|
Form – Face Perception
All faces are composed of more-or-less the same basic elements (two eyes, a nose, a mouth, eyebrows) set up in roughly the same spatial configuration. Different faces are distinguished by subtle differences in the structure of these elements and their spatial layout. Most people possess an incredible capacity for recognizing a familiar face the instant we lay eyes upon it. Further, we are capable of gleaning important information about the individual’s emotional state, age, sex, and gaze from simply looking at them. Somehow, subtle differences in facial features and their configuration encode all of this information. The exact mechanisms underlying the processes by which our brain extracts this information have not been fully established, although scientists have recently made impressive advances.
Fusiform face area
A cortical brain region containing cells that respond overwhelmingly to faces has been identified—the fusiform face area or FFA. The cells in this region do not respond exclusively to faces, but the exceptions are typically face-like objects, such as a round clock, in which case it is believed that enough of the visual elements match the encoded neural face ‘template’ to activate the face-tuned cells.
Researchers have demonstrated that the spatial relationship of facial features, such as the distance between one’s eyes, is important for face recognition. One useful model for thinking about how these cells might be tuned is this: imagine that two cells exist who care about how far apart a person’s eyes are, and each has a slightly different sensitivity curve to the magnitude of eye displacement—put ‘eye displacement’ along the x-axis and intensity of cell response along the y (measured by firing rate), one cell’s sensitivity curve would peak at the left extreme and the other at the right, and the two would overlap in the middle, which would probably represent the average in the population the individual experienced growing up. Integrating the resulting responses would give the actual distance between a person’s eyes.
It is likely that other higher level features are encoded in this way. Caricatures provide some support for this hypothesis. A caricature is an image in which the facial features of an individual are exaggerated. If the neural basis for evaluating other aspects of facial features is encoded as described above, then extremes should activate the cells better than cases that don’t deviate as far from the norm. And as it turns out, caricatures tend to activate face cells more strongly than photographs.
Resolution and Recognition
Humans have a remarkable capacity for recognizing faces at low resolutions. Take a look at the image below, can you tell who it is? Back away from your computer and try again. Now mouse over the image to reveal what shouldn’t be a surprise.
Vision scientists Harmon and Julsez published a block portrait of Abraham Lincoln in the ‘70s demonstrating this phenomenon. Three years later, Salvador Dali produced Gala Contemplating the Mediterranean Sea, which when seen from 20 meters away becomes the portrait of Abraham Lincoln (Homage to Rothko). This 6.3 x 8.3 ft oil on canvas is two paintings in one. At a distance of no more than 20 meters onlookers observe a back-facing nude gazing through a window onto a harbor. Beyond 20 meters, a second image comes into view: a portrait of Abraham Lincoln. Like the Harmon and Julsez block portrait, which is actually inset at the lower left of the painting, the image of Lincoln is composed of colored tiles whose average hue and luminance are analogous to those of the pixels one should find in a low resolution pixilated photograph of Lincoln.
Contemporary artist Chuck Close executes a similar feat in his portrait work, but rather than creating one portrait out of another holistic image (like Dail’s Gala), he grids his canvases into tiles; each of the tiles can stand on its own as an abstract composition but when the whole set of tiles is viewed at a distance, a coherent portrait is seen. What Close’s paintings tell us about the psychology of visual perception is useful and surprising: highly detailed features are not necessary to face recognition. Moreover, it is the holistic configuration of features that matters, though researchers are still working to determine which configuration measurements (such as distance between the eyes) are most important and how those spatial relationships are encoded and evaluated. Further, researchers have found that our ability to recognize faces at increasingly low resolution increases with familiarity. Hence, Dali’s choice to depict the iconic image of Abraham Lincoln, one of the most widely familiar faces of his time, contributes to the success of this double-image perceptual-switching painting.
Facial Expressions
Facial expressions are a form of nonverbal communication governed by the contraction and relaxation of facial muscles. They convey information regarding the individual’s emotional status, though voluntary action can be used to override the involuntary expression in some cases. Darwin showed that facial expressions were universal across cultures, and even drew parallels with other species.
Facial expressions are most apparent at lower-resolution, probably due to blurring caused by subcutaneous fat above the muscles, as suggested by Maragaret Livingstone in her book Vision and Art: The Biology of Seeing. In the text, she points out that this can be observed in Leonardo DaVinci’s Mona Lisa where viewing Lisa’s mouth from your periphery draws out her illusory smile. It was also found that low resolution images of a woman baring a frightened face generated stronger responses in the amygdala, the brain region responsible for vigilance response, of subjects than did the fine details. |
