About Face: Emotions and Facial Expressions May Not Be Related

For half a century, one theory about the way we experience and express emotion has helped shape how we practice psychology, do police work, and even fight terrorism. But what if that theory is wrong?


Photographs by Jesse Burke

Forty-six years ago a young San Francisco–based cowboy of a psychologist named Paul Ekman emerged from the jungle with proof of a powerful idea. During the previous couple of years, he had set out trying to prove a theory popularized in the 19th century by Charles Darwin: that people of all ages and races, from all over the world, manifest emotions the same way. Ekman had traveled the globe with photographs that showed faces experiencing six basic emotions—happiness, sadness, fear, disgust, anger, and surprise. Everywhere he went, from Japan to Brazil to the remotest village of Papua New Guinea, he asked subjects to look at those faces and then to identify the emotions they saw on them. To do so, they had to pick from a set list of options presented to them by Ekman. The results were impressive. Everybody, it turned out, even preliterate Fore tribesmen in New Guinea who’d never seen a foreigner before in their lives, matched the same emotions to the same faces. Darwin, it seemed, had been right.

Ekman’s findings energized the previously marginal field of emotion science. Suddenly, researchers had an objective way to measure and compare human emotions—by reading the universal language of feeling written on the face. In the years that followed, Ekman would develop this idea, arguing that each emotion is like a reflex, with its own circuit in the brain and its own unique pattern of effects on the face and the body. He and his peers came to refer to it as the Basic Emotion model—and it had significant practical applications. In the late 1960s, for example, Ekman realized that he could detect the microexpressions of emotion that appear on the face of a liar. Anybody trained in how to properly recognize these microexpressions, he would later argue, could detect a liar 70 percent of the time. He published his first article on the subject in 1969, and three months later the CIA came knocking, eager to learn more.

So began a meteoric rise to fame. Since that first article, Ekman has consulted for not only the CIA but also the FBI, the Department of Homeland Security, the New York Police Department, and the Transportation Security Administration, which has spent more than a billion dollars training its airport agents in techniques based on Ekman’s theories. He’s published scores of influential papers and books, and his findings have been verified and expanded upon in hundreds of studies. In 2001 the American Psychological Association named him one of the most influential psychologists of the entire 20th century. And in 2009 Time named him one of the 100 most influential people in the world.

Ekman is a giant in his field, in other words. His ideas have powerfully shaped the science of emotion for half a century. But here’s the thing: What if he’s wrong?


“Honestly, this is going to sound terrible,” Lisa Barrett told me when I asked her about Ekman and his original study. “But at first, when I read that work, I thought, Well, nobody can take this seriously. This can’t possibly be right. It’s too cartoonish.” 

Barrett is a professor of psychology at Northeastern, and for years she’s been troubled by Ekman’s ideas. People don’t display and recognize emotions in universal ways, she believes, and emotions themselves don’t have their own places in the brain or their own patterns in the body. Instead, her research has led her to conclude that each of us constructs them in our own individual ways, from a diversity of sources: our internal sensations, our reactions to the environments we live in, our ever-evolving bodies of experience and learning, our cultures.

This may seem like nothing more than a semantic distinction. But it’s not. It’s a paradigm shift that has put Barrett on the front lines of one of the fiercest debates in the study of emotion today, because if Barrett is correct, we’ll need to rethink how we interpret mental illness, how we understand the mind and self, and even what psychology as a whole should become in the 21st century.

Twenty-one years ago, Barrett had no idea she’d be wading into this debate. In 1992 she was just another graduate student studying clinical psychology at the University of Waterloo—the MIT of Canada. She had every intention of becoming a therapist. True, she was unusually engrossed in the research side of her program. But the general type of study she was doing, exploring how people’s perceptions of themselves can lead to either anxiety or depression, was the perfect choice for a future shrink.

Times were hard for her back then. Her marriage was in tatters, her thesis adviser had just left town, she was in the middle of grueling comprehensive exams, and every time she attempted to run studies necessary for her research, they failed. In one particularly troublesome experiment, no one she tested could seem to distinguish anxiety from depression—even though differentiating the two was the entire point of the experiment. “If they reported feeling sad,” Barrett told me, “they also felt anxious. And if they reported feeling anxious, they also felt sad. And I thought, Well, can’t they tell the difference?” Every paper she read told her that they were two different emotional states of mind—one based in fear, the other in sadness.

It was a puzzle. Colleagues suggested that it was probably just normal statistical error and urged her to move on. But she couldn’t drop it. She’d already triple-checked her study design and crosschecked her subjects. What was left? She eventually decided it had to be the testing measures that she and others had been using. These, she realized, were actually useless when it came to assessing whether a person felt bad and worked up about it (anxious) or bad and lethargic about it (depressed). And that, in turn, called into question many of the supposedly successful studies that her work had been designed to replicate. Barrett wrote her work up, defended her Ph.D., did a clinical internship at the University of Manitoba, and then packed up for University Park, Pennsylvania, to start life as an assistant professor of psychology at Penn State.

  • Arvid Kappas

    I am very happy to read about this important issue. This is so well written, that I am sure, many readers will want to know more about this controversy – good starting points would be the work of Alan Fridlund UC Santa Barbara, who pointed out in 1986 that Darwin had never argued for a fixed link between emotions and expressions. An excellent introduction is his 1994 book “Human facial expression: An evolutionary view.”

    Jim Russell at Boston College has also worked for many years on these issues and has interesting high-profile research out there, questioning research methods and interpretations of the universality data. He also published a nice book on the topic in 1997 on the Psychology of Facial Expression with Jose-Miguel Fernandez-Dols,

    Lastly, here is a good pointer: http://www.apa.org/monitor/jan00/sc1.aspx

    • Defenestrator

      Thank you for pointing out a huge point that was mis-stated in the article — these ideas are definitely NOT new.

  • Richard Barone

    French philosopher Merleau-Ponty published Phenomenology of Perception in 1945 and “modern” psychology is just coming around to his ideas. He was right in saying that psychology has some serious limitations.

  • http://www.facebook.com/cees.timmerman Cees Timmerman

    “But in the other three piles, the Himba mixed up angry scowls, disgusted
    grimaces, and sad frowns. Without any suggestive context, of the kind
    that Ekman had originally provided, they simply didn’t recognize the
    differences that leap out so naturally to Westerners.”

    How much do the Himba watch angry, disgusted, and sad faces? Do they have TV? Did the researches ask them to demo those emotions?

    • uniquename72

      I did the experiment on myself with the latter half of the first set and the second set of pictures, and I also could not differentiate “properly” without the words being given to me. In fact, I was WAY off. And that’s after 40 years of TV watching.

      Any demonstration (or posed photo) is really just a caricature of the emotion, rather than the emotion itself. Most of our emotions are entirely internal and often changeable based on what we’re trying to convey and to whom — which is kinda the point.

  • clifflansley

    The latter part of this article suggests that our recognition/understanding of emotions has been taught. This ignores the work Ekman did in the stone age culture of Papua New Guinea with the isolated Fore Tribe. They both displayed and recognised the core emotions using the same universal triggers – loss of valued object/person = sadness; interference with goals/values = anger; etc.

    This research addresses the paragraph early in the article:

    “I can break that experiment really easily, just by removing the words. I can just show you a face and ask how this person feels. Or I can show you two faces, two scowling faces, and I can say, ‘Do these people feel the same thing?’ And agreement drops into the toilet.”

    I am interested in seeing good research that support this?

    • RebeccaSparks

      I don’t think it ignores as much as it challenges it. The problem with that Ekman (and other universalists) design close-ended tests with the goal of being able to test cross culturally, but with the close-ended test it ends up shaping the data to fit with the multiple choice answers. That result is what the test with the Himba is supposed to be “good research” that challenges it. Arvid Kappas lists some other sources, but I am not personally familiar with them.

      • clifflansley

        A key part of the work in PNG involved descriptions that correlated with the universal triggers (e.g. “Friends have come” for happiness) and the resulting expressions from genuine emotion matched the cross-cultural database …so not a multiple closed choice.
        Please can you point me to the Himba study? I am keen to learn what you are enthusiastic about and see howit counters what the scientific community has supported for so long.

        • RebeccaSparks

          If I’m reading the article we are commenting on correctly, Lisa Berett has finished the study, but is still in process trying to get it published.

          But if you’re looking for something already published, there’s plenty cited in this lit. review.

          Barrett, Lisa Feldman. “Was Darwin wrong about emotional expressions?.”Current Directions in Psychological Science 20.6 (2011): 400-406.

          These reviews do not make the bold claim that emotions are illusions. Instead, they make the more nuanced claim that emotion categories do not have firm boundaries in nature (i.e., emotions are not natural kinds). They demonstrate that behavioral, physiological, experiential and cognitive responses are highly variable within an emotion category, and this variability can be observed even in experiments explicitly designed to produce stereotypical emotional responses. Collectively, the empirical evidence points to the need to explain this observable variability in emotional responding while at the same time understand how human perceivers deal with that variability and experience or perceive discrete categories of emotion (Barrett, 2006b). Do the relatively few positive results come from methodologically superior experiments that float to the top in a sea of misguided empirical attempts? Or does highlighting those studies, while ignoring all the contrary evidence, constitute a case of confirmatory bias?”


    • kentclizbe


      Ekman’s “research” in New Guinea was flawed beyond belief.

      He did not speak the language of his subjects. His “translator” did not speak English well. The “stone age” subjects were familiar with white men, and were accustomed to “performing” for

      treats and rewards. The pictures used to “illustrate emotions” were completely arbitrarily chosen, by Ekman. The questioning technique involved massive suggestions and biasing context setting, and much, much more.

      An excellent article debunking Ekman’s original research was published soon after Ekman’s publication. It eviscerated Ekman’s “study.” By that time, Ekman, having quickly figured out the system, had leveraged his terrible study results for more government funding, and he was off to the races, science be damned!

      Here’s the most clinical evisceration of Ekman’s entire “theory” as well as his original “research” in New Guinea:

      See page 124 for the New Guinea critique:


  • Elias

    ” Barrett’s researchers would simply hand a jumbled pile of different
    expressions (happy, sad, fearful, angry, disgusted, and neutral) to
    their subjects, and would ask them to sort them into six piles.”

    They didn’t tell the subjects what they should have based the sorting on so we still don’t know WHY they placed disgust, anger and sadness in the same pile.

    • David McShane

      I wish everyone on this thread would go to Silvan Tomkins (Affect Imagery Consciousness) from whom Ekman got his ideas in the first place. To confuse affect with emotion is fatal to this research.

      • Elias

        So what is the difference between the two?

        • David McShane

          The best way to deal with this is to become familiar with Silvan Tomkins work. His conviction and demonstration that affect biases cognition predates by decades the recent confirmation of that fact by the brain imaging techniques of contemporary neuro-scienteists.

          The following may be useful. Goggle this:
          Eric Shouse · Respond to this Article. Volume 8; Issue 6; Dec. 2005. 1. AFFECT/A
          It depends upon those who knew (somewhat) Tomkins’ work but does not quote from his
          main treatment of the subject, in the four volume AFFECT IMAGERY CONSCIOUSNESS

          In capsule form one can say AFFECT IS BIOLOGY, EMOTION IS BIOGRAPHY.

          The following is quoted from the introduction to volume 1 of Tomkins’ AFFECT IMAGERY CONSCIOUSNESS (published in 1961)* written in the late 1950’s *. Given the fact that this is 50 years old gives it a quite prophetic cast. Tomkins died in 1991. He mentored both Ekman and Carroll Izard in the miulti-cultural affect recognition research.

          Following is from AFFECT IMAGERY CONSCIOUSNESS Vol 1 p.5-6.

          “It is not just consciousness in general which has been neglected, bu the role of affecdt has also been grossly underestimated. Indeed, we might speculate that the the phenomenon of consciousness might possibly never have been so neglected had the problem been restricted to what another human being thinks. It is rather knowing how he feels that has been most strikingly avoided. This is in part a consequence of the widespread taboos on affect which are learned in childhood.

          That Behaviorism slighted the role of affects is obvious, that Psychoanalysis did is less so. But if we trace the development of Freud’s theories chrdonologically, it becomes apparent that affects play a major role in his earlier papers and a successively smaller role as Psychoanalysis evolved. The affects were subordinated to the drives. As in most psychological theories, the drives were concieved to constitute the primary motivational system, and the affects played, but comparison, a lesser role in motivation. It is our contention that exactly the opposite is the case.

          In our view, the primary motivational system is the affective system, and the biological drives have motivational impact only which amplified by the affect system. Tis niew is unusual, despite the factt hat the evidence from a wide variety of sources clearly converges towards such a conclusion”.

        • Steve Shmurak

          Elias et al,

          To help clarify the difference between
          innate-biological affective signals and biographical emotion, I suggest you watch this short video (about
          12 minutes) showing the 9 innate affects in infants:

          http://www.aynrandstudies.com/jars/v8_n1/8_1toc.asp (On the second line you will find 2
          links – “PC” or “Mac”. If you click on the appropriate one you can download the
          video to your computer.)

    • Pat Field

      I agree with Elias’s concern: WHY did they place disgust, anger and sadness in the same pile?

      way of thinking about it is to imagine the varied ways people respond
      to shame: some crumple and sadly accept that they are faulty beings
      (“attack self”); some put much distance between themselves and the
      shaming situation (“withdrawal”); some act like clowns (“shameless”) to
      trivialize their participation, or engage in the use of drugs or alcohol
      to forget (“denial/avoidance”); some get hostile with actual physical
      attack or with the rejection of other implied by an expression of
      disgust (“attack other”).

      When a psychologist asks, “How are
      you feeling?” some people will respond with what they think is going on:
      “I feel like she doesn’t like me.” A good therapist might continue
      the inquiry with asking the client to locate the feeling in his body.
      “I feel sick to my stomach when she looks at me that way,” he might
      respond. Continuing to probe, the psychologist might discover whether
      there are other similar events in the person’s life, where that bodily
      sensation was elicited in a relationship, and if so, explore how that
      relationship came to be and how it unfolded. And the upshot might be
      that the sick-stomach feeling functions to alert the client that
      something is amiss in the relationship, some disconnect that leaves the
      client feeling unloved, unwanted, for reasons beyond his comprehension.

      It will take much time and interaction with the psychologist
      for the client to be able to connect the feeling of “someone doesn’t
      like me, and it makes me feel sick to my stomach” with what elements
      have gone awry in some significant early relationship, and then to
      connect that experience with a current relationship that has at least
      this one aspect in common with the early one. And then to connect that
      insight with his own way of participating in relationship with
      significant others, so that he might consider developing other ways to
      relate and/or to make better choices in whom to be in relationship
      with–when he’s ready to make those changes.

      Then there’s the
      ability to tolerate the sick-to-stomach feeling as one goes about trying
      out new relationships, and using that feeling as information that one
      needs to elicit feedback from the other(s) with whom one is in

      What does this have to do with facial display? So
      what if I look disgusted (only slightly) when you approach me? A lot
      depends on your ability to inquire into my feelings, doesn’t it? What
      action to take? You could crumple into sadness, or run away, or take a
      drink so as not to notice, or get mad at me–or inquire.

      unpleasant feelings and adopting a healthy Interest, rather than
      dissolving into a load of negative feeling–is where learning about
      Affect Script Psychology has taken me.

      To see a brief intro to Affect Script Psychology to, please check out:
      and download the two free papers at the top.

  • Matthew Quint

    This is a great piece, and a nice reminder that all research is subject to new understanding and improvement.

    There seem to me to be three things at play:

    1. Do facial responses represent an instinctual emotion? Maybe not in all cases, but no data could ever convince me that my 6 month old’s instantaneous facial expression of “yuck” when trying certain new foods is a manifested expression designed to communicate something to me rather than an instinctual internal emotional reaction to the taste of the food.

    2. Do all culture’s read facial expressions and corresponding emotions exactly the same way? I think it is pretty clear that exactly how an experiment is conducted can lead to different conclusions on this point. However, I think it is hard to argue that there isn’t some classification level — maybe only a two-to-four (may be less) facial-emotional states that wouldn’t be interpreted the same way.

    3. What’s really going on behind the facial expression-emotion connection? This is clearly the place where it seems clear to me that individual human experiences play a role. The “why” behind a facial expression and an emotion it might reflect are hugely influences by experience and context.

  • SixnaHalfFeet

    This all reminds me of Shrodingers Cat. Perhaps emotions are a quantum state that is not known until an observation is made? Or the act of observation effects the outcome?

  • Jason Barlow

    Great article – just found out his courses are coming to boston to learn these skills. http://people-intell.com/courses/course-schedule/11-emotional-skills-and-competencies-in-boston/event_details.html

  • Shane Anderson

    Isn’t the crux of Ekman’s work that emotion presents itself in universal ways, not that it is understood the same universally?… It feels like neither she nor the writer of the article have read Ekman’s work (they may have, but the article is heavily skewing stuff).
    It says it’s not a matter of semantics, but if we named a frown-y face as Orange, and smile as Banana and used fruits to label all emotions instead of Anger, Disgust etc. it feels like it would start to fall apart,
    I mean while someone may not be able to describe or interpret what they’re feeling as Happiness, a smile is ALWAYS a reaction to something positive for that person and while what is positive to one person is completely subjective for example a killer getting pleasure from someone else’s death, that their reaction to smile, I think, seems pretty standard.

    And in this day and age, for research to be dismissed seems like it says something about the research. This isn’t, after all, the 1800’s and she isn’t John Snow battling the whole word on miasma theory, it feels like there’s more than a little self-martyring going on on her behalf.

    • kentclizbe


      Sorry, but you’re not right.

      A smile is NOT always a reaction to something positive. In many Asian cultures, smiles indicate discomfort and embarrassment.

      And that’s just one example of the multitude of cultural differences that exist in facial expressions.

      Ekman’s seminal “research” was ignorant, uninformed, poorly designed, poorly implemented, and improperly analyzed.

      Ekman is the Bernie Madoff of deception detection.

      He’s scammed his way to a huge reputation, created a marketing juggernaut that keeps rolling, and simply brazens his way past anyone who questions him.

      Why, he’s Ekman, isn’t he? He wrote a book with the Dalai Lama, didn’t he? His work was the basis for a TV show, wasn’t it?

      Yes, but Bernie Madoff was that, and more.

      Kent Clizbe

  • kentclizbe

    For the ultimate in eviscerating Ekman’s horribly flawed theory, “research” approach, techniques, and continuing marketing of his “results,” see James A. Russel’s extensive critique of the New Guinea study, here:

    See page 124 for the New Guinea critique: