Are human facial expressions universal in which emotions they express?

The latest issue of Science Advances has a provocative and clever piece of research aimed at answering a long-standing question—one considered by Charles Darwin himself: are the facial expressions associated with human emotions universal across all cultures? And, if so, is that the result of evolution? The paper suggests that, at least for a limited set of emotions, the answer is yes to both questions. You can read the paper by clicking the screenshot below, or reading the pdf here. The full reference is at the bottom.

Cowen and Keltner note that there have been a fair number of attempts to answer this question, all involving going to remote areas where there is little contact with Westerners, and seeing if people in those areas match photos of Western expressive signals (joy, anger, sympathy) with similar words in their native language. The results have been mixed. That, one would think, already suggests that perhaps facial expressions aren’t universal. (Cultural similarities could be due to either cross-cultural transmission of expressivity, or to a common evolutionary heritage, but dissimilarities among different populations already suggest that culture can’t cause similarities.)

However, the authors suggest these tests are not dispositive, and suggest a better way to see if expressions are universal. What they did was to look at ancient Mesoamerican artworks that depict facial expressions associated with different situations, isolated the facial expressions, got modern observers to guess the emotions depicted, and then asked whether Westerners would judge those expressions to be ones appropriate to the situation depicted by the artwork.  If there was a correlation between the judgements of which emotions were shown on faces, and those that other people guessed would appear in the situations depicted in the sculptures, this would suggest (but not prove) that expressions are not only universal, but perhaps universal because evolution made them universal.

The authors examined tens of thousands of Mesoamerican artifacts (going back to 1500 B.C.) archived by museums, and found 63 pieces of art that met their three criteria for this study:

  1. subjects were portrayed “within one of eight identifiable contexts” (torture, being held captive, carrying a heavy object, embracing another person, holding a baby, in fighting position, playing a ball sport, and playing music).  Examples of each of these are given in the figure below.
  2. The facial expressions could be clearly visible.
  3. The artworks (sculptures and vessels) were deemed authentic by authorities.

Here are examples of the eight contexts for emotion; see the caption for details:

(From paper): Fig. 1 Ancient American sculptures with discernible faces and contexts. (A) Captive from Tonina archeological site (Mexico, 690–700 CE). Photo credit: Mauricio Marat, Instituto Nacional de Antropología e Historia. (1 July 2019). (B) Tortured, scalped prisoner from Campeche (Mexico, 700–900 CE). Baltimore Museum of Art, Kerr Portfolio 2868, photo by J. Kerr. (C) Maya man carrying large stone (Mexico, 600–1200 CE). Kerr Portfolio 8237, photo by J. Kerr. (D) Joined couple (Mexico, 200–500 CE). Los Angeles County Museum of Art (LACMA) AC1996.146.21, gift of C. M. Fearing. (E) Maya woman holding child (600–800 CE). Princeton University Art Museum 2003-26, gift of G. G. Griffin. (F) Kneeling Maya warrior with facial tattoos and shield (Mexico, 600–800 CE), detail. Earthenware and pigment, 15.9 cm by 10.8 cm. Fine Arts Museums of San Francisco 2009.38.2, gift of G. Merriam and J. A. Merriam. (G) Maya ballplayer (Mexico, 700–900 CE). University of Maine HM646, William P. Palmer Collection. (H) Colima drummer (Mexico, 200 BCE–500 CE). LACMA, Proctor Stafford Collection, purchased with funds provided by Mr. and Mrs. Allan C. Balch.

They then isolated the faces from the contexts so that the modern subjects could just look at the faces but not see what the subject was experiencing in the artwork. The 325 subjects were asked them to describe the facial expressions using 30 categories of emotion (e.g., “awe”, “anger”, “shame”, “fear”, and so on and 13 “affective features” like “arousal” and “dominance” (see categories in figure below).

Independently, they asked 114 different subjects, also modern Westerners, to judge what emotions they would expect someone to show in each of the eight situations above. These people were not shown the artworks but simply given the eight situations verbally and asked what emotions you’d expect people to evince.  They then correlated the first subjects’ judgments on what emotions were portrayed by the faces with the expectation of what emotions should be shown from the eight situations, described in words, chosen for analysis.

Here are the correlations between the “read” of subjects from looking just at the faces with the emotions expected judged from subjects looking at the context described in words.

(from paper): Fig. 2 Accordance between emotions perceived in sculptures’ isolated face depictions and Western expectations for the emotions that unfold in eight portrayed contexts. To calculate the accordance between sculptures’ expressions and Westerners’ expectations, we correlated the participants’ average judgments of the emotions and affective features associated with each isolated face and each context across the eight contexts and divided by the maximum attainable correlation given sampling error (see Materials and Methods). Correlations are generally positive, indicating that facial muscle configurations portrayed in ancient American sculptures align, in terms of the emotions they communicate to Westerners, with Western participants’ expectations for the emotions that unfold in different contexts. Error bars represent SEs. Here, we excluded 10 emotions and 1 affective feature used seldom enough that <1/3 of the covariance in judgments was explainable, as a result of which SEs were very large.

Correlations above zero are positive; that is, the expectations of what you’d feel aligned with the subjects’ “read” of the artworks’ faces. As you see, of the 32 emotion categories assessed, 27 had correlations above zero, which alone says something (if there were no correlation, roughly half of the values would be above zero and half below). More important, only six of the 27 positive correlations had standard errors that overlapped zero, or “no correlation”; the rest were statistically significant.

This shows that, in general, the emotions expected by people hearing about the eight contexts for the artwork were those actually discerned by people who looked at the faces of the artwork. And that means that the emotions we expect to see in a modern situation were those that people read from artwork of ancient civilizations. This is turn suggests that at least some facial expressions are “universal”, and people can read them fairly accurately, even in sculptures, as reflecting what people are feeling in a situation. Because these cultures were far removed from ours, this also suggests that the facial expressions weren’t culturally inherited across the ages—that is, we don’t learn the expressions to show determination or distress or the other emotions from others, but express them innately.

One could, I suppose, argue that the Incas and Mayans simply passed on their facial expressions culturally to us as their descendants, but that’s a big stretch since the route for that kind of cultural transmission is dubious: we’re not the descendants of these Mesoamericans. Alternatively, the expressions could have been culturally inherited from our mutual ancestor who lived at least 60,000 years ago when H. sapiens left Africa to populate the world.  If you’re a blank-slater, that’s the argument you’d make.

But it seems more plausible to me that smiling and scowling and showing distress and pain are innate features of our behavior, though I can’t adduce hard evidence for it except that a). babies who haven’t been socialized to show facial expressions of joy or distress still show easily interpretable expressions, presumably before they’re socialized; b). other primates have readable facial expressions without our kind of culture; and c.) one can make a plausible evolutionary hypothesis that humans might have evolved to show expressions that other humans could read: “honest signals.” None of this adds up to a strong evolutionary-psychology argument, but one can agree, from the results above, that there is some universality across millennia in how humans show emotions on their faces.

Finally, the authors used a method called “principal preserved components analysis” to gather in one basket the main “dimensions” (combinations of emotions) that best explain the correlations shown above. From this they determined that three combinations of factors combine to explain at least five types of facial expressions on sculptures. Those expressions, the ones most easily seen as “universal, are these:

  1. “pain”, often in the context of torture
  2. “determination”/”strain”, in sculptures showing heavy lifting
  3. “anger”, usually in sculptures depicting combat
  4. “elation”, seen in “contexts of familial or social touch”, and
  5. “sadness”, as in being held captive after defeat.

The paper has other analyses, but these are the main conclusion. The authors also give several caveats, and you can have a look at those.

The tentative takeaway lesson is that, even after millennia, we can pretty accurately read the facial expressions of people from different cultures.  That is, there are some universals, at least between ourselves and Mesoamericans, in the expression of emotions. Whether that be the result of cultural inheritance or a universal code for expression that resides in our DNA (or a combination of these factors) can’t yet be discerned, but I’m betting on a largely biological-determinism explanation. And so are the authors, at least judging from their final paragraph:

The present results thus provide support for the universality of at least five kinds of facial expression: those associated with pain, anger, determination/strain, elation, and sadness. These findings support the notion that we are biologically prepared to express certain emotional states with particular behaviors, shedding light on the nature of our responses to experiences thought to bring meaning to our lives.

Here are a few more faces and the contexts in which they appear (see paper for explanation), but I think you can guess pretty well.

h/t: cesar

Cowen, A. S. and D. Keltner. 2020. Universal facial expressions uncovered in art of the ancient Americas: A computational approach. Science Advances 6:eabb1005.

24 thoughts on “Are human facial expressions universal in which emotions they express?

  1. Interesting. This is not surprising to me at all considering other primates share these kinds of behaviors with us.

  2. My facial expressions (especially that one-sided raised lip snarl of ‘don’t even think about it’ — aka the Dick Cheney face) have been understood perfectly well by dogs, horse, and hyraxes (maybe also by cats, but they’re not telling). So they’re at least a eutherian commonality.

  3. An interesting addition to the statement about babies’ facial expressions prior to their full socialization is that of people who were born blind, therefore unable to be socialized based on visual cues yet they display the same facial expressions as the sighted people do. Now, one might argue that they’ve been told what those expressions look like and are mimicking them but asking a young child who is blind to smile for the camera (as opposed to their natural, unplanned smile) and you get a weird open-mouthed grimace (personal experience on that one) rather than the typical faked for the camera smile.

  4. Pretty good experimental modeling. It depends though on someone selecting from among examples. That seems to be a potential bias. Also, the ability of every artist to produce accurate expressions is also a point of difficulty.

    Another approach would be to visit cultures around the world and photograph people in various emotional states. Someone being tortured would be somewhat challenging to find, but a good option would be forcing a few atheists to attend a Catholic Mass.

  5. All expressions could be grouped under one heading:
    “bowel movement”. This of course depends on the outcome😉

  6. I’m not a big fan of research conducted using subjects participating via Amazon Mechanical Turk. As I understand it, these people are pretty much self-selecting (though there is some basic screening for age, sex, etc.) and the more studies they participate in, the more they earn so there’s little incentive to take a great deal of time and effort especially as the amount per survey completed is very low. (Perhaps these deficiencies have been addressed since researchers started using this method of recruiting respondents via this platform, but if so I haven’t seen any evidence of this.)

    That aside, it’s an interesting paper. I couldn’t understand what “Identity” on the x-axis signified and couldn’t spot it mentioned when I read the paper (though admittedly, i only did so very quickly).

  7. Very interesting. I can easily see how “disgust” would be a result of evolution. Especially in regards to eating/drinking/smelling; it’s a very useful response to understand the myriad objects we encounter. There was that recent tweet about cats encountering durian fruit that is a great example of the universality of disgust.

  8. As others have noted, some universality among facial expressions seems extremely likely given that they show up in tiny babies, people with visual impairments, and other species of animals (I’m curious if there are any species that make facial expressions that don’t align with ours. There are certainly many animals that make no visual expressions, but I mean an entirely alternate set.)

    Anecdotally, I feel like watching my son has surprised me, in that so many things ‘just happen’, seemingly with no real teaching. I wouldn’t have considered myself a blank slater by any means but I do use behaviorism a good bit in working with those with special needs, so I didn’t realize the degree to which I assumed most things had to be taught. Seeing skills appear seemingly out of thin air was not really what I was expecting! I would have assumed that clapping was a ‘social construct’ kind of thing, for example, but shortly after my son started grabbing objects with both hands, it evolved into putting his hands together to clap. Same with pointing – he doesn’t even understand the concept of pointing at something yet, but he’ll raise his index finger and hold it up, then look at people with a huge smile. It seems like these are motor programs he was born with that are already flagged ‘important’ somewhere in his neurology, seeing as how he gets very excited about them. (Expressions, to my mind, seem even more strongly innate than that. When I first gave him sauerkraut he instantly made the classic ‘sour’ face, for example, and I can honestly say that’s an expression he’d probably never had the chance to observe other people making. And basic expressions like frowns and smiles were there pretty much from birth.)

    1. You experience with your boy what we experience with our cats! (And over the years sheep and pigs and cows, too.)

      Routines unfolding reliably.

      1. I know I sound like a gushy parent, but it just blows my mind how this stuff unfolds all by itself! One day they’re not doing something and the next – poof! – they are!

    2. Never experienced much of a baby other than babysitting and a couple cousins when much younger. I enjoyed reading your observations about your infant son. Thanks for sharing these intimate and interesting moments. I wish my parents were atheists. 🙂

      1. Ha ha, hope this is not too disappointing, but I am not technically an atheist, I am a “spiritual but not religious” type (although my husband is an atheist), so I don’t know what the heck I’m going to say to my son on the topic. I think what is most appealing about spirituality from a parental perspective is the idea of a locus of control / judgement that is neither ego-based nor peer-based nor ideology-based. If I came home from school upset because I was bullied, I took comfort in thinking “It doesn’t matter what other kids think, God loves everyone.” (I was raised a Christian.) I hope I can find a way to impart a similar concept to him so that he has a sense of worth that is not based on his own ego; the opinion of his peers; or some ideology.

  9. At least for “pain” and “exertion” it is not clear to me to what extent there are universal muscle movements that generate standard expressions. For example, I feel that by eyebrows are brought towards each other if I am lifting something.

  10. Hmm. The study and conclusions are pretty much pushing at an open door with me, but I do have a couple of cents worth.

    These people were not shown the artworks but simply given the eight situations verbally and asked what emotions you’d expect people to evince.

    Composing those descriptions of the artworks is a stage that is seriously open to injecting unconscious biases of language into it. Describing a scene of torture, for example, is pretty likely to put expectations of the subject being in pain into the contributor’s mind.
    Also, isn’t the infant (? toddler – hard to tell) in artwork E exhibiting the cranial deformity occasionally seen in mesoAmerican children as a consequence of their heads being strapped into a plank+cord vice during their growing years. “Coneheads” in the not-Dan-Ackroyd sense. It’s a relatively young piece though (600-800 CE), so no evidence of great antiquity of this practice.

  11. Even I, an instructor in the humanities–but a great fan of evolution–think it odd that the “social” explanation is always the default. But why “learn” that particular expression for an emotion?

    As the owner of four cat sibs (discovered behind the barn), hard genetic explanations seem commonsensical. You watch a cat raised in captivity “learn” such behaviors as scratching in litter, chasing after small furry things, chuckling at birds behind the windows, grooming themselves–all without a mommy or a daddy to “teach” them–and you realize just how ordinary evolutionary explanations for behavior are.

    1. The social explanation is the default for only one reason: because people are poised to attack you (on ideological grounds, really) if you make the evolutionary explanation the default one. Truly, though, for any trait that could be culturally passed on, one must consider cultural inheritance. But facial expression is unlikely to be culturally inherited for the reasons that I and some of the commenters note.

Comments are closed.