Other species have more finely tuned disease radars, relying primarily on the sense of smell. And previous research, Axelsson noted, has shown that animals can sniff sickness in other animals. (A Canadian hospital enlisted the help of an English springer spaniel trained to smell bacterial spores that infect patients.)
Yet while there is some evidence that an unhealthy person gives off odors that another individual can identify as sickness, the face is our primary source of “social information for communication,” Axelsson said.
He and his colleagues, a team that included neuroscientists and psychologists in Germany and Sweden, injected eight men and eight women with a molecule found in bacterial membranes. Like animals — from insects to mammals — people react very strongly to this substance, lipopolysaccharide. “People did not really become sick from the bacteria,” Axelsson said, but their bodies did not know the bacteria weren't actually attacking. Their immune systems kicked into action, complete with feelings of sickness. The subjects, all white, received about $430 for their trouble.
The scientists photographed the subjects two hours and 10 minutes after the injection, around the time participants said they felt the most unwell. They also photographed the subjects on a different date after they received a placebo injection of saline solution.
Axelsson and the team asked 60 students recruited from universities in Stockholm to assess the photos. The students looked at the portraits one at a time. A sick face and healthy face from the same individual never appeared consecutively. Observers were given five seconds to look at each photo and identify the person in it as sick or healthy.
The students' detection of immune responses was somewhat more accurate than sheer chance: On a scale of 0.5 to 1, 0.5 being completely random, the observers averaged a score of 0.62. The students had the most difficulty with three subjects, which is to say they had a tough time noticing any immune response. Still, they reliably judged the 13 others as sick, Axelsson said. (Of the 1,215 images identified as “sick,” nearly 800 were correct conclusions.)
Another group of 60 students was asked to rate pale lips, droopy mouths or other potentially telltale features in the portraits. “The change of skin color seemed to be the most robust” signal, Axelsson said. Heavy eyelids were also a strong cue related to illness.
“I am surprised,” said psychologist David Perrett, a researcher at the University of St Andrews in Scotland who was not involved with the research. No one had studied whether humans can sense “experimentally induced sickness” by looking at faces, he said, and “sickness judgments turn out to be far more reliable” than other visual judgments — for instance, gauging someone's personality from a neutral expression.
If we look at familiar faces, we are probably even more accurate at observing sickness because we know what color someone's complexion should be or what their usual expression is, the researchers noted.
“We will be much better as other cues come into play helping resolve any ambiguity,” Perrett said. “A sick person will likely become withdrawn from conversation and would hunch up.”
Looking for signs of illness on someone's face might seem like a scientific curiosity. But Axelsson sees an avenue to improving public health. If people can more quickly recognize the features that suggest a contagious disease, his thinking goes, a higher proportion of society can avoid infection. Waiting until someone is coughing or sneezing might be too late. “We are trying to tap into these first cues,” he said.
Axelsson and his colleagues are hunting for ways to improve the accuracy of detection. The psychologist said he hypothesized that observing disease was a learned behavior. The next step, he said, is to study whether doctors and other medical professionals are more accurate when rating sick faces than untrained students or the rest of us.