Why is small talk so hard when you're an introvert?

Small talk is hard partly because you are running a prediction about it that turns out to be wrong. In a set of experiments on commuters, people asked to talk with a stranger reported a more positive journey than people asked to sit alone — while a separate group of commuters, asked to predict the same situation, expected the opposite.

That is not a claim that you are secretly an extrovert, and it is not a suggestion that anything about you needs correcting. It is a narrower and more interesting point: the forecast people make about a specific situation appears to be systematically off, and being the sort of person who avoids that situation does not seem to protect you from the error. If anything, it is where the error lives.

The study, and what it actually did

Nicholas Epley and Juliana Schroeder ran a series of experiments published in the Journal of Experimental Psychology: General in 2014, under the title Mistakenly Seeking Solitude. The design fits in a sentence: approach people about to commute, assign them to talk to a stranger, sit alone, or behave normally, and ask afterwards how it went — then separately ask a different group of commuters to predict how each of those would feel.

On trains (97 people completed the experience version), those told to connect with a stranger reported a more positive commute than those told to sit alone, with an effect size of d = 0.63. The conversations ran 14.2 minutes on average. On buses (75 people), the same pattern appeared at d = 0.56, with conversations averaging 9.8 minutes.

The prediction groups went the other way. Train commuters asked to forecast (66 people) expected solitude to be the more pleasant option, d = 0.41. Bus riders (61 people) predicted the same, d = 0.37.

Both sets of bars in the chart above are real effects. They simply point in opposite directions — which is the whole finding.

The part that is specifically about you

Here is the experiment that matters most if you would describe yourself as an introvert.

In a taxi study, Epley and Schroeder split participants by their existing habits: people who routinely talked to drivers, and people who rarely did. The habitual talkers predicted that connecting would be pleasant. The people who rarely talked predicted the opposite — that solitude would be better — with a large effect, d = 0.93.

And then the actual experience went the same way for both groups. Connection was rated more positively regardless of which camp someone came from, also at d = 0.93.

So the difference between the groups was not in what they experienced but in what they expected. That is a useful distinction: the reluctance is not evidence that the thing would go badly for you specifically. The reluctance is the part the research found to be miscalibrated.

Why the forecast comes out wrong

The same paper offers a mechanism, and it is oddly reassuring.

When 64 train commuters were asked about interest in conversation, they reported being significantly more interested in talking to others than they believed others were in talking to them (d = 0.44). Only 46.4% predicted that a stranger would be willing to talk to them at all.

Read that as a room rather than a train. Fewer than half the people present think anyone else wants to talk to them — while being, by their own report, more interested in talking than they assume everyone else is. Everybody is running the same arithmetic and reaching the same conclusion: that they are the only one who would like this.

The forecast is not stupid. It is made from one seat, using the only information available from that seat — your own discomfort, and the neutral expression of the person opposite.

The other finding people skip

A fifth experiment, run in a laboratory waiting room with 179 participants, quietly removes the most common excuse. Being talked to was, as the paper puts it, every bit as pleasant as talking to someone, with no meaningful difference based on who started it.

Worth sitting with if your reluctance is specifically about initiating — the sense that starting a conversation imposes on someone who would rather be left alone. In this study, the person on the receiving end had a good time too.

What this does and does not show

Being straight about the limits, because a study that gets quoted this often tends to grow in the retelling.

• These were commutes and short waits, not parties or networking rooms: a stranger, a bounded amount of time, no follow-up.

• The conversations ran 10 to 15 minutes. This is not evidence about a two-hour reception.

• They were run in a particular place and time, on people willing to be recruited for a study about commuting.

• Effect sizes around d = 0.4 to 0.6 describe a real difference across a group. They promise nothing about any individual conversation, including your next one.

• Introversion here is a self-described preference and a habit. Nothing in this research, or this article, describes a condition, a diagnosis, or a thing to be fixed.

What it supports is narrower and still worth having: for short exchanges with strangers, anticipation is a poor guide to experience, and that gap holds for people who avoid these exchanges as much as for people who seek them out.

Five things you can actually do with this

Treat the dread as a forecast, not a verdict. The feeling before is real. The research suggests it is a poor predictor of the feeling during. You can notice one without obeying it.

Use the bounded version. The situations in the study had a built-in ending — a stop, a destination, a called name. Queues, lifts, the five minutes before a meeting. A conversation with a visible exit is easier to begin.

Let someone else start. Being talked to was rated as positively as talking. If initiating is the expensive part, being reachable is a legitimate strategy: phone away, face outward, look up. That is participation.

Assume the interest is mutual. Under half the commuters thought a stranger would be willing to talk to them, while reporting more interest than they credited others with. The person you are not speaking to is, statistically, running your reasoning.

Stop counting it as a skill you lack. Nothing here measures ability. It measures prediction. Those are different problems, and only one of them is yours to carry around.

September, specifically

There is a reason this gets asked more now than in July. September puts most people in front of a fresh set of strangers at once — new terms, new colleagues, rooms where nobody has settled yet.

Which means the awkwardness is currently distributed. In a room where everyone is new, the assumption that you are the only person finding it effortful is at its least accurate all year. In November it will be a much better assumption. Right now it is nearly always wrong.

Where does your reluctance actually sit?

Six questions about which part of the exchange is expensive for you — starting it, sustaining it, ending it, recovering afterwards — and you get back one of four patterns with three things to try. Two minutes, nothing collected.

Frequently asked questions

Does small talk get easier with practice? The research here does not measure practice, so this article cannot claim it does. What the taxi experiment shows is that people who rarely talked to strangers enjoyed the conversation about as much as habitual talkers once it happened — the groups differed in prediction, not experience. That is narrower than "practice works", but it does suggest avoidance is not maintained by bad outcomes, since the outcomes here were not bad.

Is being bad at small talk a sign of social anxiety? Disliking small talk and having social anxiety are not the same thing, and Mind Decoded describes patterns rather than diagnosing anyone. Plenty of people find small talk tiring, prefer depth to breadth, or would simply rather be reading — none of which is a clinical matter. If avoidance is persistent, distressing, and limiting things you would otherwise want to do, that is worth discussing with a GP or a qualified professional rather than settling from an article.

Why do I feel drained after socialising even when it went well? Enjoying something and finding it costly are compatible, and the research above speaks only to the first. Epley and Schroeder measured how positive the commute felt, not what it cost to produce. A conversation can be pleasant and still leave you wanting silence for an hour. Treating those as two separate facts, rather than assuming the tiredness proves you did not enjoy it, makes the whole thing easier to plan around.

What do I actually say to start a conversation? The commuter studies did not script the openers, so there is no evidence-based line to offer, and any article that gives you one is making it up. What the research does suggest is that the barrier sits in the prediction rather than the content — fewer than half the commuters thought a stranger would be willing to talk at all. The opener probably matters less than the assumption behind it.

Are introverts worse at conversation than extroverts? Nothing in this research measures conversational ability. The taxi experiment found that people who rarely talked to strangers had an experience comparable to those who did it habitually — the difference was in what each group expected beforehand. Introversion, as most people use the word, describes a preference about spending energy. That is a different thing from a skill, and the two get conflated constantly.

Does this mean I should force myself to socialise more? No, and the research does not support that reading. It compares short exchanges with strangers against sitting alone in the same situation; it says nothing about how much socialising anyone should do, and measures nothing about the cost afterwards. A reasonable use is to discount the anticipatory dread when you already wanted to speak to someone. An unreasonable one is to treat it as an instruction to override your own preferences.

Connect

Questions or thoughts? Reach out anytime.

Email

subscribe to our newsletter

hello@minddecoded.com

© 2026. All rights reserved.

Terms of Service Privacy Policy Disclaimer