Stop assuming you know what others think

The **closeness-communication bias** is the tendency to overestimate how well you can read the thoughts, intentions, and feelings of people you know well. **Savitsky, Keysar, Epley et al. (2011)** demonstrated that long-term couples communicate no more accurately than strangers on objective tasks — but they *believe* they communicate far better. The culprit is familiarity: knowing a lot about someone makes confident inference feel easy, which is exactly when careful listening stops. The bias is strongest in our closest relationships, not our weakest ones.

Because familiarity breeds **false confidence**, not accuracy. When you know someone's history, habits, and preferences, your brain stops treating each new situation as genuinely unknown — it runs a shortcut based on prior knowledge. **Nicholas Epley** calls this the _curse of knowledge_: the more context you carry, the harder it is to imagine that someone might think or feel something outside your mental model of them. The result is that you stop asking and start concluding. It feels efficient. It is often wrong.

Yes — persistently and quietly. When you assume you know what someone means, you stop gathering real information. You respond to the person in your head rather than the person in the room. Over time, **Kate Murphy** (in *You're Not Listening*) argues this produces a compounding gap: each assumed interpretation slightly misses the mark, and those near-misses accumulate into misunderstanding. The person on the receiving end often senses they are not being heard, even if they cannot name why. _Being half-right most of the time is not the same as listening._

**Egocentric bias** is the tendency to see shared events primarily from your own perspective and assume others experienced them the same way. **Epley** documents how this distorts memory and judgment: two people can attend the same dinner and walk away with genuinely different emotional experiences, but each privately assumes the other felt what they felt. In relationships, this produces arguments where neither person is technically lying — each is reporting their own experience — but both are confused by the other's 'inaccuracy.' Recognising egocentric bias is the first step toward [perspective-taking](/en/blog/perspective-taking) rather than competing over whose memory is correct.

Start with the smallest possible ask: a single, open question about *this moment* rather than a grand conversation about the whole relationship. **Benson** (in *Why Are We Yelling?*) recommends speaking for yourself first — state what you observed and how you interpreted it — then explicitly invite the other person to correct you. Something like 'I read that as frustration — is that what was happening for you?' is more productive than 'you seemed angry.' It signals that you are genuinely uncertain, which gives the other person real permission to tell you something different. See also our guide on [how to ask better questions](/en/blog/how-to-ask-better-questions).

Yes, but the improvement comes from **slowing observation**, not from trusting your intuitions more. **Anne Lamott** (in *Bird by Bird*) advises watching carefully before interpreting — notice what someone *does* before deciding what it *means*. Practical steps: pause before drawing a conclusion, ask one clarifying question, and treat your first interpretation as a hypothesis rather than a fact. Research on interpersonal accuracy consistently shows that people who ask more questions and assume less are the ones who end up reading others most reliably — not the ones who trust their gut the hardest.

Because **familiarity replaces attention**. After years together, partners build elaborate mental models of each other — and then consult the model instead of the person. **Epley** found this pattern across multiple studies: the length of a relationship predicts confidence in understanding a partner, but not actual accuracy. Kate Murphy reinforces the point in *You're Not Listening*: closeness creates **complacency** — the feeling that you already know what they are going to say is the exact moment you stop hearing what they actually say. Long relationships need *more* deliberate checking-in, not less.

The **illusion of transparency** is the mistaken belief that your internal states — feelings, intentions, nervousness — are more visible to others than they actually are. **Epley and Gilovich** documented this: people who are nervous in public consistently overestimate how nervous they appear. In relationships, this runs both ways. You assume your partner *can see* that you are upset (so you do not say it), and you assume you can see their internal states clearly too. Both assumptions are wrong. Feelings that feel obvious from the inside are often invisible from the outside — which is why naming them out loud is rarely as redundant as it feels.

Once you have formed a belief about someone — 'she is dismissive,' 'he is needy,' 'they always catastrophise' — that belief filters the information you take in. **Philip Houston, Michael Floyd & Susan Carnicero** (in *Spy the Lie*) show how this happens in high-stakes interviews: an interviewer who expects deception finds it everywhere. The same mechanism operates in ordinary relationships. If you are convinced a friend is jealous, you will read neutral behaviour as confirmation. **Prior beliefs do not just colour interpretation — they actively screen out contradicting data.** The antidote is to approach each conversation as if you do not already know how it ends.

A central one. **Active listening** is specifically the discipline of suspending your interpretation long enough to take in what was actually said — rather than what you expected to hear. **Murphy** argues in *You're Not Listening* that most people are preparing their response while the other person is still speaking. That means you are listening to the voice in your head, not the voice in the room. Techniques that help: withhold your response for two full seconds after someone finishes, ask a follow-up before offering your view, and reflect back what you heard before disagreeing. Our full guide on [active listening](/en/blog/active-listening) covers the mechanics in detail.