vibes and closeness rather than anything either of you has said aloud. That's not a problem to fix. It's just useful information about how much of what feels like deep similarity is really the feeling of the friendship reflecting back at you, and how much of your actual bond is coming from somewhere else entirely, from being shown up for, understood, taken seriously, none of which needed the shared trait to exist in the first place.
Frequently Asked Questions
Does this mean shared interests don't matter at all in friendship? They matter for getting people into repeated contact with each other, which is a real precondition for closeness. They just don't appear to be what sustains the bond once it's formed; responsiveness and feeling understood carry more of that weight.
Why do people overestimate how similar they are to their friends? Closeness seems to generate a kind of retroactive story-making: once you feel bonded to someone, it's natural to read more overlap into the relationship than actually exists, because "we're alike" is a simpler explanation for the closeness than "they happen to be responsive to me specifically."
Is the exchange model of friendship completely wrong? Not completely. Chronic, one-sided imbalance really does damage relationships. The exchange model just isn't the main engine of why good friendships work; it's better at explaining the failure cases than the thriving ones.
How does this change advice for someone trying to make new friends? Use shared interests to get into the room, but once there, pay attention to whether the interaction feels responsive and understood rather than to how many things you have in common. That's the signal more worth trusting.
What should I do with the reflection exercise once I've done it? Nothing needs to change about the friendship itself. The value is mostly in noticing how much of what you call similarity is actually just the feeling of being known, which tends to make you a little more generous about friendships with people who are, on paper, nothing like you.