Featured image: AI-generated illustration of an everyday conversation, not a photograph of research participants.
The conversation ends. The guessing begins.
The part that catches my attention is what happens after the talking stops. Two people can leave the same conversation carrying different guesses about whether the other person enjoyed it, and neither guess has to be a reliable account of what happened.
Picture an ordinary first meeting, not a dramatic disaster: a new neighbor, a colleague you have barely met, someone beside you at a community event. There was enough to talk about, but afterward you remember the sentence that came out awkwardly.
It is tempting to turn that remembered moment into an answer. Perhaps you talked too much, seemed boring, or made the other person wish they were somewhere else.
What interests me is the leap between those two things: noticing your own discomfort and deciding you know someone else’s opinion. They can feel like the same piece of information, even though they come from entirely different places.
A research paper published in 2018 put that gap under scrutiny. Its question was simple enough to feel personal: after a conversation, how closely does our estimate of being liked match what the other person actually reports?
The researchers did not answer by asking people to repeat reassuring slogans. They asked both sides of the conversation.
Two people, two private sets of answers
In the original paper in Psychological Science, Erica Boothby, Gus Cooney, Gillian Sandstrom, and Margaret Clark examined what they called the liking gap. Their work included laboratory conversations, workshops, and students getting to know their college suite mates.
In an initial study, previously unacquainted people talked for about five minutes with icebreaker questions available. Afterward, they went to separate rooms and answered questions about how much they liked their partner and how much they believed their partner liked them.
The questions went beyond whether someone seemed pleasant. They also asked about interest in getting to know the other person, interacting again, and potentially becoming friends.
On average, people underestimated how positively their conversation partner regarded them. The researchers found a similar pattern in another study where participants could talk about whatever they wanted rather than follow supplied questions.
That is a narrower finding than saying everybody secretly loves you. It means the participants’ estimates tended to be less favorable than the ratings supplied by the people they had actually spoken with.
I find that distinction more useful than blanket reassurance. Nobody has to promise a perfect social world for it to be worth questioning a harsh conclusion drawn from incomplete evidence.
The uncertainty is real. The negative verdict we attach to it may not be.
The signals were not necessarily missing
One obvious explanation would be that people liked each other but gave no sign of it. If friendliness stayed entirely inside their heads, underestimating it would be understandable.
The researchers looked more closely by having trained observers watch recordings from the initial study. Those observers’ judgments predicted participants’ reported liking, suggesting that some useful signals were visible from outside the conversation.
Yet the participants’ own estimates did not track those observer judgments in the same way. The problem was not simply a complete absence of signs that the meeting had gone reasonably well.
In Yale’s account of the research, Clark described people becoming occupied with worries about what they should say or had already said. That offers a possible explanation for missing warmth that another person could notice.
The paper also examined people’s thoughts afterward. Participants’ ideas about how they had been viewed were more negative than their own thoughts about their conversation partners, and that difference was related to the size of the liking gap.
This is the detail I keep returning to: you have access to every version of the sentence you wish you had delivered. The listener heard the sentence you actually said, without a running comparison to all your imagined improvements.
That interpretation does not make every worry mistaken. It does explain why your private replay should not automatically count as a recording of someone else’s reaction.
Nor did a longer conversation necessarily settle the matter. In a study where conversations ranged from two to forty-five minutes, the researchers still found an overall tendency to underestimate liking and enjoyment.
What the finding can and cannot offer
The research extended beyond brief meetings. Among college suite mates followed through an academic year, the gap appeared across several months, although it was no longer evident at the final measurement in May.
That matters because another conversation does not always erase uncertainty immediately. At the same time, it would be wrong to turn this study into a rule about how long any particular friendship takes to feel secure.
These were particular samples and settings, including university participants and people attending conversation workshops. The findings do not establish that every culture, workplace, age group, or difficult social situation follows the same pattern.
The paper also reports that its design and analysis plans were not preregistered. It is useful evidence about a tendency, not a machine for reading an individual person’s mind.
Most importantly, uncertainty about being liked is different from a clear boundary. A refusal, repeated unwelcome contact, or explicit request for distance is not something to reinterpret as hidden affection.
I would not use this research to tell someone to ignore rejection. I would use it to question the assumption that feeling awkward proves rejection happened.
If social fear is persistent, distressing, or limiting daily life, a research article is no substitute for appropriate professional support. Being told to think positively can be especially unhelpful when the difficulty is larger than an occasional doubtful evening.
A little room before the verdict
One practical response is to separate what you observed from what you inferred. You might know you paused before answering; you do not necessarily know the other person found the pause irritating.
Another is to give your own positive reaction a clear, low-pressure expression. If you enjoyed meeting someone, saying so plainly can be more useful than hoping they will correctly decode a polite goodbye.
These are practical suggestions inspired by the findings, not interventions proven by this paper. They leave room for the other person to respond freely rather than turning connection into another performance test.
You can also allow a relationship to develop through ordinary repetition: another greeting, another shared activity, another conversation when both people seem willing. There is no need to demand immediate certainty from a first meeting.
For me, the encouraging part is not a promise that everyone will like us. It is the possibility that an imperfect conversation can still have been genuinely welcome.
You may remember the stumbled sentence more vividly than the shared laugh. Before you decide which one defined the encounter, leave a little space for the person who experienced it from the other side.
