Mistakenly seeking solitude
Nine experiments, more than 800 commuters, one quiet mistake.
Watch a morning carriage fill up and you will see a wordless agreement take hold: headphones in, eyes down, nobody speaks. In 2014, behavioral scientists Nicholas Epley and Juliana Schroeder stood on the platforms of Chicago’s Metra line and asked what that silence was costing.1
They handed commuters one of three mornings, drawn at random. Connect: find the stranger near you and start a conversation. Solitude: keep to yourself and enjoy the quiet. Control: commute exactly as you normally would. Surveys came back by mail once the ride was over.
Before you see what came back, place your bet.
Your prediction · file before viewing results
Compared with sitting quietly, talking to a stranger would make your ride feel…
Results · positivity of the commute, in effect sizes
Commuters asked to forecast these three mornings got the direction wrong, not just the size: they expected solitude to beat conversation (d = 0.41). The people who actually talked had the most positive commutes of all three groups, beating solitude by d = 0.63, and reported getting just as much done.1
Human beings are social animals. Those who misunderstand the consequences of social interactions may not, in at least some contexts, be social enough for their own well-being.
Epley & Schroeder · J. Exp. Psychology: General, 2014
“But nobody wants
to talk to me”
The silence isn’t a preference. It’s a misread.
Why do we get it so wrong? Epley and Schroeder found the engine of the mistake. Commuters rated themselves as significantly more interested in talking than they believed the strangers around them were1 — which, on average, cannot be true of everyone in the carriage at once. Every rider is quietly assuming the others would rather be left alone.
Your estimate · willingness to talk
Picture 100 strangers on a morning train. Someone tries to start a conversation with each of them. How many go along with it?
0 of 100talked
Commuters in the studies guessed that fewer than 47 in 100 strangers would be willing to talk.1 The reality: every participant who managed to try ended up in a conversation. Across the entire series, not a single person reported being rebuffed. The only failures were seven riders with an empty seat beside them.
Commuters appeared to think that talking to a stranger posed a meaningful risk of social rejection. As far as we can tell, it posed no risk at all.
Epley & Schroeder, 2014
The liking gap
You are a worse judge of your own charm than of anyone else’s.
So the conversation happens. It goes fine — better than fine, usually. And then the replay begins. Did I talk too much? That joke. Why did I tell that story?
At Yale, Erica Boothby and colleagues measured the replay. After first conversations between strangers, each person answered two questions: how much do you like your partner — and how much do you think your partner likes you?2
After a first conversation · liking on a 1–7 scale
the liking gap · people are liked more than they knowThe gap survived short conversations and long ones, labs and workshops. Strikingly, outside observers watching the recordings could read the partners’ liking just fine — the signals were right there. The people inside the conversation were too busy auditing themselves to notice.2 The shyer the person, the wider the gap.
And it is stubborn. Among first-year students sharing a dorm, the gap persisted from September through February. It closed only by May.
- SEPgap
- OCTgap
- DECgap
- FEBgap
- MAYclosed
Yale dorm mates, measured five times across an academic year: at every point except the last, people underestimated how much their suite mates liked them.2
After people have conversations, they are liked more than they know.
Boothby, Cooney, Sandstrom & Clark · Psychological Science, 2018
Overly shallow
Fine, you’ll talk. But you’ll keep it safe: weather, haircuts, television.
In 2021, Michael Kardas, Amit Kumar and Nicholas Epley tested the keep-it-light instinct in twelve experiments with more than 1,800 people, pairing strangers off for either shallow or deep conversations.3 These are real question cards from the study. A stranger sits down across from you — pick the conversation you’d rather have.
- How did you celebrate last Halloween?
- How often do you get your hair cut? Ever had a really bad haircut?
- What do you think about the weather today?
- For what in your life do you feel most grateful?
- If a crystal ball could tell you the truth about your life, what would you want to know?
- Can you describe a time you cried in front of another person?
Most people flinch at Conversation B — with a stranger, anyway. And on every measure, that flinch was wrong. The gaps between what people expected deep conversation to feel like and what it actually felt like are among the largest you will find in social psychology:
Deep conversations with strangers · expectation vs. experience, in effect sizes
Effect sizes (Cohen’s d) from Experiment 1a. Psychologists conventionally call d = 0.8 large; these run past 1.7.3
The mechanism was simple: people assumed a stranger wouldn’t care about their answers. Strangers cared. Deep talk also beat small talk at its own game — partners felt more connected after the deep questions (d = 0.48) — and walking out, participants said they wished they could have conversations like this more often (d = 1.35).3 When researchers simply told people these results, they chose the deeper questions on their own.
“Surely not here”
The obvious objection: Chicago is not the world.
The London commuter is legendary for the sanctity of silence, so that is where the theory went to be broken. In 2022, Schroeder, Lyons and Epley re-ran the experiments on trains and buses in and out of London, with larger samples and a stricter design.4
Same result. Riders assigned to talk reported more positive journeys — and learned more — than those left in solitude, and their mispredictions traced to a single belief: that the other person wouldn’t be interested. The other person was interested.
The barista
People told to have a genuine interaction with their coffee-shop barista — smile, eye contact, a brief real exchange — left in a better mood and with a stronger sense of belonging than people told to be efficient.5
Weak ties
Students clicked a counter for every interaction. On days they talked with more acquaintances than usual — the classmate, the neighbor, the familiar face — they were happier and felt they belonged more.6
Say it out loud
One gap left: when we finally reach out, we type.
Amit Kumar and Nicholas Epley asked people to reconnect with an old friend, then randomized the medium: email or a phone call.7 Left to choose, most leaned toward typing — a call, they predicted, would be awkward.
The people who actually heard a voice felt significantly more connected than the ones who typed, and reported no more awkwardness. The call didn’t even cost more time: about as long as reading and writing the email. The same pattern held when strangers met through voice rather than text — the media that carried a voice built the stronger bond.
Every gap in this story points the same way. The stranger is more willing than you think,1 likes you more than you think,2 will go deeper than you think,3 in more places than you think,4 and a voice carries more warmth than you think.7 The research has a shape, and the shape looks a lot like a phone call with someone you have never met.