What emotional intelligence actually is

Emotional intelligence has been sold as the thing that matters more than IQ. Thirty years of research say something smaller and more useful: there are real, measurable emotional skills, they predict real things, and they can be trained. Here's what survives, what doesn't, and how to work on yours.

By Reid Robison, MD. 2,918 words, with six things to try.

Somewhere in your life there is a person who is not especially brilliant and yet everyone wants on their team. They notice when a meeting has gone sideways before anyone says so. They can tell you something hard without making you feel small. When they're angry you can see it, and you can also see them deciding what to do with it. Whatever that is, it isn't on an IQ test, and in 1990 two psychologists, Peter Salovey and John Mayer, gave it a name: emotional intelligence, the ability to monitor feelings, your own and other people's, to tell them apart, and to use that information to guide thinking and action.¹

Five years later the science writer Daniel Goleman turned it into a bestseller, and the idea escaped the lab.² It was a good escape in some ways. Schools started teaching feelings. Workplaces started admitting that people have them. But the claims inflated as they travelled: that EI accounts for 80% of success, that it matters "twice as much" as cognitive ability, that it's the single best predictor of leadership.³ Those numbers were never from peer-reviewed studies, and the research that followed has been, politely, a correction.

I think the corrected version is better. It's smaller, but it's true, and it tells you what to actually practise. That's this post.

Part 1Three things called the same name

The reason arguments about emotional intelligence go in circles is that the two sides are usually talking about different things. Researchers now sort EI measures into three streams, and it's worth getting the sort right before reading any claim.^4,5

Ability EI

Salovey and Mayer's original idea, updated in 2016: four skills, measured with tests that have right and wrong answers.⁶ Perceiving emotion (what is this face, this voice, this painting expressing?). Using emotion to help thinking (which mood helps with which task?). Understanding emotion (what does annoyance become if it's ignored? what's the difference between shame and guilt?). Managing emotion (given this situation, which action is most effective?). The standard test is the MSCEIT.

Trait EI

How you see your own emotional life, rated on a questionnaire: "I'm good at reading people," "I usually know why I feel what I feel." Honest and useful, but it measures self-perception, and it lands almost entirely inside the Big Five personality traits, especially emotional stability and extraversion.⁷

Mixed EI

Goleman's version and most corporate versions: self-rated competencies that bundle emotional skills with motivation, optimism, assertiveness, self-esteem and social confidence. Everything good that isn't IQ. The bigger the bundle, the more it predicts, and the less it means.

Here's the quickest way to feel the difference.

Once you can sort them, two things become obvious. Only the first kind is an intelligence in any strict sense, something you can be right or wrong about. And the third kind is where most of the big claims come from, because if you define emotional intelligence as "being motivated, confident, optimistic and nice," of course it predicts success. You've just renamed success and put it on the other side of the equation.

Part 2What it actually predicts

The best test of a construct is what it predicts after you account for things you already know. Here, for job performance, is what the meta-analyses found. The first set of bars are raw correlations. The second set asks the harder question: how much does each EI measure add once you already know someone's cognitive ability and personality?^4,5,8

Read that second view slowly, because it's the heart of the matter. Ability EI, the real, testable kind, adds about 0.2% to our ability to predict job performance once IQ and personality are known. Mixed EI adds a respectable 14%, but when researchers took it apart, the extra came mostly from self-efficacy, self-rated performance and conscientiousness smuggled in under a new name.^9,10 Goleman's "twice as important as IQ" is not supported anywhere in the data; the authors of the biggest meta-analysis say so in their paper.⁴

The picture is similar elsewhere. Across 158 studies and 42,000 students, EI predicts grades, modestly, and sits third behind intelligence and conscientiousness; the ability measures predict about as well as self-report once you adjust for everything.¹¹ EI correlates with well-being and with mental and physical health, more strongly for the trait and mixed versions, which is exactly what you'd expect if those versions are partly measuring emotional stability.^12,13

So is ability EI useless? No. It behaves like a real intelligence: it correlates with general intelligence at about 0.3, fits in the hierarchy of mental abilities as a "broad" factor alongside verbal and spatial ability, and predicts things the other measures miss, like the quality of someone's friendships as rated by their friends.^14,15 It also has one property the self-report versions can't have: people who are bad at it don't know. Self-rated EI correlates with tested EI at only around 0.2, and the people who most overrate their emotional skill are the ones with the least of it.^15,16 That's the same pattern you see with humour, driving and grammar. If you want to know how emotionally skilled you are, ask the people who live with you, not yourself.

Part 3The skills that are real

If the umbrella term is overblown, the ribs underneath it are not. Four of them have strong evidence, and each is something you can practise.

Reading emotion, without over-reading it

People recognise basic emotions in faces across cultures better than chance, and better within their own culture than across it.¹⁷ That's the perceiving branch, and it's trainable. But the biggest update in this field in a decade is a warning: faces are not readouts. A 2019 review of over a thousand studies found that people scowl when angry only about 30% of the time, and scowl for many other reasons, so inferring a feeling from a face alone is a guess, not a measurement.^18,19 Emotionally skilled people don't read faces harder. They read faces in context, and they hold the read lightly enough to update it.

Naming it precisely

People differ enormously in how finely they distinguish feelings. Some live with three words: fine, stressed, bad. Others can tell irritation from resentment from disappointment, and it turns out that matters. People with higher emotional granularity drink less when upset, are less aggressive when provoked, and recover faster from stress.²⁰ At the extreme low end is alexithymia, "no words for feeling," which affects roughly one in ten people and makes every kind of emotion regulation harder, because you can't manage what you can't name.^21,22

Naming does something physical. When people label the emotion in a face, amygdala activity drops and prefrontal activity rises, and the effect is bigger when the label is accurate.²³ Try it on whatever you're feeling right now.

Using feelings as information

The branch people find strangest is "using emotion to facilitate thought." It means treating feelings as data about the situation rather than as noise to be removed. Patients with damage to the parts of the prefrontal cortex that connect emotion to reasoning keep their IQ and lose the ability to make good decisions; without the gut, every option looks the same.²⁵ A feeling tells you something is at stake and roughly what kind of thing. The skill is in reading that signal without obeying it. I wrote a whole post on this: Feelings are data.

Managing: choosing what to do with a feeling

This is the branch with the most clinical research behind it, because emotion regulation is what most psychotherapy is. The core findings: habitually reappraising a situation (finding another true way to see it) goes with better mood and better relationships; habitually suppressing the expression of feeling goes with worse of both.²⁶ Rumination, suppression and avoidance are the strategies most consistently tied to depression and anxiety across disorders; acceptance and problem-solving are the ones tied to health.²⁷ Dialectical behaviour therapy was built on the observation that some people need these skills taught explicitly, step by step, because they were never taught.^28,29 The next post, The ninety seconds after, is entirely about this branch.

Part 4The dark side

One claim worth retiring is that emotional intelligence makes people good. Skill at reading and managing emotion is a tool, and in a 2011 study, people with high emotion-regulation knowledge who also scored high on Machiavellianism were more likely to behave badly toward colleagues, not less, because they were better at it.³⁰ Salespeople, manipulators and the best clinicians share a skill set. What differs is what they point it at. Compassion isn't a component of EI, and the research literature is honest about that even when the training brochures aren't.

Part 5Can you train it?

Yes, moderately, and more easily than the "intelligence" label suggests. Two meta-analyses of EI training programmes, 24 and 58 studies, both found gains of roughly half a standard deviation, held at follow-up in the studies that checked.^31,32 One well-designed trial gave adults 18 hours of emotional-competence training and found, six months later, less stress, better relationships, lower cortisol and even better employability ratings from blind evaluators.³³ In schools, universal social-emotional learning programmes improve behaviour, reduce distress and lift academic performance by about 11 percentile points.³⁴

Two caveats. Most of those gains are measured by self-report, the stream that's easiest to move by making people feel good about a course. And the effects are strongest for the specific skills, like naming feelings and choosing regulation strategies, not for the vague global trait. Train the ribs, not the umbrella.

Part 6Where mindfulness comes in

You'll notice I haven't yet said the word "meditation," which is unusual for this site. That's deliberate. Mindfulness is not emotional intelligence, and the popular move of equating them does both a disservice. But the overlap is real and specific. Meditation practice trains attention to the present moment, including the moment a feeling starts, before it has a story attached, and it trains the capacity to stay with that feeling without immediately acting on it. In the language of EI, that's perceiving and managing, learned from the inside.³⁷

The proposed mechanism is executive control: mindfulness strengthens the ability to notice emotion early and to hold it without being captured, which is the precondition for every regulation strategy.³⁵ Imaging studies in anxiety show reduced amygdala reactivity and better prefrontal recruitment after an eight-week course.³⁶ Here's how the five classic EI domains map onto the three skills I use throughout this site's meditation posts, concentration, clarity and equanimity, with the practice that trains each.

The honest summary: mindfulness builds the attentional platform that emotional skills run on. It does not teach the vocabulary (naming), the models (what shame does when ignored) or the moves (how to say the hard thing). Those you learn the way anyone learns a skill: explicitly, with feedback, with other people.

Part 7A week of practice

If you want to work on this, here's what the research suggests, one skill a day. None of it needs a course.

1. Monday: three words a day. Three times, stop and name what you're feeling with the most specific word you can find. Not "stressed." Use the wheel above if you're stuck. Write them down; you're building vocabulary.
2. Tuesday: find it in the body. Each time you name a feeling, find where it is. Jaw, chest, gut, hands. The body usually knows first, and clarity about body sensation is the fastest route to catching a feeling early.
3. Wednesday: read in context. Pick one conversation. Notice what you're inferring from someone's face or tone, then ask yourself what else could explain it. Hold the read as a hypothesis. If it's safe, check it: "You seem frustrated; am I reading that right?"
4. Thursday: ask what the feeling is for. When something strong shows up, ask what it's telling you is at stake. Anger says a boundary was crossed; anxiety says something matters and is uncertain; guilt says you acted against your own values. Then decide whether the message is accurate.
5. Friday: one reappraisal. Take one irritation and find a second true way to see it. Not a nicer story, a truer one. "He's late" might also be "he's drowning."
6. Saturday: the pause. Once, when you feel the pull to react, don't, for ninety seconds. Breathe, name it, feel it, then choose. The next post is the long version.
7. Sunday: ask someone. Ask one person who knows you well: "What do I do when I'm upset that I probably don't notice?" This is the only reliable measure of your emotional skill you have access to, and it's free.

Goleman was right that these things matter and wrong about how much. The honest version is that emotional skill is one real, modest, trainable ability among several, which puts it in excellent company with everything else worth learning.

This post is educational and isn't medical advice. If emotions feel unmanageable or unsafe, that's what clinicians are for; dialectical behaviour therapy and emotion-focused therapies teach these skills systematically. In the US you can call or text 988 any time.