Hard Feelings: Science’s Struggle to Define Emotions

While it's possible for researchers to study facial expressions, brain patterns, behavior, and more, each of these is only part of a more elusive whole.

When Paul Ekman was a grad student in the 1950s, psychologists were mostly ignoring emotions. Most psychology research at the time was focused on behaviorism—classical conditioning and the like. Silvan Tomkins was the one other person Ekman knew of who was studying emotions, and he’d done a little work on facial expressions that Ekman saw as extremely promising.

“To me it was obvious,” Ekman says. “There’s gold in those hills; I have to find a way to mine it.”

For his first cross-cultural studies in the 1960s, he traveled around the U.S., Chile, Argentina, and Brazil. In each location, he showed people photos of different facial expressions and asked them to match the images with six different emotions: happiness, sadness, anger, surprise, fear, and disgust. “There was very high agreement,” Ekman says. People tended to match smiling faces with “happiness,” furrow-browed, tight-lipped faces with “anger,” and so on.

But these responses could have been influenced by culture. The best way to test whether emotions were truly universal, he thought, would be to repeat his experiment in a totally remote society that hadn’t been exposed to Western media. So he planned a trip to Papua New Guinea, his confidence bolstered by films he’d seen of the island’s isolated cultures: “I never saw an expression I wasn’t familiar with in our culture,” he says.

Once there, he showed locals the same photos he’d shown his other research subjects. He gave them a choice between three photos and asked them to pick images that matched various stories (such as “this man’s child has just died”). Adult participants chose the expected emotion between 28 and 100 percent of the time, depending which photos they were choosing among. (The 28 percent was a bit of an outlier: That was when people had to choose between fear, surprise, and sadness. The next lowest rate was 48 percent.)

And so the six emotions used in Ekman’s studies came to be known as the “basic emotions” all humans recognize and experience. Some researchers now say there are fewer than six basic emotions, and some say there are more (Ekman himself has now scaled up to 21), but the idea remains the same: Emotions are biologically innate, universal to all humans, and displayed through facial expressions. Ekman, now a professor emeritus of psychology at the University of California, San Francisco, with his own company called The Paul Ekman Group, was named one of Time’s 100 most influential people in 2009, thanks to this work.

But despite the theory’s prominence, there are scientists who disagree, and the debate over the nature of emotion has been reinvigorated in recent years. While it would be easy to paint the argument as two-sided—pro-universality versus anti-universality, or Ekman’s cronies versus his critics—I found that everyone I spoke to for this article thinks about emotion a little differently.

“It’s been said that there are as many theories of emotions as there are emotion theorists,” says Joseph LeDoux, a professor of neuroscience and the director of the Emotional Brain Institute and the Nathan Kline Institute for Psychiatric Research at New York University.

The issue at the heart of this debating and theorizing is that it’s extremely difficult to pin down what people are debating and theorizing about. Because there is no clear definition of what an emotion is.

The word “emotion” did not exist in the English language until the early 17th century. It made the hop from France to Britain when British linguist John Florio translated philosopher Michel de Montaigne’s essays; Florio reportedly apologized for including the word, along with other “uncouth termes” from the French language. Uncouth, perhaps, because, as Thomas Dixon explains in his history of the word, it referred then to agitations, bodily movements, or commotions—there could be “public emotion,” for example.

The “happy” face from Paul Ekman’s research photo set (© Paul Ekman)

For many centuries, the sorts of mental states to which “emotions” now refer were typically called either passions or affections. The ancient Greek and Roman Stoics were notoriously anti-passion; they taught that man should use reason to battle all feelings, in order to avoid suffering. The Christian theologians Thomas Aquinas and Augustine of Hippo thought that was a bit much, so they carved out a separate category of good, virtuous feelings, which they called affections—things like familial love and compassion for others—and distinguished them from “evil” passions such as lust and rage.

Around the mid-18th century or so, Dixon writes, these passions and affections were lumped together under the umbrella of emotion. In the early 19th century, Scottish philosopher Thomas Brown was the first to propose emotion as a theoretical category, opening the door for scientific research. But though he was eager to study it, Brown couldn’t define it.

“The exact meaning of the term emotion, it is difficult to state in any form of words,” Brown said in a lecture. And so it has remained.

“The only thing certain in the emotion field is that no one agrees on how to define emotion,” Alan Fridlund, an associate professor of psychological and brain sciences at the University of California, Santa Barbara, wrote to me in an email. Many modern articles on the topic start off by referencing “What Is An Emotion?”, an 1884 article by the influential psychologist William James, and go on to bemoan that science has still not answered that question. If a researcher does propose a working definition in a study, it’s unlikely that anyone but the author will use it or agree with it. The author might be categorizing emotions based on behaviors, physiological responses, feelings, thoughts, or any combination thereof.

“Semantics have to do with pointing,” says James Russell, a professor of psychology at Boston College. “By ‘emotions,’ we mean ‘those things.’”

In everyday life, the lack of a formalized definition of emotion (or any of the more specific terms that stem from it—happiness, anger, sadness, etc.) may not matter so much. It’s not as though if someone tells you she’s angry, you have no idea what she means. There’s some level of understanding there. But ask people to explain in words what an emotion is (“explain it to a robot who’s just become sentient,” is how I like to put it), and they’ll quickly become stumped.

I asked a few of my coworkers to try and got responses like “individual-specific reactions to experiences,” “sensitivity to events,” “your mind’s reaction to experience,” and, poetically, “the description of intangible human feelings, the powerful internal sensations that color our every experience.”

These definitions are all pretty good. They all feel right. But fundamentally, as that last person said, emotions are intangible. They are definitely something. They’re not nothing. And that may be good enough for life, but it’s not good enough for science.

“Psychology is really experimental philosophy,” says Lisa Feldman Barrett, a university distinguished professor of psychology at Northeastern University, and the author of the forthcoming book How Emotions Are Made. Biology, for example, is a discipline that relies solely on observations of the natural world, while psychology researchers “take common sense categories that people use in everyday life and try to treat them like scientific categories.”

“Surprise,” an emotion Ekman’s subjects often confused with “fear” or “sadness” (© Paul Ekman)

Barrett has emerged in recent years as a new voice in the field of emotion, with a unique perspective on how to think about the phenomenon. In her 2006 article “Are Emotions Natural Kinds?”, she threw down the gauntlet, positioning herself strongly against Ekman’s viewpoint that emotions are biologically basic. (The term “natural kind” refers to a group of items that are inherently equivalent.) “The natural-kind view has outlived its scientific value,” Barrett wrote, “and now presents a major obstacle to understanding what emotions are and how they work.”

According to Ekman, the evidence for universality is “extremely strong and robust, statistically.” In a meta-analysis of similar photo-matching experiments, people across cultures were able to correctly categorize emotion expressions an average of 58 percent of the time—higher for some emotions, lower for others. That is significantly greater than chance. The question is, is it enough?

Barrett says no. She doesn’t think expression categorization shows that emotions are biologically basic, and she’s not convinced these specific expressions appear every time someone feels the corresponding emotion. She points out, for instance, the subtlety and range of actors’ emotional expressions. “When was the last time you saw an actor win an Academy Award for scowling?” she asks.

She acknowledges, in her 2006 article, that “meta-analytic and narrative reviews clearly indicate that perceivers from different cultures agree better than chance on the best label to assign to posed, static, facial configurations … But above-chance accuracy is only part of the picture.”

The rest of the picture is interpretation. Either 58 percent is good enough for you, or it isn’t. If something truly universal and innate is going on, why can’t we do better than just “above-chance”?

Human error, some might say. Just because an emotion is expressed on a face doesn’t mean the person looking at the face can read it accurately. Or maybe the same expression can be read different ways by different people. Barrett suggests that priming people with stories like “this man’s child has died” might lead them to categorize a pouting face as sadness, when they might label it something else without the context.

Russell, who has also been a prominent critic of the natural-kind view of emotions, has a similar complaint. “Forcing the observer to choose exactly one option treats the set of options as mutually exclusive, which they are not. Subjects place the same facial expression … into more than one emotion category.”

As Barrett sees it, emotions are totally made up. Not that they aren’t meaningful—it’s just that words like “joy,” “shame,” and “rage” describe a whole host of complex processes in the brain and the body that aren’t necessarily related. We’ve just lumped some of these things together, and named them. She compares the concept of emotion to the concept of money.

“The only thing that holds that category together is that humans agree,” she says. “Currency exists because we all agree something can be traded for material goods. Because we agree, it has value. One of the remarkable things humans can do that no other animal can do is that we can make stuff up and make it real. We can create reality.”

One common critique of the labeling-photos approach is that the expressions in the pictures are posed. A study done in the 1980s found that when people were shown photographs of candid, spontaneous emotions, the rate of recognition went down from more than 80 percent with posed pictures to just 26 percent.

“Fear” (© Paul Ekman)

It is true, that in daily life, you probably won’t see an Edvard Munch The Scream face every time someone feels afraid. The extreme, exaggerated version of an emotional facial expression might only appear in extreme situations—when a loved one has died, or when someone is in mortal danger. For subtler emotions, Ekman’s theory goes, the corresponding expressions are subtler as well.

And people can also actively suppress their more dramatic facial expressions if they don’t want people to know what they’re feeling. What Ekman calls microexpressions are the small, quick facial movements that sometimes leak out anyway, even when someone’s trying to keep a lid on it.

To support his theory of microexpressions, Ekman has done research measuring the movement of facial muscles while eliciting emotions. (This led to Ekman’s Facial Action Coding System, a guide to facial muscle movement used by scientists and artists alike—including Pixar, he says.) The smaller movements are harder to see, which may explain why the candid expressions in that study from the 1980s were harder for subjects to recognize—subtle emotions are often the most researchers can evoke in a laboratory. Dacher Keltner, a former student of Ekman’s and a professor of psychology at the University of California, Berkeley, puts it this way: “You have to take a step back and remind yourself what these scientists are really studying. Most often we study people’s responses to film clips. I love movies, but little two-minute film clips are lower on the scale of powerful elicitors of emotion.”

Even if a split-second brow furrow is there, that doesn’t mean someone will notice it, or even read it as anger. Most people won’t, Ekman says. This is why he created tools that he now sells on his website, which claim to teach the user how to recognize these microexpressions and thus better read what emotions other people are feeling. This would be quite a power to have.

“These [tools] have been used by a variety of organizations—all of the three-letter intelligence and law enforcement agencies on a national level,” Ekman says. “My research has not been limited to the labeling of still photos.” He complains that takedowns of his work ignore this component altogether. “My critics pretend [the measurement research] wasn’t published, but it was published and it was a lot of work.”

This research is in fact the basis for the TV crime drama Lie to Me, which features a researcher who helps law enforcement by detecting deception through facial expressions and body language. “I reviewed every script,” Ekman says, “and gave them feedback, which sometimes they took and sometimes they didn’t.”

But for the most part, it’s Ekman’s fundamental idea—that emotions are the same for all humans across cultures—that tends to provoke the most criticism. Decades before either Barrett or Russell criticized his model, he was catching flak from the famous anthropologist Margaret Mead, who believed emotions were a product of culture. “[Mead] treated me rather shoddily,” Ekman says. In a 1975 issue of The Journal of Communication, Mead wrote a disparaging review of Ekman’s book Darwin and Facial Expression, calling it “an example of the appalling state of the human sciences.”

“I never found out whether she was making a pun on my first name,” Ekman says, referring to the “Paul” in “appalling.”

But emotions don’t exist in a vacuum, and for some researchers, context is everything. (Though, for what it’s worth, Ekman does concede that the basic toolkit of emotions all humans share can be influenced by experience.) “When people across cultures have the words for anger, that doesn’t mean that anger means the same thing, that it evolves in the same way, that the same situations are thought to be anger, that how anger functions in a relationship is similar,” says Batja Gomes de Mesquita, director of the Center for Social and Cultural Psychology at the University of Leuven in Belgium.

When Mesquita considers Ekman’s photos, she says, “it’s not clear to me that what these faces express is emotion. But it’s undeniably the case that what they express is relevant to emotions. I think a lot of the problems are not so much in the data, but in the inferences from those data.”

If not facial expressions, then what’s the best way to measure emotions? A 2007 paper on which Barrett and Mesquita were co-authors called for “a focus on the heterogeneity of emotional life.” The authors asserted that “language use, context, culture, or individual differences in prior experience will produce variation in whether emotions are experienced, which emotions are experienced, and how they are experienced.” There are a number of methodologies researchers can use to capture this heterogeneity, from brain imaging to measuring physiological responses, but learning what someone actually feels, Barrett says, is hard to do with anything other than self-report—asking people to describe how they’re feeling or answer questionnaires.

“Anger” (© Paul Ekman)

“The gold standard is self-report,” says Maria Gendron, a postdoctoral research fellow in Barrett’s lab at Northeastern. “Because it doesn’t make assumptions.”

Of course, this methodology is up for debate as well. “The memory for emotional experience is highly unreliable,” Ekman says. “If [self-report] is the method that’s used, I won’t read the article.”

One problem, as many scientists pointed out to me, is that language—particularly the language of emotion—is inconsistent. “If someone says, ‘I’m really anxious to see you,’ what they’re really saying is, ‘I’m eager to see you,’” Ekman says. “If they’re anxious about seeing you, that means they’re highly disturbed mentally at the prospect of seeing you. The layman uses these words very sloppily.”

On the biology side, some researchers are trying to identify structures and systems in the brain where emotions come from. One scientist, Jaak Panksepp, a professor of neuroscience at Washington State University, has identified seven circuits of neurons that he says correspond with seven basic emotions. Panksepp’s work is congruous with Ekman’s on the universality issue, but he actually takes it even further—he works with animals, and says there’s something about emotions that’s biologically basic not just to humans but to all mammals.

LeDoux, the NYU neuroscientist, is somewhere in the middle. He thinks responses to stimuli are hardwired into the brain, which lines up with Ekman and Panksepp. But like Barrett, he thinks that the conscious brain and the analysis that goes on there are necessary for the experience of emotion. By this logic, since we can’t know what animals are experiencing, there’s no way to know if animals have emotions.

He emphasizes the role human consciousness plays in studying things like emotion. (What consciousness is, and how it works, is a whole other contentious question.) “In physics, it doesn’t matter whether people believe the sun rises or not,” he says. “That has no impact on the movements of the planets and stars. Whereas in psychology, people’s ideas about how the mind works influence the subject matter. Our folk psychology, in other words, can’t be divorced from the science.”

Consider the amygdalae, the two little oblong nuggets, one on each side of the brain, that are widely considered to be the seat of fear. A recent episode of the NPR show Invisibilia featured a woman who suffers from a rare disorder that left her amygdalae calcified. The patient, who goes by the initials S.M., does not report experiencing fear, a fact that would seem to solidify the connection between anatomy and emotions. But in 2013, researchers were able to trigger a fear response in S.M. and other patients with amygdala damage by having them inhale carbon dioxide. This makes the body feel like it’s suffocating, and the so-called “fearless” patients panicked, much as anyone would.

“Everybody had a headline about this—‘Fearless Woman Feels Fear,’” LeDoux says. “The only reason you’d be surprised by that is if you think fear comes from the amygdala.”

LeDoux defines fear as what happens in the conscious brain in reaction to the  response to danger from the brain’s survival circuit. If that’s the case, then a person’s experience of fear comes not from the amygdala itself, but from the brain structures responsible for cognition.

“Feeling afraid only occurs in organisms that can be conscious that they are in danger,” he wrote in a paper published in January. When I spoke to him, he added, “If we tell people the amygdala is directly responsible for fear, we’re giving the wrong message.”

For his part, even Ekman would no longer say that facial expressions alone equal emotion. “Thirty years ago, I was emphasizing facial expression, and I might have said to you: ‘Expressions are emotion,’” he says. “[But] it’s not a single phenomenon. It’s a group of organized phenomena. Some theorists have emphasized one.”

Ekman now considers physiology, appraisal, subjective experience, and antecedent events (you have an emotion about something) to be distinctive characteristics of emotion, along with facial expression and a few other factors.

“Sadness” (© Paul Ekman)

Still, “at the heart of ‘emotion’ is the experience of emotion, and this can’t be measured,” Fridlund writes. Recorded, maybe, but not measured. “This leaves scientists studying ‘emotion’ trying instead to measure everything around it.”

Russell makes a similar point. He thinks that emotions are best studied by measuring their components—facial expressions and nervous system activation, as well as behavior and internal feelings. But he says it’s going too far to add all these things together and call the result “emotions.” “We pull out certain clusters of those and name them,” he says. “When your physiology is high, you’re in danger, and your face goes into a gasp, you say, ‘Oh, that’s fear.’ I think as scientists we’re not going to do well defining clusters. They’re too vague.” Better, he says, to just ask: “Under what conditions do the facial muscles contract in a certain way?” rather than saying that contraction signals an emotion.

Even if there’s no consensus on what emotions are, there’s at least some overlap in what scientists think they involve. In 2010, Carroll Izard, who, along with Ekman, contributed greatly to the universal basic emotions theory, surveyed 34 emotion researchers on their definitions of emotion. While “no succinct synthesis could capture everything in the 34 definitions of ‘emotion’ given by the participating scientists,” he writes, here is the description Izard came up with, based on the things that had the highest agreement:

Emotion consists of neural circuits (that are at least partially dedicated), response systems, and a feeling state/process that motivates and organizes cognition and action. Emotion also provides information to the person experiencing it, and may include antecedent cognitive appraisals and ongoing cognition including an interpretation of its feeling state, expressions or social-communicative signals, and may motivate approach or avoidant behavior, exercise control/regulation of responses, and be social or relational in nature.

Izard then goes on to say that “the foregoing noteworthy and highly pluralistic description of the structures and functions of emotion is not a definition.” The scientists agreed more on what emotion does than what it is. (It seems from my research that there’s some disagreement, too, on what an emotion isn’t. States like “hungry” or “sleepy” are usually excluded, but while one researcher might call “love” an emotion, for example, another might say it isn’t a brief enough feeling to qualify.)

It’s strange that in a field as uncertain as emotion research, there is so much contention. I have rarely heard scientists get so sassy talking about their research. Ekman accused some of his critics of having careerist motives. “If you challenge someone who’s well-established, that can get you press coverage. [Barrett]’s done press releases, that’s what gets her coverage.” When I ask what’s wrong with press releases, he says, “I don’t do it. I’ve never done it … There’s other ways to get recognition for your science.”

Panksepp says he feels he’s often dragged into debates even though he sees his work on the “primary level” of the brain as a foundation for researchers like LeDoux and Barrett, who emphasize cognition, to build on. “I see myself as providing great assistance to other [scientists] if they desire such assistance,” he says, adding, “People are always competing. That’s the way it’s always been and will always be.”

Keltner, the Berkeley psychologist, says, “I think we’re always going to battle over what the broad construct of emotion is. There’s something about emotion that produces these disputes. It may be that we think we’re getting down to the essence of human nature.”

When there isn’t an agreed-upon definition for what researchers are seeking, science can look like a kind of religion. People commit to different paths to look for the same thing. Some become certain that their path is the right one. Others are agnostic—certain only that things are uncertain. Still others are content to ignore the unanswerable questions and focus on analyzing things that don’t resist analysis. Data is data, true enough, but individuals can interpret it however they please.

Ekman himself models “contempt” (© Paul Ekman)

Russell likens the naming of emotions to a sort of psychological astrology. “Lots of cultures have recognized constellations, named them, and made up stories about them. People who believe in astrology still think they influence people. But in astronomy, those stars don’t have any particular relation to each other.”

Fridlund sees the emotion field as a kind of Rorschach blot “on which psychology is pretext, but ideology is subtext.” He describes the Mead/Ekman feud, for example, as primarily one of ideology. He thinks Ekman’s universality theory was an attempt to bring psychology back from Mead’s idea of cultural diversity to a “feel-good Kumbaya message” where “we’re deep-down all the same.”

Science is not always a set of answers to questions, a collection of hard-won facts about how the world works. Sometimes the scientific method spans decades, centuries even, every study a drop in a bucket that might never be filled. It’s hard to know how close emotion researchers are to a solution, or if there even is one. “Philosophically, it’s arguable that ‘experience’ is not anything intrinsically measurable,” Fridlund writes. “This may make it forever off-limits to science.”

It would be kind of nice to think that in this age of answers, there might be a forever question. Not about God or the meaning of life, but just about humans and how we work. Maybe emotions are just the collection of physiology, behavior, and situational context, nothing more. But maybe there’s something more to them than just that—a deeper meaning that emerges from the constellations we create, something transformative and, ultimately, unknowable.