A quiet train carriage can feel like a shared agreement: look at a screen, keep your bag close and let everyone preserve a little private space. In a 2014 Chicago-area field experiment, researchers tested whether that instinct accurately predicts the more enjoyable journey.
It did not, at least for this sample. Commuters randomly assigned to talk with a stranger rated their ride more positively than those assigned to remain alone. A separate group, asked to imagine the same conditions, predicted the opposite.
This is one paper, not a universal rule about every commuter or every conversation. The study used self-reported mood and pleasantness on one morning route, and a welcome exchange between willing passengers is not the same thing as unwanted attention.
What the commuters were actually asked to do
Psychologists Nicholas Epley and Juliana Schroeder recruited 118 adults just before they boarded inbound morning trains at the Homewood, Illinois, Metra station. In the experiment reported in the Journal of Experimental Psychology: General, each commuter was randomly assigned to one of three conditions.
Those in the connection group were told to speak with the person who sat beside them and try to learn something interesting about that person. Those in the solitude group were asked to enjoy their solitude and avoid connecting with anyone. The control group received no special social instruction and simply followed its normal routine.
Eighty-nine per cent returned the survey. After the researchers removed eight people who could not follow their assigned instruction, including seven connection participants who had nobody sit beside them, the analysed sample contained 97 commuters. Their average age was 49, and 61 per cent were women.
Random assignment matters. It makes the different instruction the most plausible cause of the difference found between these groups. It does not, however, make a relatively small sample from one suburban Chicago station representative of every transport system, culture or passenger.
The conversation condition produced the more positive ride
After the journey, participants rated how pleasant it had been compared with their usual commute. They also rated positive and negative feelings. The researchers standardised and combined those responses into a positivity score.
The connection group reported a significantly more positive experience than the solitude group. The statistical comparison was t(94) = 2.49, p = .02, with an effect size of d = .63. In plain language, the observed difference was unlikely to be random noise under the study’s model and was moderate in size within this sample.
The normal-commute group landed between the connection and solitude groups. It did not differ significantly from either. That midpoint is important: the experiment supports a specific connection-versus-solitude contrast, not the stronger claim that talking was proven superior to every ordinary commute.
For the people who did talk, the conversations averaged 14.2 minutes and were rated 5.1 on a six-point pleasantness scale. Productivity ratings did not differ significantly across the three conditions. The exchange seemed to make the ride feel better without producing measurable evidence, on the researchers’ scale, that conversation had made commuters less productive.
A separate group expected exactly the reverse
The prediction result came from Experiment 1b, which involved another 66 commuters. These were not the same people who had completed the journey experiment. They read one of the three instructions and imagined how they would feel if they had to connect, remain alone or commute normally.
The forecasters expected the connection condition to feel less positive than solitude, t(64) = 2.46, p = .02, d = .41. They also expected it to be less positive than the control condition. Their ranking therefore ran in the opposite direction from the key result reported by the travellers who actually followed the instructions.
That separation between samples is easy to lose in a memorable headline. No individual commuter was shown to predict misery and then personally discover delight. The evidence instead compares the average forecasts of one group with the average reported experiences of another.
A University of Chicago account of the study captured the surprise, but the paper’s careful version is more useful: commuters underestimated how positive a stranger conversation would be on average under the conditions tested.
Why a quiet carriage may send the wrong signal
Silence is visible; willingness is not. When nearly everyone keeps to themselves, each passenger can read the carriage as evidence that other people prefer privacy. Yet the same pattern can emerge when several people would accept a conversation but nobody wants to risk being the first to begin.
The researchers explored this as a coordination problem. People may know that they themselves could enjoy a conversation while doubting that the stranger beside them would. Because both people can make the same cautious inference, mutual openness can produce mutual silence.
Related experiments help explain why that forecast may be sticky. In research on the “liking gap” after conversations, people systematically underestimated how much their partners liked them and enjoyed their company. We have direct access to our own hesitation and awkward thoughts, but far less access to another person’s warmer private assessment.
Evidence on weaker relationships also gives brief encounters a place in the wider social picture. A series of studies found that interactions with acquaintances were associated with happiness and belonging. A separate field experiment at a coffee shop found that customers instructed to create a genuine social interaction with a barista reported more positive affect and belonging than those instructed to be efficient. That coffee-shop result does not turn service workers into emotional resources; it suggests that even a small moment of mutual recognition can change how an ordinary transaction feels.
Later train research sharpened the explanation
A later field experiment with London-area train commuters replicated the positive experience of talking: passengers assigned to speak with a stranger reported a more positive journey and said they learned more than those assigned to solitude or a normal commute.
But the London work revised part of the original explanation. Those commuters did not necessarily expect the conversation itself to be worse. They expected the attempt to start one to be less pleasant because they anticipated that other people would not be interested. The feared cost sat at the threshold, in the possibility of rejection, rather than in the exchange once it was under way.
This is a useful refinement rather than a contradiction. Forecasting errors can change across places and times, and the social norm of one rail network may not match another. The later experiments suggest that “people think conversation is unpleasant” is too blunt. Some people may expect a pleasant talk while still believing that initiating it will be unwelcome.
The finding offers permission, not a social obligation
Nothing in these results means that headphones are antisocial, that a quiet commuter is mistaken, or that a passenger owes a stranger access to their attention. Solitude can be restorative. Noise, fatigue, disability, cultural norms, safety and unequal power can all change whether an approach feels friendly or intrusive.
The study also did not test a technique for making anyone talk. The gentlest interpretation is to become slightly less certain that a brief opening will be rejected. An unforced comment or question can leave room for the other person to engage, answer briefly or return to quiet. For readers who do want to open a conversation, our guide to asking a real question and following up explores that listening-first approach.
The Chicago experiment changes the confidence with which avoidance can be treated as an accurate forecast. Sometimes silence genuinely is the kinder, safer or more restorative choice. Sometimes it merely feels safer in advance, while a low-stakes exchange makes an ordinary ride warmer than either person expected.