Charisma Is a Learnable Skill: the 7 things that make people want to keep talking to you

Seven behaviours, ordered from the least settled research to the most replicated, and every one of them can be learned.

There is a man at every party who has taken up a position beside the food and has no intention of leaving it.

I know him intimately, because for about fifteen years I was him. It is not a bad post. You get first refusal on the good crisps, there is a wall behind you, which is more consoling than anyone likes to admit, and you have an unobstructed view of the phenomenon that this article is about: the one person in the room who appears to be generating their own weather.

You know the one. They arrive slightly late. Within four minutes there is a small crowd standing where a bookcase used to be. They are not the best looking person present and they are frequently not the funniest, and if you eavesdrop, which I have done extensively and without shame, you notice something quietly deflating. They are not saying anything clever. They are asking a man in a terrible jumper about his dog, and the man in the terrible jumper is having the best evening he has had in a month.

For fifteen years I assumed this was a factory setting. Some people came with it, like double-jointed thumbs, and the rest of us stood near the crisps and went home and conducted the audit.

You know the audit. It begins in the car, or on the walk to the station, and it concerns the thing you said about the schools. It is thorough. It is unsparing. It is conducted by a prosecutor who was also, inconveniently, the defendant, and it has never once returned a verdict of acquittal.

I want to tell you two things about that audit before we go any further, because they change the whole shape of what follows.

The verdict was in before you spoke, and it was kinder than you think

In 2006, Janine Willis and Alexander Todorov at Princeton flashed photographs of strangers’ faces at people for a tenth of a second, then asked for judgments on trustworthiness, competence, likeability and a handful of other traits. The judgments made after a hundred milliseconds correlated highly with the judgments people reached when given as long as they wanted. Extra time did not overturn the verdict. It mostly increased how confident people were in it.

Which is a genuinely unpleasant result if you stop there, and everybody stops there. It has been used to sell an enormous quantity of nonsense about handshakes.

So let us not stop there, because sitting immediately next to it is one of the loveliest findings in social psychology, and almost nobody has heard of it.

Erica Boothby, Gus Cooney, Gillian Sandstrom and Margaret Clark ran a series of studies on people who had just finished talking to each other. Strangers paired in a laboratory. First-year students landed with a roommate they had not chosen. Adults at a personal development workshop, which I mention only because it must have been a difficult afternoon. In every setting, people underestimated how much their partner had liked them and enjoyed their company. The researchers called it the liking gap. Part of the cause, they found, is that we are savage critics of our own performance in a conversation, in a way that the other person, who was busy having the conversation rather than grading it, simply is not.

Put those two findings side by side and something rather wonderful falls out of the gap between them.

The room decided about you almost instantly. And the room liked you more than you have ever believed it did.

The audit in the car has been marking its own homework, harshly, in a subject where it was never the examiner.

That is your actual starting position. Not zero. Somewhere already comfortably above zero, accompanied by a lifelong private conviction that it is below.

Why a health publication is spending eight thousand words on this

Because the stakes here are not social. They are cardiac.

Julianne Holt-Lunstad and colleagues pooled seventy studies on isolation and mortality. Social isolation carried a 29 per cent increased likelihood of death, loneliness 26 per cent, living alone 32 per cent, figures the authors placed alongside well established risk factors. The risk from a lack of social relationships came out larger than the risk from obesity.

Nobody in the history of the world has been talked out of loneliness by a percentage, and I am not going to try. I mention it so you know why a publication that usually writes about visceral fat and step counts has spent three weeks on the question of how long to look at someone. The capacity to make a stranger glad they spoke to you is not a soft skill. It sits upstream of your friendships, which sit upstream of your last twenty years. It simply happens to be the one skill nobody ever sat you down and taught, on the grounds that everyone assumed you either had it or you did not.

Seven behaviours follow. They run from the least settled evidence to the most replicated, which means the ground gets firmer under your feet as you go, and the last one rests on a randomised trial.

They are not rivals and they do not work alone. Each one can quietly cancel another, and I will show you exactly how as we go. Take up the whole sofa, radiate ease, ask nothing about the person opposite, and you have not become magnetic, you have become a peacock. Ask flawless questions while folded into yourself like a deckchair in a shed and you have conducted an interview. The person generating their own weather is running five or six of these simultaneously without a single conscious thought, which is precisely why watching them has never taught you anything.

Some of it works in the first ten seconds. Some of it is a craft with a syllabus and takes weeks. I will tell you which is which, which is more than the people selling instant magnetism will do.

---

1. A nod is a receipt, and most of us write terrible ones

Watch yourself in a conversation this week, at the moment somebody else is doing the talking.

Your head is going. Up, down, up, down, a steady metronomic agreement, and somewhere around the fourth beat you will catch yourself doing something else at the same time: assembling. You are nodding at a story about a difficult sister-in-law while quietly drafting your own contribution about your difficult sister-in-law. The nod has become entirely decorative. It has detached from your attention and is running on its own small motor, like an indicator left on after the turn.

Everybody on the receiving end of that nod can feel it. Not consciously. But they can feel it in the same way you can feel when the person opposite is reading their phone under the table, and they will conclude, without ever forming the thought, that this conversation is not going anywhere and they should probably go and find someone else.

Which raises the question of what a nod is actually for, and the answer is smaller and better than agreement.

Takayuki Osugi and Jun-ichiro Kawahara, working at Yamagata and Hokkaido Universities, made short clips of computer-generated figures doing one of three things: nodding, shaking their heads, or standing there like a rendered coat stand. They showed the clips to forty-nine people and asked for ratings on likeability, approachability and physical attractiveness.

Nodding raised likeability by around 30 per cent and approachability by around 40 per cent. Head shaking did nothing whatever to either. And the effect on judged physical attractiveness came in below 10 per cent, which tells you something precise: the nod was not doing anything to the face.

Nodding primarily increased likability attributable to personality traits, rather than to physical appearance.

Jun-ichiro Kawahara, Hokkaido University

So a nod is not agreement and it is not charm. It is a receipt.

It is the small physical acknowledgement that a sentence arrived intact, and its function is to let the speaker continue without stopping to check whether they have lost you. It costs nothing to issue and it is enormously reassuring to receive, which is why it is the first thing to break down when you stop listening.

Two things follow, and they are more specific than you would expect from something this small.

The first is timing. Almost all of us nod during other people’s sentences, which is when the nod is least useful and most obviously automatic. The useful nod comes at the end of a thought, once something has actually landed, and it is slower than feels comfortable. One nod, held a fraction longer than you want to hold it.

The second is the triple nod, and I would like you to go looking for this in yourself because it is astonishing how often it appears once you know it is there. The rapid yes-yes-yes, three quick dips, often accompanied by a small intake of breath. It does not mean go on. It means wrap it up, I have somewhere to be with my sentence, and it is one of the most efficient ways in the English-speaking world to make someone decide to stop talking to you for the rest of the evening.

Now, the evidence. Forty-nine participants, all in Japan, judging animated figures rather than actual humans in an actual room. It is the thinnest ground in this article, which is why it comes first, and it survives here only because everything after it keeps confirming the same thing from firmer ground.

Nodding helps.

But you can nod perfectly, on the beat, for forty minutes, while thinking about parking.

2. Three point three seconds

There is a point at which eye contact stops feeling warm and starts feeling like a procedure. Everyone has crossed it. Almost nobody knows where it is.

You know the sensation from both sides. There is the person who will not look at you, whose gaze keeps sliding off to check the door, and who leaves you with the impression that you are a delay rather than a conversation. And there is the other one, the one who has read something, who locks on and stays locked on, and whose face you find yourself studying for signs of a plan. Somewhere between those two there is a correct amount of time. You have been estimating it by feel your entire life.

In 2016, a team led by Nicola Binetti at University College London decided to stop estimating. They set up at the Science Museum in London and recruited visitors off the floor, roughly five hundred of them, from fifty-six countries, aged eleven to seventy-nine. Each one sat in front of a screen showing an actor looking directly at them, and pressed a button at the moment the eye contact stopped feeling comfortable.

There is something magnificent about this. Five hundred people, on a day out, queueing to be stared at by a stranger on a monitor so that science could establish precisely when it gets weird.

The number they got is 3.3 seconds. Give or take 0.7.

The clips ran from a tenth of a second to just over ten, and the preferences clustered tightly in the middle of that range. But the finding that actually changes your behaviour is the one underneath: preferred gaze duration did not depend on gender, and it did not depend on personality.

Sit with that for a moment, because it removes a job you have been doing for years. You can stop trying to read whether the person opposite is an eye contact person. There is no such category. They are running approximately the same clock as everybody else in the room, including you.

Which means 3.3 seconds is a budget per look. It is not a total for the conversation.

That distinction is the entire technique, and it is where nearly all the popular advice goes wrong, because the popular advice on eye contact was written by people who believe a conversation is a contest. Some version of hold their gaze until they look away first has certainly reached you by now. That is dominance advice. It is advice for winning, and winning is not the activity taking place at a dinner party. The far end of what Binetti’s team tested was ten full seconds, and the person determined to hold anywhere near it does not read as confident. They read as somebody who has recently finished a book about it. There is also a persistent trick involving mentally tracing a triangle across the other person’s face, which has the practical effect of making you look as though you are inspecting them for a leak.

Eye contact is not an endurance event. It is contact and release, and the release is the part that matters, because it is what makes each return feel like a decision rather than a lock.

So: look, land, hold it a beat past comfortable, glance away as though you are thinking, which you probably are, and come back. That is all of it. It feels absurdly mechanical the first three times, and then it disappears into you and never surfaces again.

The limitation is one the authors state themselves, and it is worth knowing. This was a screen. The actor was not looking back in any meaningful sense, so it is a one-directional measurement of a two-directional event. Treat 3.3 seconds as a well-measured centre of gravity rather than a stopwatch setting.

Eye contact helps.

But you can look directly into someone’s face for the entire duration of a story and not hear a single word of it.

3. The starfish in the lavatory

Go and stand at the edge of a party and count the drinks.

Not the number of them. The height. There are people holding a glass at waist level, loosely, the way you hold something you have forgotten about. And there are people holding a glass pressed flat against the sternum, both hands, elbows pinned to the ribs, and what they have built, without any awareness of having built it, is a small defensive wall out of a gin and tonic.

Once you have seen it you cannot unsee it. It is everywhere. The ankles wound around the chair legs at a dinner party. The hands tucked under the thighs, palms down, which is the single most common thing an anxious person does with their hands and which reads instantly, from across a room, as bracing. The folded arms that the folder will tell you are just comfortable, and which are in fact the most legible sentence their body has produced all evening.

We know this matters. What we were told about why it matters turned out to be wrong.

In 2012 Amy Cuddy gave a talk about power posing that became one of the most watched pieces of psychology ever recorded. Stand like a starfish in a lavatory cubicle for two minutes, the argument ran, and your testosterone rises, your cortisol falls, and you walk into the interview a different animal. A very large number of people have since stood in a very large number of lavatory cubicles.

Then Eva Ranehill and colleagues ran a bigger replication. Power posing did shift how powerful participants said they felt. It did not shift their behaviour and it did not shift their hormones. The claim that made the talk famous, that these poses could meaningfully change the outcomes of a life, was not supported. Joe Simmons and Uri Simonsohn then went through the original evidence in forensic detail and concluded there was not much left once selective reporting was accounted for. Later meta-analyses, pooling first dozens and then more than a hundred studies, do find differences between expansive and contracted displays on self-reported and behavioural measures. On hormones and physiology, they find nothing.

So the cubicle is out. Whatever the starfish is doing in there, it is not chemistry.

And yet something did survive that wreckage, and it deserves to be heard properly, because it is the more useful finding of the two and it has spent a decade being drowned out by the louder one that collapsed.

It survived precisely because it never made any claim about you. Nothing about hormones. Nothing about your inner state or your sense of personal power. It is a claim about the two seconds before you have said anything at all, and about one measurable thing you are doing with your body right now, in the chair you are sitting in, which determines whether the person opposite reads you as someone worth staying with.

It was tested frame by frame on video, on people who had four minutes to decide whether they ever wanted to see each other again. Then it was tested a second time, somewhere considerably harder to fake, on three thousand people who had no idea they were in an experiment at all.

It took me two readings to believe the size of it.