Reading other people's minds: how to notice your own guess

Epigraph

“gaining new information”

Tal Eyal, Mary Steffel, Nicholas Epley, Perspective Mistaking, 2018; short abstract excerpt. Its context is understanding another person accurately, rather than a guarantee about every conversation.

What this article is about

A message has been read, but there is no reply. That is enough for an entire conversation to appear in your head: the person dislikes your proposal, has decided to withdraw, or thinks you are intrusive. They have said none of those things. Yet you proceed as though you had received an official explanation: cancel the meeting, answer more coldly, protect your dignity in advance. The other person then notices your coldness, and the guess acquires fresh confirmation. From outside, this resembles an exchange of information. In reality, one participant is talking to a version of the other that they have created themselves.

We need the ability to infer intentions in order to live together. The problem begins when the distinction between inference and knowledge disappears. This article follows that transition: what information we actually received, what filled the gap, what can be clarified, and what choice remains when clarification is impossible. It does not prohibit intuition or require trusting everyone. It concerns a precision that lets us protect ourselves while allowing another person to differ from our explanation.

I. Silence contains less information than our account of it

Begin with an imagined scene. You sent a colleague a draft at ten in the morning. By evening, they have not replied. The observable fact is limited: there is no answer in this channel yet. Even a read receipt adds little: the application registered the opening of the message. It does not establish how much the person read, what they understood, when they intend to return to the text, or what tasks they are doing now. Your hurt is real as an experience; it need not wait for a protocol of evidence. But its reality does not automatically make your explanation of the silence correct.

The statement “they are ignoring me” can express different claims. First: no reply is arriving, and I find that unpleasant. Second: the person is deliberately withholding an answer to show my insignificance. The first describes an event and its effect. The second adds intention, purpose, and an assessment of the relationship. We often move between them without noticing because ordinary speech compresses a chain of inferences into one verb. Unpack that verb, and something previously obvious becomes unknown.

This does not mean that every explanation is equally plausible. The colleague may previously have warned about their workload, may consistently fail to respond specifically to your requests, or may explicitly have said they will not discuss the project. History changes the strength of the inference. But it is more useful to preserve history as particular actions: three agreed deadlines passed without a reply; two requests received no response. That record supports a decision about collaboration. A story about secret contempt may add nothing to the decision while requiring constant defence against an invisible opponent.

There is a simple test of the boundary. Ask which part of the claim could appear in an external observer's record and which requires access to thoughts. The observer could see the email, the time, and the subsequent answer, and perhaps hear a participant's explanation. They cannot see the motive directly. Even the participant's explanation remains their account of a motive, assessed alongside behaviour and circumstances. This test does not turn life into a trial. It restores the scale of our words, especially when an important conversation or decision depends on a few of them.

II. A guess becomes convincing when something important is at stake

Our own guesses rarely sound inwardly like cautious suppositions. They arrive with bodily readiness to respond, an image of a familiar scene, and memories of similar humiliation. Someone may therefore say, “I can just feel that they are angry.” Two messages are joined here: I experience tension; I explain it through another person's anger. The first deserves attention. The second needs a basis. Separating them does not devalue feelings. It prevents feelings from performing the role of an information source about another mind.

We fill gaps particularly quickly where personal meaning is large. An ordinary “later” becomes a rejection of closeness; a brief comment becomes a verdict on competence; a missing invitation becomes proof of being unwanted. Each inference concerns both the other person and who we become beside them. Testing that account is difficult: abandoning it sometimes means returning to uncertainty, and a definite piece of bad news may seem more bearable than an open question. This is an authorial description of a possible logic, rather than an established explanation of every reader's behaviour.

A confident account also saves the effort of choosing. If the person has rejected me, I need not decide how to ask for clarity and withstand the answer. If they are jealous, I can avoid considering their substantive objection. If they are punishing me with silence, my silence in return looks like protection rather than one possible action. The convenience of an explanation does not make it false. It shows why testing should include its consequences: what action it already justifies, what information it makes unnecessary, and what alternative I have stopped noticing.

It helps to ask not “why am I so suspicious?” but “what would change if this account were wrong?” Perhaps I would send an ordinary reminder instead of an accusation. Perhaps I would acknowledge that the comment was specific and fair. Perhaps nothing would change: the deadline was missed regardless of the reason, and the work must still go to someone else. That last case matters. Not every unknown must be uncovered. Sometimes precision means leaving the motive unestablished while making a firm decision from behaviour that is already known.

III. Imagining another person's position does not mean knowing it

The advice to “put yourself in their shoes” can make an explanation more humane. Instead of a malicious colleague, we imagine someone with a tired child, a difficult project, or their own fear of making a mistake. Yet imagination retains its original material: we use what we already know or regard as plausible. A kinder account is not necessarily a more accurate one. We can replace an accusatory invention with a compassionate invention without acquiring any new information.

The author-hosted abstract by Eyal, Steffel, and Epley, 2018 reports that imagining another perspective did not consistently improve accuracy across 25 experiments; conversation improved it in the final experiment. Only the abstract was read. The finding guarantees neither truthful replies nor conversational safety.

This distinction changes the form of care. Sometimes care means refraining from assigning the feelings we consider appropriate. “That must hurt, perhaps?” allows correction. “I know it hurts; you are just hiding it” closes correction in advance. Even a benevolent explanation can deprive someone of authorship of their own experience. This is especially visible across cultural habits: restraint, directness, a pause, and a smile need not serve the same purpose in every family and social setting.

We can use imagination to prepare options rather than deliver a verdict. At a minimum, it helps us formulate a question that can accommodate several answers. “Are you not replying because you don't value my work?” invites either confirmation of an accusation or self-defence. “I need to know whether you can read the draft by Friday” seeks necessary information. If the relationship itself is troubling, a direct question could be: “I've noticed that we communicate less; do you notice that too?” No answer completes understanding forever. But the conversation now has a chance to change the original script.

IV. Ancient advice and modern verification meet at only one point

In The Analects, II.10 and II.17, Legge translation, II.10 trusts the readability of character; II.17 distinguishes knowing from not knowing. We retain distance from the first proposition and compare the second with precision in our own wording. This is a moral framework, not a contemporary protocol for measuring another person’s thoughts.

Epictetus’s Enchiridion §45, Higginson translation, limits adding a verdict without knowing an action’s basis. This is Stoic ethics, rather than a rule to endure harm.

There is a useful tension between these approaches. One directs attention to actions; another limits the added evaluation; the modern study stresses new information. Remove the differences, and we get an attractive slogan: every tradition teaches us to understand people. Preserve them, and a more precise question appears: what does observation tell us, what does conversation add, and when does moral judgment become necessary? We can use the sources together without requiring them to confirm a single doctrine.

Suppose an employee publicly claims someone else's result. We may not know whether they intended humiliation or acted carelessly. But authorship and the public claim can be established separately. Correcting the record, restoring the author's name, and changing the process need not wait for access to the motive. The practical point of intersection is narrow: do not invent the invisible portion of the case. That coexists with assessing visible actions. Philosophical caution becomes useful when it supports more precise action rather than postponing necessary action until impossible complete clarity arrives.

Imagine three participants in a shared project. One awaits an answer; the second still awaits initial material from the third. Silence readily becomes a story of personal disregard although the work has a separate chain of responsibility. Ask which material is absent and who can provide it. Everyone need not be declared conscientious beforehand. The organisation obscured by a story of hidden attitudes needs making visible. Clarification may reveal a repeated violation, which can be named specifically, or a broken handoff, which can be corrected. Either way, a new agreement concerns actual shared work rather than only an inward assessment of one colleague. It permits checking subsequent performance and distinguishing agreement to participate from our supposition that everyone has already accepted the role assigned to them.

V. Alternatives are for testing, rather than excusing

A list of possible explanations can be irritating. Someone describes rudeness and is told that the speaker was tired, had a difficult day, or struggles to express care. This can sound like a request to consider the harm insignificant. An alternative serves a different purpose: to separate the basis of an assessment from a supposition about its cause. A rude remark remains rude; an agreement remains broken. An unknown motive does not cancel consequences or create an obligation of endless indulgence.

Two or three genuinely different accounts are enough for testing. The colleague has not replied because they lack time, did not realise a reply was required, or do not want to participate. The distinction helps if the possibilities lead to different questions or deadlines. Ten fantastical stories generally expand the internal serial drama. An account that cannot be distinguished through any available observation contributes little to choice. Explanations that absorb every answer are particularly unhelpful: denial means concealed anger; admission means the guess is proven.

An honest test permits an outcome we dislike. The other person may really be withdrawing. Their answer may be evasive. They may not want to discuss the relationship. The original supposition then gains some support, yet precision remains: they said they do not want to meet; they did not give a reason; I choose to stop inviting them. This wording does not provide the desired final picture of their character. It does spare us the repeated task of extracting proof of our own unwantedness from every subsequent message.

There is an opposite trap: demanding evidence of perfect benevolence from everyone around us. “I do not know why they do this” quietly becomes “I am not entitled to find this unacceptable.” The right to establish a boundary rests on what is happening and your conditions of participation. If every conversation ends in insults, you can reduce contact without a theory about the other person's hidden trauma. If a business partner again breaks a promise, you can choose another partner. Uncertainty about motives and certainty about your boundary can coexist; neither need wait for the other.

VI. A question can invite correction or close it off

Clarification begins long before the question mark. “Why are you displeased with me again?” has already established displeasure and repetition. The other person must first contest the premise, and the conversation can become a struggle over who describes reality correctly. A more open form names an observation and its personal effect: “You went quiet after my suggestion, and I didn't understand what that meant.” Then we can ask about substance: is there an objection, is more time needed, or did the person want to change the subject?

Do not hide your own request inside an investigation of someone else's psychology. If you need a deadline, ask for one. If you want support, name that wish. If a remark hurt, explain which part. Otherwise a conversation about intentions becomes an indirect demand for changed behaviour. The person may spend a long time assuring you of good feelings while the necessary agreement never appears. A precise question ties the unknown to the actual task that brought you into the conversation.

The other person retains the right not to answer or not to know. They may themselves be unsure why they have called less often. They may remember an event differently. Such answers do not automatically prove either sincerity or deception. They establish the limit of the information received. It helps to hear a correction fully before explaining why it does not fit your account. If only the expected answer is permissible, the external conversation becomes another scene in the internal play: the other person is physically present, but their words cannot alter the script.

A question is not appropriate in every relationship. Direct clarification can sometimes create a risk of pressure, retaliation, or further humiliation. We can then use documents, witnesses, independent advice, or our own decision to establish distance. The discipline of checking is not a rule to “always ask.” It involves choosing an available and proportionate information source. Where no safe source exists, an honest statement of uncertainty is better than artificial confidence. It permits us to stop blaming ourselves for lacking clarity that the situation does not yet allow.

VII. A guess can help create the next event

Return to the unanswered message. You decide the colleague disregards you and speak curtly at the meeting. They notice your sharpness, stop offering help, and you read this as the revelation of their true motive. In this imagined scene, the original account influences your behaviour; your behaviour influences a new reaction; the reaction strengthens confidence in the account. This is a possible interaction sequence, not a proven universal mechanism of every conflict. It matters because the latest fact is not always independent of actions based on the first guess.

The hardest task is seeing our participation without turning it into self-accusation. “I may have influenced the conversation” does not mean “I caused everything that happened.” The other person remains responsible for their response. The environment has conditions of its own. But the possibility of participation opens a small choice: return to a specific question, acknowledge your sharpness, propose a discussion time. This is neither a duty to reconcile first nor a promise of a good outcome. It is a way to test whether changing one accessible link changes the interaction.

Instead of another dry answer, you might write: “I spoke sharply yesterday because I was upset about not receiving a reply. It still matters to me to know whether you can read the document.” This acknowledges your action without assigning their motive and brings the conversation back to the task. If clarity still does not arrive, you have a new fact obtained under more definite conditions. If an answer comes, it can be considered separately from the earlier script. Either way, the test supplies material for revision rather than merely an opportunity to win an argument.

Sometimes revision must go in a favourable direction. The person replies with a substantive comment, yet we keep looking for a trace of contempt. We need not demand immediate happiness, but we can give the fact a place in the history. Yesterday there was no reply; today there was one; the explanation was this; trust is returning gradually, or is not returning. A sequential record preserves the unpleasant event without making it govern every later meaning forever. Precision helps both to detect harm and to notice change when it actually occurs.

VIII. Sometimes the main question concerns ourselves

We may try for a long time to discover what someone else thinks because our own decision seems permissible only after their verdict. Can I leave a project if my partner still values me? Can I ask for quiet if someone close did not intend offence? Can I decline a meeting if the person is good? In these cases, an unknown motive becomes a condition of the right to act. But your participation, time, and boundaries remain yours without a complete explanation of someone else's inner world.

Separate two questions: “What can I learn about them?” and “What do I need for my decision?” The second is often narrower. Repeated missed deadlines can be enough to change a workflow; your available resources can be enough to reduce a load; finding the present conversation unsuitable can be enough to request another form. This does not release us from agreements or consequences. It releases us from having to prove someone else's badness before acknowledging our need or suggesting a change.

A particular form of respect appears here: allowing the other to remain partially unknown. We may know someone sufficiently for friendship or collaboration while not knowing the meaning of every pause. Closeness does not require complete access to thoughts. It may include asking, hearing a refusal, revising an earlier account, and agreeing again. When the unknown stops automatically meaning threat or personal defeat, conversation gains room. This is a possibility, not a standard of maturity: in some relationships, uncertainty really is too costly.

Looking again at the screen with the read message, you now see two unfinished processes. One concerns the colleague: what they are doing and when they will answer. The other concerns you: what meaning you have already assigned to the pause and what you intend to do. You need not wait indefinitely. You can clarify, set a limit on waiting, or transfer the task. But an absent answer no longer obligatorily equals your unwantedness. This small distinction changes the quality of action: it rests on a situation that can be discussed instead of a verdict carried in silence.

IX. Benevolent certainty also needs room for an answer

Mind reading does not always begin with suspicion. Sometimes we are certain a person will like a gift, consent to help, or want to join a family celebration. The account looks generous and rarely comes under scrutiny. Its consequences may nevertheless be difficult: help intrudes on another person's arrangements, a gift creates an awkward obligation, a celebration demands unavailable strength. The giver's intention and the recipient's experience are separate parts of the event. We can retain a good intention without assigning its result in advance.

Imagine an adult daughter buying her parents travel tickets without asking. She knows they once dreamed of visiting that country. An earlier desire does not establish a present one: health, plans, attitudes to travel, or financial priorities may have changed. Instead of a surprise, the parents receive the task of explaining a refusal without hurting their daughter. The problem in this scene is neither too much love nor ingratitude. Another person's agreement entered the plan before entering the conversation. A question before the purchase could have preserved more freedom for everyone.

Even knowledge of a stable habit requires adjustment to the particular occasion. A friend likes visitors but wants no company today. An employee usually accepts difficult tasks but now chooses a lighter load. A partner previously asked for advice but wants to be heard in this conversation. We need not erase the relationship's history at every meeting. However, history helps create an expectation rather than removing the right to change. A clarification can be brief: “Would help with the solution be more useful now, or a chance to tell me about it?” It matters only if both answers are truly permissible.

A benevolent guess may also be right. The friend will welcome the visit, the parents will want the trip, and the advice will be timely. Testing does not require exposing every success as chance. It helps identify the expectation's actual basis: a recent request, a joint plan, explicit consent, or only our notion of what is good. The more costly an action and the harder refusal becomes afterwards, the more reason to clarify first. This is a proposed editorial principle of proportion, not a demonstrated psychological law. Its practical meaning is simple: care allows the recipient to say which care suits them and which becomes an additional task today.

There is another question about the chosen criterion. If you check a response deadline, the answer concerns that deadline. If you check whether working together is possible, information about conditions of participation is needed. Every clarification need not simultaneously test love, respect, and character. One question can remain small while sufficiently supporting the next action. Separating criteria helps us accept a concrete answer without requiring it to confirm the entire relationship’s history.

What this means in practice

Choose one recent episode in which you confidently explained another person's motive. Record the event, your account, and your experience separately. For instance: no answer since Tuesday; I suppose the proposal is uninteresting; I feel disappointed and anxious. Add one genuinely possible alternative. Decide which information would change your next step. If you only need a deadline, do not conduct an investigation into the relationship. If the relationship itself matters, name the observable change and leave room for an answer.

Before clarifying, define your own boundary: how long you will wait, what you will do without an answer, and which channel is safe. Afterwards, reread the original record. Note what was confirmed, what changed, and what remains unknown. Do not make the exercise into permanent monitoring of every facial expression. Its purpose is to restore choice where a guess has already chosen for you. If testing increases inward preoccupation without supporting a useful decision, you can stop and return to the concrete task.

A successful result may be a more precise sentence rather than confidence in someone else's goodness: “I don't know their intention, but I know which agreement I need.” Sometimes this clarity supports conversation; sometimes it supports distance. Revising an account and protecting a boundary do not conflict. Together, they let us interact with a real person who may answer unexpectedly rather than a person whose answers are already entirely written in our head.

Keep this record brief and reconsider it after a new event. It supports a specific decision rather than becoming a dossier on another person or proof of your own infallibility.

Key quotation

“getting perspective, not simply taking perspective”

Eyal, Steffel, Epley, Perspective Mistaking, 2018; excerpt from the abstract’s final sentence. The conclusion concerns the accuracy studied; safe and truthful conversational conditions require separate assessment.

Connections between traditions

Comparison	Actual similarity	Significant difference
Confucius II.10 ↔ checking behaviour	Attention to actions	Moral instruction is not a protocol for checking another person’s thoughts
Confucius II.17 ↔ precise wording	Knowing distinguished from not knowing	Moral cultivation differs from scientific measurement
Epictetus §45 ↔ checking motives	Limiting added judgment	Stoic ethics does not replace responsibility for consequences
Eyal and colleagues ↔ asking the person	Value of new information	Experimental findings do not guarantee truthful answers everywhere
Compassionate imagination ↔ dialogue	Considering another perspective	An imagined perspective and an obtained account have different evidential bases

Primary sources

Confucius. The Analects, James Legge translation, PG3330; II.10, II.15, II.17. The named passages and adjacent context were read; a moral and philosophical source.
Epictetus. The Enchiridion, Thomas W. Higginson translation, Liberal Arts Press 1955 edition, PG45109; §45. The selected paragraph was read, rather than the complete Stoic corpus.
Tal Eyal, Mary Steffel, Nicholas Epley. Perspective Mistaking, 2018, JPSP 114(4), 547–571, DOI 10.1037/pspa0000115. The author-hosted abstract was read; the tool did not obtain the complete PDF. Rights UNKNOWN; no full-text republication.

About the authors

Andrey Zarinsh is the author and founder of United Wisdom. He explores inner transformation, the unity of traditions, intercultural dialogue, and contemporary spiritual discipline. Instagram: @andrey.zarinsh.

Maria Evilkut is a coauthor of United Wisdom. Alongside her research, she seeks to test these ideas in practice and observe how principles of unity and mutual understanding work in real life between people and cultures. Instagram: @ilkutki.