Web edition: September 21, 2012
Print edition: October 6, 2012; Vol.182 #7 (p. 20)
Let’s take a minute to turn faces upside down.
Pick any face. Ignore beards, glasses, hairdos or lack of any hair to do, and upend the facial features of Charles Darwin, Ray Charles or anyone named Charlotte who reads Science News.
People who normally remember or match a face perfectly well have trouble when it is standing on its head. But before there’s a chorus of “well, obviously,” let’s try turning dogs upside down, too.
Most people who don’t breed dogs or judge shows don’t recognize an individual dog nearly as well as a person’s face to begin with. And when pictures of poodles and Irish setters flip upside down in quizzes of learning and memory, people struggle a bit more than they do with the natural versions. But scores drop only modestly with these flipped-dog pics, compared with the dramatic drop for facial flips.
The disproportionate decline in remembering inverted faces has shown up in a variety of recall tests, with comparison groups from dogs to bridges, airplanes, stick figures, even clothing from 17th and 18th century paintings. Upside-down faces are where quiz scores really slump, and researchers view that slump as one of the signs that test-takers are actually experts at face perception.
A dog is a dog in any orientation. Same for other organisms and objects. But right-side-up faces apparently are so compelling that people have become especially masterful at recognizing the human visage. Know-it-at-a-glance holistic techniques behind this mastery fail when the world turns upside down.
The world of face-perception science didn’t exactly turn upside down last winter. But it developed a novel slant when two evolutionary biologists from the University of Michigan argued that a type of paper wasp, like people, shows signs of using specialized approaches to recognize faces of its own kind.
Paper wasps aren’t mammals, or even vertebrates. Before this study, the notion that a creature so distant from humankind in the tree of life could possess face expertise was weirder than an upside-down Darwin. Now the wasp development has added some sizzle to the endeavor of establishing what face-perception abilities other creatures may actually have. Emerging patterns in the animal world may reveal what drives the evolution of remarkable face prowess.
“The search is on,” says neuroscientist Winrich Freiwald of Rockefeller University in New York City.
While some researchers continue to invent tests (and debate how to interpret test results) for probing facial aptitudes among humankind’s primate cousins, other efforts have pushed beyond primates. Sheep, as well as those paper wasps, appear to have some special face skills. And faces may be important among rodents in ways that demand a more ticklish view of what face perception means. When it comes to face smarts, researchers are finding that the size of an animal’s brain may not matter as much as the company it keeps.
You’re good, really
Extreme skill at learning faces comes so naturally to many people that it can be hard to recognize as anything special.
But consider how many faces of family, friends, coworkers, classmates, teachers, neighbors, store clerks, actors and even long-dead historical figures on money, stamps and book frontispieces you can easily recognize. Learning to identify all those people by their hands would be a daunting challenge.
One argument that faces are something special for human perception points to newborns’ phase of facial fascination. In a classic test, babies averaging just 9 minutes old turned their blurry gazes farther to follow a schematic eyes-nose-mouth setup than to follow a disk with higgledy-piggledy features.
When babies grow up into children, at least three parts of the brain, and perhaps more, respond more strongly to human faces than to other objects, Freiwald says. Just how such skills develop as an infant ages, and how nurture might enhance nature, is still a matter of furious discussion.
Another nod to specialization comes from studies of people who suffer brain injuries that damage their powers of perception. Some people end up essentially blind to faces, a condition called prosopagnosia. “Not surprisingly, many prosopagnosics have impairments with both faces and objects,” says Bradley Duchaine of Dartmouth College.
But extensive testing has found people who struggle with just faces. A woman known in scientific papers as PS survived a blow to the back of her head after being hit by a bus. She recovered well enough to sail through tests of naming objects, but had to depend on haircuts, voices and other nonfacial cues to name children in the kindergarten class she taught, or even members of her own family.
Her difficulties and the accompanying stress emphasize just how important face perception is in society. A face serves as a badge of identity, a meter of welfare, a novella of reactions and intents: all important aids in negotiating the alliances and enmities of social living. So it’s easy to imagine how evolution may have favored skill for remembering faces.
Then there are the neuroscience tricks, such as the upside-down face test, that highlight a specialized system for face recognition by causing that system to fail. Take the bit of inverted spookiness called the Thatcher effect, honoring the former British prime minister whose portrait stars in a storied demonstration of the phenomenon. (Look to the image at right to see the effect on a more generic face before reading the spoiler ahead.)
At first glance, a test’s upside-down portrait comes across as no more than a flipped version of an ordinary person. Rotating the image right-side-up, however, creates a growing queasiness that turns B movie–gruesome as the brain can finally perceive the details properly. The eyes and mouth sit upside down relative to the rest of the face. Even partly right-side-up, a face that hadn’t been disturbing now looks nightmarish.
Messing with orientation isn’t the only way to tease the face-perception system. A person trying to identify the upper part of a composite face as, say, George Clooney’s works fast and accurately when the Clooney half appears misaligned from another celeb’s mouth and jaw. When the halves line up to form the familiar facial oval, perception falters; the face-specialized system apparently tries to identify the half-and-half face as if it were indeed one whole.
Many of humankind’s primate relatives live a complex social life too, and some of them also show interest and skill when it comes to faces.
Very early in life, chimps and at least one kind of monkey react to faces, although the babies are flexible about whom they gawk at. Rhesus macaques just 3 days old lip smack and stick out their tongues if a human adult does so in front of them, Pier Francesco Ferrari of the University of Parma in Italy and colleagues have found. In 2009 in Current Biology, Ferrari and colleagues also reported baby macaques gazing into their mothers’ faces and lip smacking when mom did. Yet the babies don’t make faces at a gesturing human hand or a spinning disk.
Macaques as old as 2 years still show some kind of interest in faces after a face-free start in life. Yoichi Sugita of the National Institute of Advanced Industrial Science and Technology in Tsukuba, Japan, worked with Japanese macaques raised for six months to two years without seeing faces. (Their human caregivers wore hoods.) When researchers finally presented pictures of faces, the macaques looked longer at the visages of other macaques, and of humans too, than at pictures of cars, alarm clocks and other unfamiliar objects.
Some other nonhuman primates show signs of face expertise during picture tests. Chimps appear to perceive faces better in a right-side-up orientation than in an upside-down portrait. Most studies testing for this inversion effect found it, Lisa Parr of the Yerkes National Primate Research Center in Atlanta reported in a 2011 review in Philosophical Transactions of the Royal Society B. In their version of the test, Parr and colleagues judged that chimps’ upside-down difficulties were greater with chimp faces than with pictures of species they had never encountered (capuchin monkeys) or objects with which they had little to no experience (cars, for example).
Parr and her colleagues also worked out a way to test chimps on half-and-half faces, in an echo of the human tests. When the upper part of a portrait of one chimp’s face sat weirdly fused atop the mouth and jaw of a different individual, about two-thirds of test chimps matched the mash-up to a picture of the top chimp instead of the bottom one. (Attending to eyes may be particularly important.) When Parr skewed the halves and broke the illusion of a new whole face, test chimps dropped closer to evenly dividing their matches between the whole faces contributing to the composite.
Not just primates
Sheep likewise react to faces in sophisticated ways. Yes, sheep. In the Oxford Handbook of Face Perception, published in 2011, neuroscientist Keith Kendrick and a colleague hailed sheep as dispelling a long-held notion linking specialized face perception only with primates.
Sheep may not have the brainy luster of the primate lineage, but they do lead social lives and they do attend to faces. During his years at the Babraham Institute in Cambridge, England, Kendrick and his colleagues trained sheep in such protocols as using pictures as clues to which arm of a Y-shaped enclosure holds a reward at its end. With such procedures, the researchers established that sheep can remember the faces of 50 other sheep for at least two years and can distinguish subtleties between an actual face and a computer-morphed image that is only slightly different. And that’s just a crumb of the case that Kendrick, now at the University of Electronic Science and Technology of China in Chengdu, has made for sheep as face experts.
In the classic test of distinguishing between upside-down pictures of faces of their own breed, sheep showed the diagnostic dip in performance that people and chimps do. Yet for pictures of buckets or portraits of an unfamiliar sheep breed, right-side-up versus upside-down didn’t make much of a difference.
Sheep even showed signs of distinguishing between facial expressions. When presented with a picture of a sheep face captured after a stressful bout of isolation or a picture taken during apparent calm, sheep tended to favor the contented sheep.
Paper wasps don’t have any facial expression — the hard outer skeleton of an insect’s body can’t move — but they do have faces. And the common paper wasp, like distant primates, appears to have specialized facial skills, Elizabeth Tibbetts of the University of Michigan and Michael Sheehan, now of the University of Arizona in Tucson, reported in Science last year.
Taking a close-up look (with respectful distance) at the faces of Polistes fuscatus reveals a riot of stripes, almost-eyebrows, quasi-mustaches, forehead swatches and cheeky splashes of yellow and brown. These colorful countenances reportedly stare back at the huge face of a human drawing near. Part of why Tibbetts loves working with the paper wasps, she says, is that they track her as she walks toward them in the lab and “they look at you in ways that other insects don’t.”
These paper wasps are also the only insects known so far to recognize familiar individuals of their own species by their faces, she says.
To explore the limits of a queen wasp’s face skills, Tibbetts and Sheehan pasted a picture at each crossbar end of a T-shaped wasp passage. Pictures showed wasp faces, geometric patterns or some nonfacial object that a wasp might hunt, such as caterpillars. One picture in each pair designated a refuge from an electric zap that researchers shot through the floor. (Unfortunately, Tibbetts says, the team couldn’t figure out how to train wasp queens by offering food rewards, as has been done with the sheep. Food is not particularly motivational since queens can go without it for a month.)
In the study, the paper wasps learned to distinguish between two faces faster than they learned about geometric patterns or other pictures. Jumbling the features slowed down the learning and destroyed any particular advantage wasps normally showed in face learning. Thus, wasps too must have their own version of a specialized face-recognition system, conclude Tibbetts and Sheehan.
Studies in a related wasp added a social dimension to the story. P. metricus queens flunked a test for recognizing individual faces of their own kind.
It makes sense that faces wouldn’t matter much in this species compared with P. fuscatus, Tibbetts says. Queens of the face-indifferent species usually found colonies on their own, so there’s not much pressure to recognize potentially mutinous royal housemates. In contrast, face-smart queens of the common paper wasp regularly cluster with other young queens to build a joint nest and lay eggs there. These clustering queens fight and establish a dominance hierarchy from one top queen on down. So face expertise should help individuals keep track of whom to fight and whom to cower from among scrappy, ambitious royalty.
Tibbetts is now working on understanding how face skills develop throughout life and how experience nurtures inborn capacities. Testing her ideas would be a problem in people; raising babies away from faces or with loving but masked mothers — don’t even think about it. Paper wasps and other facially deft but not so emotionally charged animals may offer insights.
Much of the study of faces relies on what they look like, which is not so strange a research bias for visually oriented humans. But that preoccupation may mean that science is missing whole other worlds of face perception, such as that of nocturnal, burrow-dwelling rodents, says Michael Brecht of Humboldt University in Berlin.
Rats live in a smelly, touchy world, and Brecht and his colleagues have been studying what he calls their “amazing” whisker-sensing abilities. Rats meeting whiskers-to-whiskers may perform an unappreciated exchange of facial expressions, he and his colleagues proposed late last year in Behavioral Neuroscience.
When rats meet, they sniff both ends (no suggestion yet of a specialized rear end perception system). But front-end encounters prevail after initial investigation. Rats align nose to nose, though trimming the whiskers can throw off the alignment. Then the rats sweep their whiskers forward and back. Among aggressive rats, the whiskers angle more forward and sweep faster than in apparently more serene rodents. Thus the whisker touch, which gives rats so much information about their world, may be detecting a whisker-position equivalent to sneers and stick-out-your-tongue taunts. Instead of just studying how animals look at each other’s faces, Brecht suggests a fuller picture of specialized facial skills would include facial touching, too.
Even focusing on just ho-hum sight, more animal groups remain to be tested. Patterns may point to the evolutionary history of facial expertise, says neuroscientist David Leopold of the National Institute of Mental Health in Bethesda, Md. People, chimps, sheep and paper wasps almost certainly did not inherit their powers full-blown from a mutual common ancestor, he notes. Yet ancestors of some or all might have passed down building blocks of those skills.
After reviewing what’s known so far about face perception in nonhuman animals, Leopold has sketched out some speculations that he hopes biologists will someday be able to test. Perhaps far back in time, he says, ancestors developed capacities for watching faces of possible predators. Though the ability to track the direction of another creature’s gaze doesn’t seem widespread, maybe ancient ancients started with some kind of interest in eyes. And then certain conditions may have nudged branches of descendants toward more elaborate attentions, not so much to predators but to the faces of their own kind.
At this stage it’s just the start of a story, but Leopold predicts that tales of animals that evolved in social groups among plenty of faces will make a good long chapter.
M. Brecht and W. A. Freiwald. The many facets of facial interactions in mammals. Current Opinion in Neurobiology, Vol. 22, April 2012, p.259. doi: 10.1016/j.conb.2011.12.003 [Go to]
A. Calder et al. (editors) Oxford Handbook of Face Perception. Oxford University Press, 2011.
P.F. Ferrari et al. Current Biology, Vol. 19, November 3, 2009, p. 1768. doi:10.1016/j.cub.2009.08.055. [Go to]
D.A. Leopold and G. Rhodes. A comparative view of face perception. Journal of Comparative Psychology, Vol. 124, August 2010, p. 233. doi: 10.1037/a0019460. [Go to]
M. Moscovitch, G. Winocur and M. Behrmann. What is special about face recognition? Nineteen experiments on a person with visual object agnosia and dyslexia but normal face recognition. Journal of Cognitive Neuroscience, Vol. 9, Fall 1997, p. 555. [Go to]
L. Parr. Evolution of face processing in primates. Philosophical Transactions of the Royal Society B. Vol. 366, June 12, 2011, p. 1764. doi: 10.1098/rsb.2010.0358. [Go to]
M.J. Sheehan and E.A. Tibbetts. Specialized face learning is associated with individual recognition in paper wasps. Science, Vol. 334, December 2, 2011, p. 1272. [Go to]
Y. Sugita. Face perception in monkeys reared with no exposure to faces. Proceedings of the National Academy of Sciences, Vol. 105, January 8, 2008, p. 394. doi: 10.1073/pnas.0706079105. [Go to]
A.J. Tate et al. Behavioual and neurophysiological evidence for face identity and face emotion processing in animals. Philosophical Transactions of the Royal Society, Vol. 381, December 2006, p. 2156. doi: 10.1098.rsb.206.937. [Go to]
B. Bower. Face memory peaks late, after age 30. Science News, Vol. 179, January 1, 2011, p. 16. Available online: [Go to]
B. Bower. Primates get a neural facial. Science News Online. December 18, 2008. [Go to]
B. Bower. Monkey moms and babies communicate from the start. Science News, Vol. 176, November 7, 2008, p. 15. Available online: [Go to]
B. Bower. Faces of perception. Science News. Vol. 160, July 7, 2001, p. 10. Available online: [Go to]
D.Y. Tsao and M.S. Livingstone. Mechanisms of face perception. Annual Review of Neuroscience. Vol. 31, July 2008, p. 411. doi. 10.1146/annurev.neuro.30.051605.094238
Test my Brain website: [Go to]
Teens take home science gold at Intel ISEF
One of the most abstract fields in math finds application in the 'real' world
Fine-tuning of technique used in other animals could enable personalized medicine
Simulation suggests long-term effect on sea level not as dire as some predictions
Coverage of the 2013 American Association for the Advancement of Science meeting
The Year in Science 2012
Three-part series on the scientific struggle to explain the conscious self
Tables of contents, columns and FAQs on SN Prime for iPad