Hack 93. Understand What Makes Faces Special
We have dedicated neural machinery for recognizing faces from just a few basic features arranged in the right configuration.
It's an important evolutionary skill to be able to quickly and efficiently recognize faces that are important to us. This allowed our ancestors to conform to the social hierarchies of the groups in which they lived, to keep checks on who was stronger and who was weaker than they were, and to track potential mates.
While faces are very important things to recognize, they are also all remarkably similar. Eyes, noses, and mouthsand it is these features that we rely on most when we discriminate between facesall look pretty much alike, and the ratios of the spacing between them do not leave too much scope for differing widely either. Nevertheless, it is remarkably easy for us to distinguish between faces.
10.2.1. In Action
Take a look at the two pictures in Figure 10-1.
Figure 10-1. Two upside-down faces, but you should have no problem recognizing who it is1
While you might detect some sort of difference between them, the odds are that both will look like pretty normal upside-down pictures of a face (and you might well be able to identify who it is, too). Now turn the book upside down. The face on the right is a grotesque: its eyes and mouth have been inverted. But you probably didn't notice this (and it certainly is not as striking as when the faces are the right way up). This is a neat demonstration of the fact that faces are normally processed holistically. When they are the right way up, we "understand" faces as a whole based on their internal components; turning them upside down disrupts this ability. We then have to rely on componential encoding instead and judge the face simply in terms of the individual items that make it up. This makes it much harder to detect that something is "wrong" than when we are able to use holistic processing. While, of course, we rely on differences in hairstyle and color and other factors when identifying people in the real world, experiments have shown that we rely most on the central features of faces.
Another example of the way in which we are "primed" for the ability to recognize faces is how difficult it feels to look at the face shown in Figure 10-2.
Figure 10-2. It's difficult to look at this double face
This is because the two sets of internal features are competing with each other to allow us to make sense of the face. Neither set can win, so our visual system can't settle on the stimuli and make sense of it the way it would with a normal face.
10.2.2. How It Works
So how does the brain recognize faces? It turns out that there is a section of the brain that is specialized for recognizing facelike stimuli. In imaging studies,2 it has been shown that a section of the fusiform gyrus, which borders the temporal and occipital lobes, is more active when participants view images of faces than when they view other pictures. This area is now termed the fusiform facial area. It is specialized for viewing faces in an upright orientation, suggesting that faces as they are normally seen are treated as a specialized type of object by the brain. It is easy to recognize partial or degraded images, though, such as with low-quality CCTV images.3 Again, this would make sense in allowing us to identify people in low lighting or among other objects such as trees.
When looking at faces, our eyes dart most around the mouth and eyes [Hack #15], the two features that are absolutely essential to depict a face in a cartoon. Although other featuresparticularly hairare used to recognize individuals, it is the mouth and eyes that define a face. Experiments with subliminal stimuli have shown that faces, especially emotional faces, can be processed by the brain at exposures too quick for conscious appreciation. Even a face shown to you too fast to be consciously seen influences your feeling of familiarity with it; mere exposure [Hack #82] is good enough. This is another sign of specialized neural networks for face processing in the brain.
10.2.3. In Real Life
There is some debate concerning whether this "face recognition" system is actually specialized for this purpose or is a general categorization system, with categorizing faces just one ability we've picked up through experience. For example, people who are expert at recognizing species of birds or types of cars show activation of the fusiform facial area when shown pictures of their specialist subjects.4 Another type of evidence comes from studies of people who have developed prosopagnosia, an inability to discriminate faces, even though the ability to discriminate other types of objects seems intact. Interestingly, some people who are experts with some category of objectsheep, for instancehave been known to retain the ability to discriminate between the specialty objects even when they lose the ability with faces.5
Whether or not the fusiform facial area is specialized for faces, it is clear that we are very good at identifying faces. Most researchers agree that by 3 or 4 months of age babies are skilled at facial discrimination.6
We are so good at identifying faces that our brains are primed to see them everywhere. We see them in clouds (where they aren't) and in smileys (where they are). Our ability to perceive faces so readily from a relatively small amount of information in the smileys or emoticons used in text and online messagingthe fact that we can readily understand that :-) is intended to represent a happy face and ;-) a winkis a legacy of how important face recognition has been to us evolutionarily.
10.2.4. End Notes