You Should Talk to Strangers. Science Says So.

We are crap at predicting how it feels to talk to strangers

We nearly all typically predict that talking to strangers will be awkward and unrewarding.

The evidence says we are wrong every time.

In a field experiment with London commuters, those assigned to speak with a stranger reported significantly more positive experiences than those who sat in silence, yet participants predicted in advance they would prefer solitude.

This “misconnection” bias operates because we incorrectly assume others are uninterested in speaking to us and we predict that silence will be better than conversation. We are wrong on both counts.

The missed opportunity extends far beyond the commute: the coffee queue, the lift ride, and the waiting room are all micro-encounters with positive potential that we habitually, and incorrectly, opt out of. The practical implication is not just about pleasant journeys but about weak ties, belonging, and the basic human experience of feeling that the world is warmer than we feared.

On a Tuesday morning in 2014, a group of commuters boarded trains in Chicago. Some were told to strike up a conversation with the person next to them. Others were told to sit quietly and keep to themselves. A third group was simply left to do whatever they would normally do. Afterwards, all of them reported on how they had found the journey. The results were unambiguous. Those who had spoken with a stranger reported a significantly more positive experience than those who had sat in silence. They rated the journey as more pleasant, more enjoyable, and more interesting. And yet, when a separate group of commuters was asked in advance which condition they thought they would prefer, an overwhelming majority chose silence.

This is the central finding of Nick Epley and Juliana Schroeder’s remarkable 2014 paper, “Mistakenly Seeking Solitude,” published in the Journal of Experimental Psychology.

General, and it captures something important about how we relate to one another and why, despite everything, we so often choose not speak to each other. The experiment was replicated with bus passengers and repeated in London with comparable results. Across different cities, different transit systems, and different cultures, the same pattern held: people predicted they would prefer solitude, chose solitude when left to their own devices, and yet reported significantly better experiences when they were obliged to connect instead.

We are, it turns out, systematically wrong about what makes a good journey, and the same error is likely operating across a much wider range of daily social encounters than the morning commute.

The Commuter Experiment

The elegance of Epley and Schroeder’s methodology is worth dwelling on. They were not asking people about whether they valued connection in the abstract, a question that would invite socially desirable responses. They were measuring the actual experience of specific social interactions in specific, mundane contexts, and comparing those experiences to people’s advance predictions. The gap between prediction and reality was consistent and substantial.

What makes the finding particularly interesting is not merely that conversation was more pleasant than silence. It is that people actively expected silence to be better, and those expectations shaped their behaviour. When left to choose, commuters reliably sat alone, looked at their phones, and avoided eye contact with the person sitting eighteen inches away from them. They did this not because they were unfriendly or misanthropic but because they had made a rational prediction based on their own mental model of social interaction, and that model was wrong. (Check our our posts on predictive processing for more of our thoughts on this).

The London replication is worth noting specifically because British culture is often held up as particularly resistant to conversation with strangers. The stereotype of the British commuter, silent, newspaper raised, radiating a force-field of don’t-talk-to-me, is internationally recognised. And yet the pattern held. Even among a population with strong cultural norms against unsolicited interaction, the actual experience of conversation with a stranger was more positive than the experience of sitting in carefully maintained isolation.

Why We Expect Rejection

The obvious question is why we get this so consistently wrong. If talking to strangers is reliably more pleasant than sitting in silence, why do we so reliably predict the opposite?

Epley’s broader programme of research, summarised in his 2013 book Mindwise, offers a compelling account of why our social predictions are systematically mis-calibrated in a particular direction: we consistently underestimate others’ interest, warmth, and similarity to ourselves. (Check out Dan Gilbert’s work on “mis-wanting” and Laurie Santos’ work too.)

The problem is partly that we are trapped inside our own perspective (or our predictive processing model as we think of it). When we imagine initiating a conversation with a stranger, we feel the awkwardness of the approach, the uncertainty about how it will be received, the risk of rebuff (we predict these things). We simulate our own anxiety with considerable accuracy. What we fail to simulate adequately is the other person’s experience: their own desire for connection, their likely warmth towards us, the pleasure they are also likely to take from a good conversation. We project our imagined embarrassment without adequately accounting for the reciprocal human interest that would, in reality, make the encounter work.

There is also a status asymmetry in how we model social risk. The fear of rejection or awkwardness looms large in imagination; the cost of missed connection is invisible. We do not feel the pleasure of the conversation we did not have. We do not experience the warmth of the stranger we avoided. The asymmetry in how we experience potential gains and losses drives us towards the apparently safe option of silence, even when silence is not actually the better choice. As Epley’s work demonstrates, we are not suffering from a lack of confidence in the abstract; we are making a specific, systematically incorrect prediction about a specific type of interaction.

The Evidence We Get Wrong

The 2021 follow-up paper, “Hello, Stranger?”, extended the original findings and examined the mechanisms of the error more closely. Participants reported concerns before initiating conversations with strangers: worry about being rejected, uncertainty about what to say, anxiety about the interaction becoming awkward or running dry.

These concerns were genuine and felt. But once the conversations actually began, the concerns proved largely unfounded. People found things to talk about. The interaction was warmer than anticipated. The other person was more interested and more engaging than expected. The concerns were not irrational given the model of human interaction the participants were operating from; the model itself was simply inaccurate.

This is consistent with Epley’s broader findings in Mindwise. We are, he argues, remarkably poor at reading others’ inner states (and as some of our other posts show, our own ones), and our errors are not random. They are directionally biased. We assume others are less interested in us, less warm towards us, and less similar to us than they actually are.

These errors are not benign mis-calibrations; they have real consequences for how often we reach out, how often we initiate connection, and how isolated or connected we feel as a result. The stranger on the train who looks absorbed in their phone may be just as open to conversation as we are, and just as unlikely to start one for exactly the same reasons.

Beyond the Commute

Epley and Schroeder were careful to frame their findings in terms of commuting, but the implications extend to every context in which we routinely encounter strangers and routinely choose not to engage with them. The coffee queue. The waiting room. The lift ride. The queue at the post office. The seat next to someone at a conference lunch. These are all micro-encounters that most of us navigate in a state of deliberate non-engagement, and if the commuter research is generalisable, we are probably wrong to do so in most of them.

The implications are not trivial. If you interact briefly with a handful of strangers on any given day, and those interactions are consistently more positive than anticipated, the cumulative effect on mood, on sense of connection, and on the basic feeling that the world is a warm and interesting place rather than a cold and indifferent one, is substantial. Research on what psychologists call “micro-interactions” suggests that these brief, low-stakes encounters contribute meaningfully to daily wellbeing, and that their absence is one of the under-appreciated costs of the kind of social withdrawal that has become more common in urban life.

There is something almost paradoxical about the fact that the very density of urban environments, which places us in constant physical proximity to large numbers of other people, seems to intensify rather than reduce our social isolation. We have more strangers around us than any generation in human history, and we make less contact with them. The norm of urban anonymity, the collective agreement that we will be physically present with one another while maintaining the fiction that we are alone, is so deeply embedded that breaking it feels transgressive. But the research suggests that transgressing it is reliably pleasant.

What This Means for Weak Ties

The connection between Epley and Schroeder’s work and the sociological literature on weak ties is worth making explicit. Mark Granovetter’s classic research demonstrated that the acquaintances and loose contacts at the periphery of our social networks, the people we know slightly, from different contexts, with different knowledge and different networks of their own, are often more valuable for spreading information and opportunity than our close friends. But weak ties begin somewhere. The acquaintance who turns out to be a useful professional contact was once a stranger.

If we are systematically avoiding the conversations that would allow weak ties to form, we are not merely missing a pleasant journey. We are impoverishing the social fabric of our own lives in ways that have real consequences for opportunity, information, and belonging. The misconnection bias that Epley identifies is not just a source of unnecessarily dull commutes. It is a structural impediment to the formation of the weak-tie networks that give social life its texture and its range.

Practising the Uncomfortable

Epley has been asked in interviews whether his findings mean that everyone should be trying harder to talk to strangers, and his answer is characteristically nuanced. The research does not suggest that people are miserable and that talking to strangers would fix them. It suggests that people are making systematically inaccurate predictions about a type of social interaction, and that correcting those predictions might lead them to make different choices. Whether those different choices would improve their lives depends on context, personality, and a range of other factors.

That said, there is a reasonable case for treating our resistance to conversation with strangers as something worth examining rather than simply accepting. The specific fear that underlies the avoidance, that the other person will not want to talk, that the conversation will be awkward, that we will be rejected or ignored, is precisely the fear that the research shows to be systematically overstated. We are not being appropriately cautious about a genuinely risky social behaviour. We are being falsely cautious about a behaviour that is, on the evidence, reliably rewarding.

The practical implication is not a prescription to corner your fellow commuters. It is something more modest: a recognition that the predicted discomfort of reaching out is not an accurate guide to the likely experience of the interaction itself, and that the norms of urban anonymity we have collectively constructed are not serving us as well as we assume. The stranger next to you probably wants to talk about as much as you do, is probably as worried about being awkward as you are, and would probably enjoy the conversation about as much as you would. Someone has to go first.

Learning More

To go deeper, explore these resources on people-shift.com:

• Wellbeing
• Subjective Wellbeing
• Employee Engagement
• Bowling Alone
• Dunbar’s Number

People Shift View: The misconnection bias Epley describes is one we ourselves feel a lot (all those tube trips). We like what Nick Epley says and have enjoyed listening to him talk about this topic. We also think connection really matters and that we should all get better at it.