You’re Sick, We Can See It All Over Your Face

By Carl Engelking | January 3, 2018 4:12 pm

A composite, layered image of 16 individuals (eight women) photographed twice in a cross-over design. These photos average the facial features of all 16 people into a single face. Can you tell which person is sick? A or B? (Answer at the bottom of this article. (Credit: Axelsson et al)

Humans seem to possess an uncanny ability to read sickness on others’ faces, even in the earliest stages of an infection.

No kidding, you might say. Who couldn’t pick out a poor soul who’s been in the throes of the flu, red nose and all? But our ability to detect sickness is far more sensitive than that, according to a study by John Axelsson and his team from the Karolinska Institutet in Stockholm, Sweden. Our face-assessing abilities are, perhaps, so sensitive, that we might even detect signs of sickness in another person’s face long before they know they’re sick.

It’s well known that humans immediately judge attractiveness, trustworthiness, dominance and basic emotions in a face; so, why not sickness?

In a simple experiment, Axelsson injected healthy volunteers with either a placebo or lipopolysaccharide (LPS), an endotoxin that kicks off an inflammatory stress response in the body—it simulates the physical response to being sick. Researchers snapped photographs of each person just two hours after receiving an injection and after blood tests revealed the inflammatory response had started.

Then, entirely new sets of observers were brought into the lab to scrutinize the images. In the first trial, observers simply indicated if they thought a person was sick or not. On a scale of 0 to 1, with .5 being pure chance, observers scored a .62, which means their ability to detect a sick person from a glance wasn’t random. In a second trial, observers homed in on the specific facial cues that indicated a person was sick.

Based on their responses, people who received the LPS injection appeared to have paler lips and skin, a more swollen face, mouths with drooped corners, hanging eyelids, redder eyes and less glossy skin. Of all of these indicators, pale lips were the most apparent sign of sickness noted by observers. Researchers published their study Wednesday in the journal Proceedings of the Royal Society B.

Subtle Cues

Evolutionarily speaking, detecting and avoiding sick individuals is a survival imperative. Animals with powerful olfactory abilities have the luxury of detecting the chemical signatures of illness in urine and feces. And while there’s some evidence that humans can smell a sick person, our visual diagnostic abilities might be far more refined than our noses’.

Let’s keep in mind, people were spotting sickness in others just hours after receiving an infection. This is an indication that the subtlest cues in a person’s face can alert us to sickness, before their nastiest, phlegmiest, wheeziest symptoms emerge. And this early detection ability is key, because people tend to be most contagious in the earliest days of their ailment, which could be days or weeks before they reach for the Kleenex. Therefore, our sensitivity to changes in the face may provide fair warning to stay away.

Now, to be clear, a success rate of .62 is better than chance, but it certainly didn’t floor researchers. That the predictive power of their ratings was low wasn’t surprising, as the observers were only exposed to photographs for few seconds, the researchers explain.

In reality, people may be even more receptive to sick-face cues in person, especially if they’ve seen a person before—in good health.

“There is a need to further test how accuracy can be improved, for example through learning, and whether identification is similar across diseases and ethnic groups,” researchers wrote in their study.

Moving forward, researchers would like to narrow down facial cues that are exclusive to sickness, and not, say, outward signs of fatigue or basic emotions like anger and fear. For example, there’s a fair amount of overlap in changes that occur to faces of tired people and sick people. Their research also begs further questions: Do certain diseases affect facial features in specific ways? Could health professionals better identify a sick person with facial-recognition practice?

The earlier contagion is detected, the easier it is to contain. So, let’s extrapolate even further: If researchers can identify all the nuance of a sick face, it’s easy to foresee a facial-recognition app that can provide a quick check-up via a cell phone.

If that theoretical technology were to ever arrive, kids can about forget sticking a thermometer on a light-bulb to feign a fever and play hooky. Moms will just pull out their iPhones, snap a photo and send their deceptive offspring off to school.

And, in case you’re playing, the sick person—who received the LPS injection—in the top image is (A). Were you correct?

CATEGORIZED UNDER: Health & Medicine, top posts
MORE ABOUT: personal health
  • Nitin

    I Think its A

    • Debsvoice

      I think it’s B… we have an answer?

  • Franz Egli

    As a child I grew up with two sisters who got sick several times just like me. I claim having learned to distinguish a healthy face from a sick one by training during my childhood and not by my genes. The authors don’t lose any word about a possible effect of training onto their findings. Moreover, the poor statistical signal of 0.62 (0.5 = completely random) doesn’t corroborate their interpretation of the data. I would expect a proper discussion which deserves the term “discussion”, meaning a discussion taking into account all possible alternative interpretations and thoroughly weighting them against the present one. It’s a pity that delivering the basis for a sensational story is obviously more important than doing a proper science job!

    • Norm Farris

      Totally agree, either the study and/or the article lacked rigor.

    • Marie Kho

      Having actually read the article, I am confused by your comment. Nowhere does it imply or state the ability is genetic.

      As for why 0.5 is considered random, it is because there are 2 choices 0.5 would be the average results for guessing. (This isn’t determined by author but generally accepted tgat if you guess and there are 2 choices, you have a 59/50 chance ir 0.5 chance of guessing correctly. It is math. Now the article states that though the results are above “random” guessing are “they are not flooring”. Mainly because it is only slightly above chance.

      The article is clearly a piece to entice interest in the possibilities of further research. Research needs funded, if people find the possibilities worth investing in, then you can get your discussion. This obviously wasn’t a in depth scientific study but a hypothetical question being asked and test designed to find out if this question merited more in depth and costly research. Studies like this are done all the time, to determine merit. This is obviously not the final research but the first.

      The article clearly stated that this was a first step and more questions will be asked moving forward to refine results. And the research attached cleary show this was very much a scientific research project and not a fluff piece.

      And if you clicked on the hyperlink to “their study” (the red letters) you might get some of the “proper science job” you state is lacking.

  • Thomas Buzzi

    May be some hidden bias in whoever posted the pics but it seems the one on the left is not as sharp as the one on the right. That and the droopy eyelids had me choosing A

    • John Thompson

      It’s not a real image.
      So right off the bat the study is flawed.
      Why not just use actual pictures of people who are sick or not?
      Why have a composite image?
      So what they were actually studying is if you could detect differences in two composite images – no word on what if any modification of the images happened.
      It’s difficult see any differences in the image – neither look sick to me.
      This is why people can’t really avoid sick people – early stages of infections often show few signs such as a red, runny nose.
      And things like coughing and sneezing are what really let us know someone is sick.
      If we had some ability to detect and avoid people with infections/illness early on, that seems to be contradicted by the history of disease and outbreaks of pandemics.
      Often when reading about things like the 1918 Spanish Flu Pandemic, people didn’t know who had it or who brought it into their towns. (People forget that many small towns here actually went into quarantine, but they didn’t notice mailmen being sick with the flu, and that’s how it spread into even rural towns that literally went into full blockade of outsiders. Seems like if they had the ability to detect illness early on, they would have blocked the infected mailmen from coming in. Ironically, they were often fumigating the mail itself.)

  • Esther Webb

    The palor of the skin and the deeper coloring under the eyes caused me to choose A. I never gave much thought to seeing the illness, but have felt that I knew before hand when a child would die while young. Twice this occured and one of them was from an accident.

    • Necromancer

      It is a gift you have. I’m an empath but i never learned to control it as a child.

      • Amaya Hiko

        Nothing to do with. A pale complexion and a slightly stressed face.

        It is an automated reaction to a slight exhaustion. That simply proofes – we are adapt at noticing low energy levels

  • Pedro_Schwartz

    I didn’t find it obvious, but I chose “A” based upon the just slight downturn of the corners of the mouth, which suggested a degree of discomfort.

  • Necromancer

    I chose “A” immediately

  • timbalionguy

    This might have a primal survival function in 1.) Identifying a family member who is not feeling well, so we can help. or 2.) Identifying weakness in a adversary, or a weak individual of another tribe/species who might be an easy meal.

  • Antonio Urbizu

    Yes, I was. I pick A and it was some Ques that naturally and without thinking I picked up. Also, it was 50/50 chance of guessing.

  • L Rowan McKnight

    I was correct. It’s subtle, but it’s all over his face.

  • k5sat

    I thought the B picture looked brighter and happier.
    I think at face value (pardon the pun) if these were two different folks most of us would have guess right 50% of the time.

  • DodgeMiniVan

    I chose Character A because there is a lack of pigment color in his lips and face also, the image appears rather sallow.



Briefing you on the must-know news and trending topics in science and technology today.

See More


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar