How to Build Empathy Without Assuming You Know What Others Feel
Empathy begins with an uncomfortable fact: you do not directly experience another person's mind.
You hear their words, notice their behaviour, interpret tone and expression, remember what you know about their circumstances and compare some of it with your own experience. From those pieces of information, you construct an explanation of what the other person may be thinking or feeling.
Sometimes that explanation is remarkably accurate.
Sometimes it is projection.
Someone becomes quiet during a meeting and you conclude they are angry. They were concentrating. A friend declines an invitation and you assume they are offended. They are exhausted. A colleague appears unusually formal and you interpret it as dislike when they are worried about an unrelated problem.
The difficulty is that empathy often feels more certain than the evidence justifies.
Psychological research also does not treat empathy as one perfectly unified ability. Researchers distinguish processes such as understanding another person's perspective, sharing or resonating with emotional states, and experiencing concern for another person's welfare. These processes can overlap, but they do not always occur together. Research has even found evidence that empathic emotional sharing and concern for another person's welfare are psychologically distinguishable rather than simply two names for the same response. (pubmed.ncbi.nlm.nih.gov)
A practical approach to empathy therefore needs imagination and restraint.
The most useful habit may be:
Treat your interpretation of another person's experience as a hypothesis that the other person is allowed to correct.
That turns empathy from confident mind reading into disciplined curiosity.
Better empathy often comes from getting perspective, not merely imagining it
The phrase “put yourself in their shoes” sounds like the essence of empathy.
It can be useful because it interrupts the assumption that your own perspective is the only reasonable one. Before condemning an employee as careless, you may ask what pressures or information were available to them. Before dismissing someone's fear as irrational, you may consider what risks the situation contains from their position.
But imagination has an important limitation:
you are still the person doing the imagining.
Your assumptions, experiences and expectations travel with you into the other person's supposed shoes.
This creates one of empathy's most common errors. Instead of asking what they feel in their circumstances, you imagine what you would feel if you were in what you believe their circumstances to be.
Those are not necessarily the same thing.
Research gives unusually strong support to this caution. In 25 experiments, Tal Eyal, Mary Steffel and Nicholas Epley tested whether deliberately taking another person's perspective improved accuracy when predicting thoughts, feelings, preferences and other mental states. They found no consistent evidence that perspective-taking instructions improved interpersonal accuracy; if anything, perspective taking sometimes increased confidence without increasing accuracy. In a final experiment, however, getting the other person's perspective through direct conversation did improve accuracy. (pubmed.ncbi.nlm.nih.gov)
The distinction is extremely useful.
Perspective taking asks:
“What do I imagine this person thinks?”
Perspective getting asks:
“What does this person say they think?”
The first can prepare you to ask better questions.
The second gives you information you did not previously possess.
Another meta-analysis by William McAuliffe and colleagues examined experiments in which participants were instructed either to imagine the perspective of a distressed person or to remain objective. Interestingly, instructions to imagine the person's perspective did not reliably increase empathic concern relative to receiving no instructions, although instructions to remain objective reduced concern. (pubmed.ncbi.nlm.nih.gov)
So perspective taking should not be treated as a magical empathy switch.
The better principle is:
imagine enough to become curious, then gather information.
Suppose a colleague says, “I had to redo the report three times.”
You could immediately conclude that they are angry about the extra work.
A better response might be:
“Was the main problem having to redo it, or that the requirements kept changing?”
Perhaps they answer:
“No. I didn't mind revising it. The frustrating part was that nobody explained why.”
Your first interpretation was wrong.
That correction is not a failure of empathy.
It is empathy working properly.
Research on empathic accuracy also suggests that what people actually say is one of the strongest available sources of information about their internal states. In experimental work comparing different channels of information, verbal content contributed more to accurate inference than visual nonverbal cues alone. (pubmed.ncbi.nlm.nih.gov)
That should make us cautious about popular claims that emotionally skilled people can simply “read body language.”
Facial expressions and posture matter.
Tone matters.
Context matters.
But words matter enormously too.
And when the person themselves can clarify, asking is often better than decoding.
Curiosity makes empathy more accurate than immediate judgment
Human beings make rapid explanations of other people's behaviour because doing so is efficient.
Someone misses a deadline:
irresponsible.
Someone interrupts:
rude.
Someone refuses to help:
selfish.
Someone reacts strongly:
overdramatic.
Sometimes those judgments are correct.
The problem is how quickly a description of behaviour becomes a conclusion about character.
Empathy creates a useful delay between the event and the verdict.
Instead of immediately asking:
“What kind of person behaves like this?”
ask:
“What else could explain this behaviour?”
Maybe the deadline was missed because expectations changed.
Maybe the person interrupting believed the decision was about to be made without important information.
Maybe the colleague refusing additional work is already overloaded.
Maybe the strong emotional reaction connects to something you do not yet know.
You do not need to invent a sympathetic explanation and decide it must be true.
That would simply replace one assumption with another.
Curiosity means keeping several explanations available long enough to gather better evidence.
Questions help.
“What was the hardest part from your side?”
“What were you expecting to happen?”
“What mattered most to you?”
“What constraint was I missing?”
“What did you want us to understand?”
These questions invite the person to supply information rather than requiring you to manufacture it.
Short paraphrases can test your understanding in the same way.
“So the problem wasn't the additional work itself. It was getting the change after you'd already completed the first version.”
The person can say yes.
Or they can say:
“No, that's not quite it.”
That “no” is valuable.
Empathy should make correction easier, not make the empathic person increasingly committed to their first interpretation.
Your own similar experience can help—and distort
Shared experience can make people feel understood.
Someone describes caring for an ill parent and you have done the same. A colleague describes being made redundant and you remember your own redundancy. A friend talks about moving to another country and you recognise some of the uncertainty.
Similarity provides emotional vocabulary and possible explanations that may otherwise be difficult to imagine.
But it can also create projection.
Research involving romantic partners has found evidence of both accuracy and projection when people estimated their partners' emotional experiences: people could perceive their partners accurately while simultaneously allowing their own emotions to influence what they believed their partners felt. (pubmed.ncbi.nlm.nih.gov)
Having experienced something similar therefore does not mean you know what this person feels.
Sometimes shared experience can even interfere with recognition. Experimental research has found circumstances in which recalling one's own similar negative experience reduced accuracy in recognising another person's emotions. (pmc.ncbi.nlm.nih.gov)
So when someone tells you something difficult, notice the impulse to answer:
“Exactly the same thing happened to me…”
Your story may help.
But first ask whether the person has actually finished telling theirs.
A useful internal question is:
“Am I sharing this because it will help them feel understood, or because their story has reminded me of something I want to talk about?”
That small distinction prevents empathy from quietly becoming self-reference.
The same discipline matters when encountering experiences outside your normal social circle. Memoirs, oral histories, high-quality interviews, literature, long-form journalism and direct conversation can broaden the range of lives you have encountered. They may help you notice assumptions that once felt obvious.
But one story should not become a stereotype.
One woman's experience does not reveal what women feel.
One immigrant cannot represent migration.
One autistic person's experience does not establish a universal autistic perspective.
Exposure should expand the number of possibilities you can imagine, not convince you that you have finally decoded an entire group.
Understanding someone does not require agreeing with them
Empathy becomes politically and personally difficult when the person you are trying to understand holds a view you strongly oppose.
That is precisely where the skill becomes useful.
Before arguing against someone's position, try identifying its internal structure.
What does the person value?
What outcome are they afraid of?
Which evidence do they consider credible?
What trade-off are they willing to accept?
What would have to be true for their position to make sense from their perspective?
This is different from asking whether their conclusion is correct.
You can understand the reasoning behind a position and still reject it.
You can tell an employee:
“I understand why changing the deadline feels unfair after you've already rearranged your week. We still cannot move the regulatory filing date.”
The understanding is real.
The decision remains unchanged.
Empathy should improve the accuracy with which we understand other people's interests. It does not require handing those interests veto power over every other consideration.
That is also why empathy should be distinguished from agreement, compassion and rescuing.
Cognitive empathy generally refers to understanding another person's perspective or mental state.
Affective empathy usually refers to sharing or resonating with another person's emotion.
Empathic concern involves caring about another person's welfare.
Researchers debate the exact taxonomy, but these processes are sufficiently different that combining them into one vague idea can create confusion. Reviews of empathy research specifically distinguish affective empathy from emotional contagion, sympathy and empathic concern even though the processes can occur together. (pmc.ncbi.nlm.nih.gov)
A doctor may understand that a patient is terrified without becoming equally terrified.
That emotional separation can help the doctor remain capable of treating the patient.
A manager can understand why an employee is struggling without taking responsibility for solving every difficulty in the employee's life.
A friend can care deeply while still saying:
“I can't do that for you.”
Empathy without boundaries easily turns into over-responsibility.
Boundaries without empathy can become indifference.
Mature relationships often require both.
Be careful with the idea of “empathy fatigue”
Helping professions create real emotional strain.
Doctors, nurses, therapists, social workers, carers and emergency personnel may face repeated exposure to suffering, trauma, high workload and moral stress. People commonly describe the resulting exhaustion as compassion fatigue or sometimes empathy fatigue.
But the terminology deserves care.
A major meta-narrative review of the healthcare literature found that “compassion fatigue” had become widely used despite significant conceptual problems. The authors found little evidence that being an especially compassionate caregiver itself makes someone more susceptible and argued that occupational stress, workload and related job conditions were often being compressed into an imprecise “cost of caring” narrative. (pubmed.ncbi.nlm.nih.gov)
More recent reviews likewise identify individual, organisational and system-level contributors to compassion-fatigue outcomes among healthcare professionals. (pmc.ncbi.nlm.nih.gov)
The practical implication is important.
Do not tell exhausted carers that they simply “cared too much.”
Staffing, workload, traumatic exposure, inadequate recovery, role ambiguity and organisational conditions matter.
Empathy does not require absorbing another person's distress until you are unable to function.
Effective care can include emotional connection alongside enough self-other distinction to continue thinking, deciding and recovering.
Empathy can be trained—but training works better when it changes behaviour
If empathy involves habits of attention, interpretation and response, at least some of those habits should be trainable.
Research broadly supports that conclusion, although effect sizes and durability vary.
A meta-analysis by Emily Teding van Berkhout and John Malouff examined 18 randomised controlled trials involving 1,018 participants after excluding one extreme outlier. Empathy training showed a medium overall effect, estimated at g = 0.63, falling to .51 after adjustment for possible publication bias. The authors also found substantial variation according to population and measurement method. (pubmed.ncbi.nlm.nih.gov)
A larger 2024 systematic review and meta-analysis examined 110 eligible empathy-training studies and separated outcomes into affective, cognitive, motivational, behavioural and composite forms of empathy. The picture was more nuanced: training effects differed across dimensions, some declined over time, and socially oriented approaches were generally the most effective across categories. (pubmed.ncbi.nlm.nih.gov)
That is more useful than saying:
“Empathy can be taught.”
Specific components appear trainable.
But different programmes train different things, and improvement on a questionnaire does not necessarily mean permanent transformation of every aspect of empathy.
For everyday development, focus on behaviour.
Practise asking one additional question before giving your opinion.
When someone makes a decision you do not understand, ask what constraint mattered most.
When a friend is upset, resist offering a solution until you understand what outcome they actually wanted.
When you think you have identified someone's feeling, verify it instead of announcing it as fact.
After difficult conversations, ask:
Did I interrupt?
Did I start giving advice before I understood the problem?
Did I assume that my similar experience meant theirs was identical?
Did I ask enough questions?
Did I allow the person to correct me?
Did the conversation become about me?
This turns empathy into something observable.
Feedback helps for the same reason.
You may intend to be an excellent listener but routinely interrupt because you become excited.
You may believe your advice communicates care while friends experience it as pressure.
You may think you are accurately reading subtle emotion when other people experience your interpretations as intrusive.
Empathy improves when the person you are trying to understand has some influence over your assessment of whether you succeeded.
A practical empathy loop begins and ends with uncertainty
For ordinary conversations, empathy can be reduced to five connected moves: notice, ask, listen, reflect and verify.
Notice that something may be happening without deciding immediately what it is. A colleague is quieter than usual. Your partner seems tense. A customer reacts much more strongly than expected.
Then ask for information. “Is something about this worrying you?” “What part is causing the problem?” “What were you expecting instead?”
Next, listen for more than isolated facts. What outcome did the person want? Which constraint matters? Which emotion are they naming themselves? What are you inferring that they have not actually said?
Then reflect your understanding tentatively: “It sounds as though you were less upset about the decision than about finding out after everyone else.”
Finally, verify: “Is that right?”
That last step protects the entire process.
Without verification, empathy can become an increasingly elaborate story you tell yourself about someone else.
With verification, misunderstanding becomes correctable.
This approach is especially valuable during disagreement because you can separate two questions:
“Do I understand your position?”
and:
“Do I agree with your position?”
The answer can be yes to the first and no to the second.
Empathy therefore does not require emotional certainty.
In fact, one of its most mature expressions is recognising how uncertain our access to another person's internal world really is.
Research on interpersonal accuracy reinforces that point. People tend to believe that adopting another person's perspective will help them understand that person better, yet experimental evidence shows that simply imagining another perspective does not consistently increase accuracy. Direct information from the person does. (pubmed.ncbi.nlm.nih.gov)
So the most empathic sentence is often not:
“I know exactly how you feel.”
It is:
“I want to understand. This is what I think I'm hearing—tell me where I've got it wrong.”
That combines imagination with humility.
You still notice.
You still infer.
You still use your experience.
But you remember that another person's mind belongs to them, not to your interpretation of them.
Empathy grows when we become better at constructing possible models of other people's experience and equally willing to revise those models when new information arrives.
The practical goal is therefore not perfect emotional certainty.
It is better calibration: fewer reflexive judgments, more useful questions, more careful listening and a lower attachment to being right about what another person feels.
That version of empathy is not only kinder.
It is more accurate.
