How to Read Attraction: The Psychology of the Signal

There’s a window, near the end of a long phone call that’s gone well, where someone could just say it — we should do this again — and the call would go on a little differently, or it wouldn’t, but the thing would at least have been said out loud. The window is short. Four seconds, maybe five, somewhere after the conversation has clearly started winding down and before either person has actually said goodbye. Almost nobody uses it. The call ends instead the ordinary way, with whatever logistics were already pending and a vague, mutual agreement to talk soon, and the four seconds get folded into everything else that didn’t happen on the call, indistinguishable from a dozen smaller silences nobody remembers passing.

The timing matters more than it logically should. The same sentence, sent ten minutes later by text, says something different — it arrives already edited, already reconsidered, with the live, unrehearsed version of it gone. A signal offered inside the window costs more precisely because it couldn’t have been workshopped first, and that’s most of what lets it count as a signal at all, rather than a decision arrived at later, alone, and then merely reported.

Not because the feeling isn’t there. The feeling is usually right there, available, unmistakable to the person having it. What’s missing isn’t the feeling. It’s a decision about what it would cost to say it out loud and be wrong.

The Master Hub for this category already took up the receiver’s half of this problem — the threshold a receiver brings to whatever arrives, set by what they’re already hoping for rather than by the strength of the signal itself. The sending half is what this series is actually about. A sender is running a threshold of their own, on the way out, before any receiver gets involved at all.

A sender weighing those four seconds of silence is pricing two different ways of being wrong. Say it, and risk being read as more invested than the moment can bear — a false alarm, to borrow the language signal detection theory takes from radar operators deciding whether a blip is a plane or static. Stay quiet, and risk losing something that was never wrong at all — a miss, the real signal that never got sent because sending it felt like the riskier of the two silences. Almost everyone defaults toward the miss. A miss is invisible to everyone, including, eventually, the person who made it. A false alarm is witnessed.

Not every signal in this series works the same way, and the difference matters before any individual case gets examined. At least four distinct problems hide inside the single word “signal,” and confusing them is most of what makes attraction feel harder to read than it should.

The first is interpretation: a real signal has been sent, and the only open question is whether it gets correctly read on arrival, or lost the way most quiet, indirect signals get lost — the kind of thing that gets noticed three days later, replayed, and only then understood for what it was. The second is discrimination: something that looks like a signal — warmth, attention, a remembered detail — turns out, on closer inspection, to be a baseline trait rather than a targeted one, indistinguishable on the surface from the real thing until someone checks whether it shows up for everyone else too. The third is channel: signals travel through more than one medium at once, and the body commits to a read before language is even available to commit anything — a held gaze a half-second too long, a torso angled toward someone before any decision about it has been reached. The fourth is production: the sender’s own side of things, the deliberate choice to send something on purpose, in a vocabulary nobody sat down and taught, assembled out of whatever happened to be lying around — a joke, a deflection, an invented reason to stay five minutes longer than the conversation strictly required.

Reading a posture as discomfort and reading a withheld answer as discomfort are not the same skill, even though both get filed under “signs.” This series treats them as four separate, related problems, not one problem asked four different ways — and most of the territory below sorts cleanly into one of the four. Each carries its own version of the cost-of-being-wrong problem from the opening, just applied to a different part of the transaction: misjudging interpretation costs a connection that was actually there; misjudging discrimination costs mistaking ordinary politeness for a promise; misjudging channel costs trusting a body that hasn’t caught up to its own decision yet; misjudging production costs missing a deliberate move when one finally, actually arrives.

Knowing the sender’s threshold exists doesn’t make it easier to set correctly, for the same reason knowing about a blind spot doesn’t make it visible. What decides whether someone speaks up runs almost entirely on stakes that have nothing to do with the actual signal: how much a previous miss cost, how recently a false alarm went badly, what this particular silence is worth compared to the last one. Two people in the exact same situation, with the exact same feeling, can land on opposite decisions for reasons that have nothing to do with how strong the feeling is — one has a recent false alarm still costing them something, a confession six months ago that didn’t land the way they’d hoped, the other doesn’t, and the strength of what they’re each feeling is nearly beside the point.

This means a missed signal usually isn’t a failure of feeling, and it isn’t a failure of courage either, in the simple sense the word implies. It’s a threshold that was calibrated by something that happened in a different relationship, sometimes years earlier, now quietly setting the price of speaking up in this one. The fix, if there is one, isn’t bravery. It’s noticing that the price being paid was set by a transaction that already closed.

The version of this that produces the most genuine loss isn’t one person miscalibrated and the other reading correctly. It’s two conservative thresholds meeting at the same low-stakes evening: both people interested, both pricing a false alarm higher than a miss, both choosing the miss, both walking away from an evening that produced, on paper, nothing. Two correct readers of an incomplete instrument, each one waiting for a signal the other had already quietly decided not to send.

Most of the social judgment attached to this difference is really judgment about an instrument setting, narrativized after the fact as character. Someone with a low threshold gets called confident, decisive, someone who knows what they want. Someone with a high one gets called cautious at best, oblivious or cold at worst. Both labels describe the same calibration, read backward from its outcome, as if the setting had been a virtue or a flaw all along, rather than a price quietly paid somewhere else, on a different night, by a different version of the same person.

More people will arrive at this series by typing some version of does he like me into a search bar, alone, late enough at night that the question feels more urgent than it would at noon, than will arrive at any other series in this category. What they’re looking for, technically, already exists — has existed for years, on dozens of other sites that answer the four words and stop there.

Every other series here also assumes something this one hasn’t yet established: that a signal got sent, or didn’t, and that someone is now living with the consequence of how it landed. Attachment styles activate around contact that already happened. Mixed signals require an earlier signal clean enough to later contradict. Even people-pleasing’s appeasement and overthinking’s loop need a relationship already underway to attach themselves to. This is the piece of the picture set before any of that has anything to run on yet — the reason it gets searched more than the rest put together, and the reason a four-word answer was never going to be the whole of what’s actually there.

The four problems sorted out above don’t just organize this series internally — they’re the ground floor everything after this one quietly stands on. The Gender Psychology series can ask why men and women tend to encode signals differently only once it’s established, here, that the encoding takes more than one form at all. The Mixed Signals series can describe a wire carrying two messages only because a single, clean signal — the kind this series spends its time on — has already been described first. Nothing later gets to skip the work this one does.

It’s also worth saying what this series isn’t training anyone to do, since that boundary gets misread constantly. Emotional intelligence — what happens after a signal is noticed, how it gets metabolized and acted on without making things worse — is a different series’ territory entirely, and assumes this one’s job is already finished. A signal correctly read and then mishandled is an Emotional Intelligence problem. A signal never correctly read in the first place is this one.

Attachment depends on this series even more directly than the others, just earlier in the sequence. A style can only activate around contact that already happened, which means whatever this series gets wrong about how that initial contact got read — suppression mistaken for disinterest, a baseline trait mistaken for a targeted one — becomes the starting condition the Attachment series has to work with, not a separate problem it gets to correct for later.

Some of the sender’s silence is exactly the kind from the opening — saying nothing instead of saying something that could be checked later. A man three weeks into seeing someone will sit through an entire dinner without once saying the thing he’s apparently sure of, that this is good, that he wants it to continue, and somehow expect the wanting to register as obvious anyway, on the strength of having shown up, paid for dinner, and laughed at the right moments. That particular silence is one of the most reliable patterns in this category, and one of the most reliably misread, for exactly the reason the opening laid out; the rest of this series takes it apart on its own terms, in detail. The same withholding shows up in its more commonly female-coded form too, where indirectness itself ends up doing most of the work a direct statement would otherwise have to do — a question answered with a question, an invitation extended sideways instead of head-on. Why the coding splits this way at all is the Gender Psychology series’ question, not this one’s. Both versions are still answering the same question from the opening, what it costs to be wrong, just dressed in whichever vocabulary the situation handed each person first.

A different problem shows up once someone is simply, evenly nice to everyone in the room. Someone three dates in remembers an order at the coffee shop down to the modification, and there’s no way yet to know whether that’s about the person sitting across from him or just how he treats the barista, the coworker, whoever held the elevator that morning. Warmth distributed identically to twelve people carries a different kind of information than warmth aimed at one, and telling a targeted signal from a baseline trait — on both the more commonly male-coded and female-coded versions of this confusion — is its own piece of work, distinct from the suppression problem above it. This isn’t really a sender problem at all, which is what makes it genuinely different work from the pair above it; the threshold being misjudged here belongs to whoever’s doing the reading, not whoever’s being read.

The body, meanwhile, has usually already decided something language hasn’t caught up to yet. A hand reaches half a second before its owner has consciously decided to let it. A torso angles toward someone across a crowded room before any sentence about why has been assembled. A separate piece in this series takes up that channel specifically: what commits, physically, before language is even available to commit anything, and how much earlier that commitment usually arrives than anyone involved would guess if asked directly. None of this requires anyone to be lying. A body can commit honestly to something its owner hasn’t yet decided to admit to, which is exactly what makes the channel worth treating as its own problem rather than folding it into whichever spoken signal eventually catches up to it. The lag between the two is often where the most reliable information in the entire exchange is sitting, unread, because everyone in the room is still waiting for the slower, verbal version to arrive first.

And some of what belongs here isn’t about reading a threshold at all, but about what happens once someone stops calibrating it and actually crosses it on purpose. A joke gets aimed at one specific person at the table instead of just told to the table generally, and the aiming is most of the actual content of the message — the deliberate, active choice to flirt, in a vocabulary nobody sits down and teaches anyone, assembled out of whatever was already lying around. The male-coded version of that choice gets its own treatment. So does its counterpart: a direct question left to sit, unanswered, for exactly long enough that the not-answering becomes the answer — restraint used as the move itself, not as a failure to make one. Both are choices, not leaks, which is the real dividing line running through this entire series: some signals get away from a person, and some get sent on purpose, and almost nothing about reading either one correctly tells you, in the moment, which kind you’re looking at.

None of these four problems waits politely for the others to finish. A single dinner can run all four at once: he says nothing about wanting to see her again, which is suppression; he’s unusually attentive in a way she can’t yet tell is general or specific, which is discrimination; he leans in slightly without registering that he has, which is channel; and he risks one aimed joke near the end of the night that the first three behaviors never once attempted, which is production. Four separate problems, the same evening, the same person, none of them waiting for the others, and no single one of them telling the whole story on its own.

At a friend’s housewarming, a man notices a woman he’s spoken with twice already standing alone near the bookshelf for the third time in twenty minutes. Instead of walking over, he refills his own drink for the second time, taking the long way past her without stopping.

He’s walked within ten feet of her twice now without stopping. That’s not nothing — someone who genuinely didn’t care wouldn’t bother engineering even that. But it’s also not a signal, not in any form she could repeat back to a friend, or to him, as evidence of anything. He’s found the one move that lets him register interest to himself without producing anything he could be held to.

This is the same threshold from the opening, just walked instead of spoken. The cost of crossing the room and saying something is the cost of a false alarm made physical and witnessed by an entire party. The cost of the slow lap past the bookshelf is close to zero, recoverable, deniable even to himself if anyone asked. Most of what reads as hesitation in this category isn’t really about whether someone feels something. It’s about finding the cheapest available action that still counts, to the person doing it, as having tried.

Not everyone defaults to the cheap option. Some people cross the room anyway, every time, seemingly without running the calculation at all — and the more accurate accounting is that they’re not without the calculation, they’ve simply already paid for a few false alarms early enough in life that the price stopped feeling unfamiliar. A low threshold is cheaper, not braver, for whoever has already absorbed the loss a few times and found it survivable.

Signal detection theory has a name for an instrument tuned to minimize false alarms above almost everything else: a conservative threshold. It trades away a lot of valid detections to avoid the embarrassment of a wrong one, and the trade is rational, person by person, even when a room full of people making the same rational trade is collectively terrible at producing the thing everyone in it is nominally there for.

Whether any of this registers on her end is a separate question, running its own version of the problem from earlier in this piece — her threshold for what counts as enough evidence to act on is exactly as invisible to him as his is to her, shaped by whatever she’s already paid for in some other room, with some other person, on some other night. Two thresholds, set independently, by two different histories, deciding between them whether anything happened at this party at all. If hers happens to be conservative too, this becomes the double miss named above, in its smallest possible form: two people, one bookshelf, neither one wrong about anything, and nothing on paper to show that either of them was ever in the room.

None of this addresses what happens once a signal actually lands and contact follows from it — that territory belongs to the Attachment series. It doesn’t address what happens when a signal that clearly landed once starts contradicting itself later, which is the Mixed Signals series’ job. This series stays in the narrower window before any of that: the four seconds, repeated, in every form they take before anyone involved has decided to call it anything yet.

It’s also, probably, not where the search that brings most people to this series in the first place gets fully answered — and that isn’t a failure of the question, or of whoever typed it at midnight hoping for something more usable than what a dozen other sites already offered. A real answer to does he like me was never going to fit in a single word. It was always going to be two thresholds: his, calibrated by costs that have nothing to do with her; hers, calibrated by costs that have nothing to do with him; and whatever gets called a signal sitting in the gap between two instruments that were never built to match.

There’s a partial answer to whether a threshold like this can actually move, and it isn’t willpower. Thresholds shift the way most calibrated instruments do: with feedback, repeated enough times that the old price stops feeling like the only available price. A handful of false alarms that turned out to be survivable — said, and not catastrophic — tend to do more to lower someone’s threshold than any amount of correctly being told they’re too cautious. Which means the people most stuck at a conservative setting are usually the ones with the least recent evidence that being wrong is survivable, not the ones with the least courage.

Go back to the end of the call. Nothing about those four seconds gets easier once the mechanism behind them is named — knowing the price was set by an old transaction doesn’t refund it. What changes, if it changes anything, is what a missed window means afterward. Not proof the feeling wasn’t real. Not proof the moment wasn’t worth pursuing. Just a threshold, calibrated somewhere else, by someone with their own currency to protect, doing on this particular evening exactly what it was built to do.

What reading someone else’s threshold actually requires, once it’s clear neither side is running the calculation on purpose, is what the rest of this series takes up one case at a time — not as a single skill learned once, but as four different instruments, each demanding its own kind of attention: the silence that has to be told apart from indifference, the warmth that has to be told apart from a baseline, the channel that commits before language does, and the choice, finally made on purpose, to cross the four seconds and say it.