Our brains are exquisitely tuned to perceive, recognize and remember faces. We can easily find a friend's face among dozens or hundreds of unfamiliar faces in a busy street. We look at each other's facial expressions for signs of appreciation and disapproval, love and contempt. And even after we have corresponded or spoken on the phone with somebody for a long time, we are often relieved when we meet him or her in person and are able to put “a face to the name.”
The neurons responsible for our refined “face sense” lie in a brain region called the fusiform gyrus. Trauma or lesions to this brain area result in a rare neurological condition called prosopagnosia, or face blindness. Prosopagnostics fail to identify celebrities, close relatives and even themselves in the mirror. But even those of us with normal face-recognition skills are subject to many illusions and biases in face perception.
DR. ANGRY AND MR. CALM
Massachusetts Institute of Technology vision scientist Aude Oliva and University of Glasgow researcher Philippe G. Schyns created this illusion by producing hybrids of two images. The left picture shows Dr. Angry, and the picture on the right is Mr. Calm. But if you step away from this page, you will see that appearances can be deceiving. Nice Mr. Calm becomes Dr. Angry, and nasty Dr. Angry turns out to be a pretty decent fellow after all.
Fine details become blurred at a distance, leaving you with only the overall shapes and shadings of the images: what vision scientists refer to as the low-spatial-frequency content of an image. When you move closer, the images are once again dominated by their fine details, which are referred to as high spatial frequencies. The illusion works because the face on the left is composed of a high-spatial-frequency angry face combined with a calm face in low spatial frequencies. The right face is exactly the opposite: a low-spatial-frequency angry face with a high-spatial-frequency calm face. When the images are blurred (by stepping away), the different layers of the hybrid are revealed.
“MONA LISA” SMILE
Mona Lisa's captivating smile (left) is perhaps the most renowned art mystery of all time. Margaret Livingstone, a neurobiologist at Harvard Medical School, showed that Mona Lisa's smile appears and disappears owing to different visual processes used by the brain to perceive information in the center versus the periphery of our vision.
Look directly at Mona Lisa's lips and notice that her smile is very subtle, virtually absent. Now look at her eyes or at the part in her hair, while paying attention to her mouth. Her smile is now much wider. The movement of our eyes as we gaze around Mona Lisa's face makes her smile come alive, flickering on and off as our perception of it changes.
The center and periphery of the visual field have this differential effect on perception because the neurons at the center of our vision see a very small portion of the world, giving us high-resolution vision. Conversely, the neurons in the periphery see much larger pieces of the visual scene and thus have lower resolution.
This is what happens in the eye while viewing Mona Lisa: the eye focuses light that is reflected from the painting onto the retina, upside down and backward (above). Adjacent photoreceptors within the retina are activated by adjacent points of light reflected from the painting.
THE DA VINCI CODE OF PERCEPTION
Mona Lisa's smile can be explained by the fact that images are blurred in the periphery of our vision, so that her smile is only seen when blurred. Livingstone solved this mystery by simulating how the visual system sees Mona Lisa's smile in the far periphery, the near periphery, and the center of our gaze (above, left to right). The simulation was done in Adobe Photoshop by simply blurring and deblurring the painting to simulate the change in resolution from the center of vision to the far periphery. The smile appears on the left and center panels (far and near visual periphery) but is gone on the right panel (center of gaze). The effect is similar to the hybrid images of Dr. Angry and Mr. Calm and is likewise explained by the fact that different retinal neurons are tuned to different spatial frequencies. In a sense, Leonardo da Vinci painted the Mona Lisa as a hybrid, with a happy Mona Lisa superimposed on a sad one, each having different spatial-frequency content.
Surrealist painter Salvador Dalí also experimented with combining high- and low-spatial-frequency content in a single image (right). The title of the painting says it all: Gala Contemplating the Mediterranean Sea, which at Twenty Meters Becomes the Portrait of Abraham Lincoln (Homage to Rothko). The finer details of the painting, such as the edges of the colored blocks, blur when you view the painting from a distance or squint your eyes—and you can then see the low-spatial-frequency shapes and shading that make up Lincoln's face.
ILLUSION OF SEX
This illusion, created by psychologist Richard Russell, won third prize in the 2009 Best Illusion of the Year Contest. The side-by-side faces are perceived as female (left) and male (right). Yet both are versions of the same androgynous face (see http://illusioncontest.neuralcorrelate.com/2009/the-illusion-of-sex). The two images are identical, except that the contrast between the eyes and mouth and the rest of the face is higher for the face on the left than for the face on the right.
This illusion shows that contrast is an important cue for determining the sex of a face, with low-contrast faces appearing male and high-contrast faces appearing female. It may also explain why females in many cultures darken their eyes and mouths with cosmetics: a made-up face looks more feminine than a fresh face.
Cat Woman (right), created at Aude Oliva's M.I.T. laboratory, is a hybrid image of a woman (left) and a cat. At close range, Cat Woman has a cat's face. But at a distance, coarse features obscure the whiskers, fur texture and other details. Simply superimposing a transparent cat's face on a woman's face would not produce the same effect; this illusion works only by combining two images that differ in their spatial detail—one fine and one coarse.
MARGARET THATCHER ILLUSION
This illusion by vision scientist Peter Thompson of the University of York in England was critical to our understanding of face perception. When the illusion was discovered in 1980, scientists already knew that faces were difficult to recognize upside down. But the assumption was that because the brain always sees faces right side up, the face-recognition cells were optimized for right-side-up faces. This assumption was partially true, but the Margaret Thatcher illusion went further to show that the brain does not simply process and store representations of whole faces; rather it recognizes representations of individual facial features such as the mouth and eyes.
The top and bottom row of Thatcher images are identical to each other but flipped vertically. The top row looks like two upside-down Thatchers, no problem there. But the bottom row looks like a Thatcher on the left and a horrible mutant on the right. The reason is that whereas the left column depicts normal faces (although the upper face is upside down), the right column shows Frankenstein-ish composites of Thatcher with only the eyes and mouths flipped vertically. The Thatcher at the upper right does not freak you out, because the eyes and mouth are right side up (although the overall face is upside down), and your face-perception neurons therefore see them as “normal” (even though they do not match the rest of the face).
The bottom right image, on the contrary, is creepy because the eyes and mouth are upside down and thus all wrong, despite the fact that the face as a whole is right side up. Harvard neuroscientists Winrich Freiwald, Doris Tsao and Livingstone have now found neurons in the brain that respond to specific face features such as mouths and eyes, confirming the predictions that were made from this illusion several decades earlier.
TONY BLAIR ILLUSION
Vision scientist Stuart Anstis of the University of California, San Diego, created this illusion in 2005 to celebrate the 25th anniversary of the Thatcher illusion. Anstis reasoned that if face-detecting neurons prefer right-side-up facial features, they should also be selective for other evolutionarily stable aspects of faces. He tested this idea by comparing positive and negative images of Tony Blair. Because we have evolved to see faces only in positive contrast, it follows that the perception of individual facial features should fail if shown in negative. As with the Thatcher illusion, showing the whole face in negative (top left) makes it less recognizable than the normal face (bottom left). Using positive images of the mouth and eyes overlaid on a negative face does not look particularly grotesque either (top right). But a positive image of Blair with a negative mouth and eyes (bottom right) is just as horrid as the upside-down mouth and eyes in the right-side-up Thatcher.
Our nervous systems are hardwired to detect and process faces rapidly and efficiently, even with scarce details. Pictures such as the ones shown above are often referred to as Mooney faces, after cognitive psychologist Craig Mooney, who used similar images in his research on perception. Mooney faces illustrate how little visual information it actually takes to “see” a face.
The artist who created the movie poster for Premonition understood this phenomenon (opposite page, bottom). The tree branches, leaves and birds in the poster form only the barest outline of actress Sandra Bullock's face. Our brains fill in the gaps and construct a finished face from sparse visual content.
Our face-detection neural machinery can be overloaded. There is a man's face hidden in this image. But before we spill the beans about its location, look around and see if you can find it yourself. It's difficult! Don't give up too quickly: finding the face may take you a few minutes the first time you look. But once you have seen it, you will always find it immediately in every subsequent search. Given up? It's in the lower left quadrant near the bottom edge, about one third of the way across the image from the left.
HOLLOW MASK ILLUSION
This hollow mask created by sculptor Bryan Parkes gives the eerie impression that Albert Einstein's face is following you as you move around the room (below). The mask is placed in front of a window, with its open back facing toward you, so that sunlight illuminates the plastic face. Although the mask is concave, your brain assumes that all faces are convex. While a convex face would look in only one direction, Einstein's hollow face seems to look forward when the viewer is directly ahead, but at an angle when the viewer moves sideways. In another demonstration of this well-known illusion, when a hollow mask rotates on a turntable, it appears to turn opposite to the actual direction of the turntable.
Vision researcher Thomas Papathomas of Rutgers University created an interesting variation on this illusion by attaching three-dimensional eyeballs and a nose ring to a hollow mask. As shown in these three frames from a movie of the rotating mask, the eyeballs and nose ring appear to rotate in the opposite direction to that of the mask (above). This illusion won third prize in the 2008 Best Illusion of the Year Contest. You can view the movie at http://illusioncontest.neuralcorrelate.com/2008/rolling-eyes-on-a-hollow-mask.
SURROUNDED BY FACES
Because our brains are so good at detecting faces, we sometimes see them where they do not exist. Were you ever scared as a child by strange faces popping up from an abstract wallpaper design or formed by shadows in the semidarkness of your bedroom? Ever noticed that cars seem to have faces, with the headlights as eyes and the grilles as mouths? These effects result from the face-recognition circuits of our brains, which are constantly trying to find a face in the crowd. These circuits are so powerful that we see faces in an old telephone, a bowling ball, a roped-off room, a USB drive, a faucet and a log (from upper left).
THE MANE DIFFERENCE
Visual illusions showcasing politicians are all the rage. At first sight it looks like Al Gore standing behind Bill Clinton, but notice that Gore is really a doppelgänger Clinton, only with Gore's gorgeous head of hair (left). A set of face features (Clinton's) mixed with a different set of features (Gore's hair) isn't easily recognized as being misplaced.
Superman relies on the same illusion to protect his identity: thanks to a pair of glasses, a change of clothes and a different hairstyle, nobody in Metropolis realizes that he and Clark Kent are the same person (below).
Gaze at the angry face (left) for about 30 seconds while looking around the face from the eyes to the mouth, to the nose, back to the eyes, and so on. Then look at the center face. It looks scared, right? Now look at the scared face (right) for 30 seconds and then look at the center face again. This time it is angry! In reality, the center face is a 50–50 blend of an angry and a scared face.
Created by Andrea Butler and her colleagues at the University of British Columbia, this illusion shows that our visual-processing system adapts to an unchanging facial expression by temporarily becoming less responsive to it. As a result, the other facial expression dominates when you view the blend. This adaptation occurs in higher-level brain circuits, rather than in the retina, because the illusion works even if you view the left or right image with one eye only and then look at the center image with your other (unadapted) eye.
RACE FACE ILLUSION
While viewing composites of racially black (left) and white (right) faces that reflect exactly the same amount of light, psychologist Mahzarin R. Banaji of Harvard University noticed an interesting illusion: the white face appears lighter. Banaji and Daniel T. Levin of Vanderbilt University have proposed that the distortion occurs because abstract social expectations about skin tone influence our perception of faces.
FAT FACE THIN ILLUSION
Peter Thompson, who discovered the Thatcher illusion, has now identified a new illusion that he calls the “fat face thin illusion.” In this example of the illusion, the photographs are identical, but the upside-down face appears strikingly slimmer than the right-side-up version.
One possible explanation is that it is easier for the brain to recognize distinctive facial features, such as chubby cheeks, when they are viewed in the normal upright position. But that does not explain why thin faces don't look fatter—or thinner still—when viewed upside down.
FOCUS ON FACES
Facial expressions play a key role in our everyday social interactions. Even when watching movies or looking at photographs, we spend most of our time looking at the faces they portray. Our intense focus on faces is at the expense of other potentially interesting information, however. Take a quick look at this woman and child. Their smiling faces suggest they are having a good time. But is that it? Look more closely, and you may notice that the girl has an extra finger on her right hand: something that you probably missed at first because your attention was fixed on the faces.