Scientists and children belong together in another way. The new research shows that babies and young children know and learn more about the world than we could ever have imagined. They think, draw conclusions, make predictions, look for explanations, and even do experiments. Scientists and children belong together because they are the best learners in the universe. And that means that ordinary adults also have more powerful learning abilities than we might have thought. Grown-ups, after all, are all ex-children and potential scientists.
Walk upstairs, open the door gently, and look in the crib. What do you see? Most of us see a picture of innocence and helplessness, a clean slate. But, in fact, what we see in the crib is the greatest mind that has ever existed, the most powerful learning machine in the universe. The tiny fingers and mouth are exploration devices that probe the alien world around them with more precision than any Mars rover. The crumpled ears take a buzz of incomprehensible noise and flawlessly turn it into meaningful language. The wide eyes that sometimes seem to peer into your very soul actually do just that, deciphering your deepest feelings. The downy head surrounds a brain that is forming millions of new connections every day. That, at least, is what thirty years of scientific research have told us.
Piaget concluded that babies aren't just born in possession of adult knowledge, either from a past life or from DNA. Instead, Piaget thought that children must have powerful learning mechanisms that allow them to construct new pictures of the world, pictures that might be very different from the adult picture. When we learn about the world, when we do science. for example, we don't just hit the right answer once and for all. Rather, there is a very gradual unfolding sequence of corrected errors, expanded ideas, and revised misconceptions as we approach more and more nearly to the truth. That was what the Piagets saw as they watched their babies make their way through infancy.
But Piaget also thought that learning was just as rooted in biology as any innate idea in the genetic code. He often used the metaphor of digestion: babies' minds assimilated information the way babies' bodies assimilated milk. For Piaget, learning was as natural as eating. This idea is the second element in the new developmental science.
Vygotsky saw that adults, and especially parents, were a kind of tool that children used to solve the problem of knowledge. in contrast to our—probably necessary—parental megalomania. Vygotsky noticed, for example, how adults, quite unconsciously, adjusted their behavior to give children just the information they needed to solve the problems that were most important to them. Children used adults to discover the particularities of their culture and society.
But Vygotsky also thought that the adult influence on children's minds was fundamentally biological, a part of our basic human nature. He emphasized the role of language. Language is a natural, biological, and unique feature of human beings, but it is also the medium by which we transmit our cultural inventions. Just as Piaget saw that learning was innate. Vygotsky saw that culture was natural.
The theories that did dominate psychology, especially in America, were Freudianism and the behaviorism of psychologists like B. F. Skinner. Both theories had lots of things to say about young children. But like Aristotle with the teeth, neither Freud nor Skinner took the step of doing systematic experiments with children or babies. Freud largely relied on inferences from the behavior of neurotic adults, and Skinner on inferences from the behavior of only slightly less neurotic rats. And like the philosophers, Freud and Skinner got the developmental story wrong, too. Freud saw children as the apotheosis of passion, creatures so driven by lusts and hungers that their most basic perceptions of the world were deeply distorted fantasies. Skinner's view was that children were the ultimate blank tablets, passively waiting to be inscribed by reinforcement schedules.
You're lying in bed in the labor room of the hospital and you're about as exhausted, as utterly worn out, as you'll ever be. Giving birth is this peculiar combination of determination and compulsion. It's you pushing, and you push in a more concentrated, focused way than you've ever done anything, but in another sense you don't decide or try to push or even want to. You are just swept away by the action. It's like a cross between running a marathon and having the most enormous. shattering, irresistible orgasm of your life.
And then suddenly, in the midst of all this excitement and action, agitation and exhaustion, there is a small, warm body lying on your chest and a tranquil, quiet, wide-eyed face looking up at yours. Maybe it's just the natural endorphins flowing through you once the actual pain is gone, but instead of collapsing, as you might expect, you feel a kind of intensified alertness. You're preternaturally awake, and everything is clearer and sharper than usual. And through the next night or two, when the nurses have finally left you alone, and the helpful husband has gone home to get some sleep and tell the relatives, you lie with the baby in your arms and inhale that peculiar, sweet, animal, newborn smell, and you look, and look for an hour at a time, at the small, still somewhat squished face. And the baby, perhaps also under the spell of the endorphins, alert as he won't be for some days to come, looks at you. And then and there—before the sleepless nights and diapers and strollers and snowsuits have kicked in—that gaze seems to signify perfect mutual understanding, complete peace, absolute happiness.
That's the romance of it, anyway. The romance doesn't come, sadly, with every birth, just as the parallel romance of true love doesn't come with every sexual encounter. But, just as with true love, it is one of the great gifts of life and seems more than worth the risk of disappointment and the reality of pain.
But why should you believe us instead of those benighted experts who thought babies couldn't really see? How can we say we actually do know what babies think? With the help of videotape, scientists have developed ingenious experimental techniques to ask babies what they know. One whole set of techniques has been designed to answer two simple questions: Do babies think that two things are the same or different? And if they think they're different, do they prefer one to the other? You can present babies with pairs of carefully controlled events and see whether they can differentiate between them and which they prefer to look at or listen to. For instance, you can show babies a picture of a human face and a picture of a complicated object, like a checkerboard. Then an observer. who doesn't know what the babies are looking at, records their eye movements. By analyzing the babies' eye movements, you can see which picture they looked at longer. You can take the same idea a bit further by getting babies to suck on pacifiers that turn on different video- or audiotapes and determining which tapes they are willing to do some work for. You can see. for instance, if they will keep a tape of their own mother's voice playing longer than a tape of a stranger's voice.
Finally, you can exploit the fact that babies, like the rest of us, get bored. If you show babies the same old same old over and over, they stop looking and listening. Change the tape to something new, and they perk up and take notice. Developmental scientists call this boredom "habituation." So, for instance, you can show a series of different happy faces, and the babies will gradually lose interest; they'll habituate. Show them a new happy face, and they hardly look any longer. But show them a sad face, and they start to stare again. This means that babies somehow know that the happy faces are the same and the sad face is different.
Using these sorts of techniques we can show that at birth. babies can discriminate human faces and voices from other sights and sounds, and that they prefer them. Within a few days after they're born, they recognize familiar faces, voices. and even smells and prefer them to unfamiliar ones (it even looks as if they recognize their mother's voice at birth based on the muted but still audible sounds they hear in the womb) They'll turn toward a familiar face or voice and even toward a pad that has been held close to their mother's skin and turn away from other faces, voices, and smells.
There are other reasons to think that even very young babies are especially tuned to people. Babies flirt. One of the great pleasures in life is to hold a three-month-old in your arms and talk absolute nonsense. "My, my, my," you hear your usually sane, responsible, professional voice saying, "you are a pretty bunny, aren't you, aren't you, aren't you, sweetums, aren't you a pretty bunny?" You raise your eyebrows and purse your mouth and make ridiculous faces. But the even more striking thing is that that tiny baby responds to your absurdities. He coos in response to your coo, he answers your smile with a smile of his own, he gestures in rhythm with the intonation of your voice. It's as if the two of you are engaged in an intricate dance, a kind of wordless conversation, a silly love song, pillow talk. It's sheer heaven.
But aside from being sheer heaven, it's also more evidence that babies spontaneously coordinate their own expressions, gestures, and voices with the expressions, gestures, and voices of other people. Flirting is largely a matter of timing. If you look around at a party, you can tell who's flirting just by looking at them, without even hearing a word. What you see is the way two people time their gestures so they're in sync with each other and with nobody else in the crowded room. She brushes her hair off her face, and he puts his hand in his pocket; she leans forward eagerly and talks, and he leans back sympathetically and listens. It's the same way with babies. When you talk. the baby is still; when you pause, the baby takes her turn and there's a burst of coos and waving fists and kicking legs. Like imitation, baby flirtation suggests that babies not only know people when they see them but also that they are connected to people in a special way. Like grown-up flirtation, baby flirtation bypasses language and establishes a more direct link between people.
Other experiments also show that one-year-olds have a radically new understanding of people. What happens when you show a baby something new, something a little strange, maybe wonderful, maybe dangerous—say, a walking toy robot? The baby looks over at Mom quizzically and checks her out. What does she think? Is there a reassuring smile or an expression of shocked horror? One-year-olds will modify their own reactions accordingly. If there's a smile, they'll crawl forward to investigate; if there's horror, they'll stop dead in their tracks.
Again we can show this quite systematically. For instance, a grown-up can look into two boxes. She looks into one box with an expression of joy and into the other with an expression of complete disgust. Then she pushes the boxes toward the baby, who has never seen inside the boxes. Nevertheless, the baby figures out something about what is inside just by looking at the experimenter's face: the baby happily reaches into the box that made her happy but won't open the box that disgusted her. The baby doesn't just understand that the other person feels happy or disgusted, but also understands that she feels happy about some things and disgusted about others.
Alison and Andy designed an experiment to test this idea further. First they set up an imitation game: you give the toy to me and I'll give it to you; you put the sticker on my hand and I'll put it on your hand. Children are very good at this and love doing it. Then Alison and Andy put a screen on the table between the experimenter and the child. The experimenter hid a toy from the child by placing it on her side of the screen. Then she gave the toy to the child and asked him to hide it from her. To do this correctly, the child had to put the toy on his side of the screen so that he could see it and the experimenter couldn't. But the youngest children, twenty-four- and thirty-month-olds, would often put the toy on the experimenter's side of the screen so that it was hidden from their own sight but was completely visible to the other person. And the toddlers actively experimented with this problem. They would walk over to the experimenter's side of the table to see how the screen looked from that side. Or they would invent ingenious ways of avoiding the problem, like hiding the toy behind their back so that it was hidden from everyone. Just as with the picture at the end of the tube, they couldn't seem to get their minds around the idea that they could see the toy but someone else couldn't.
Before they are three, though, children do learn about the differences between what they see and what other people see. A thirty-six-month-old, barely turned three, will always hide the toy correctly on his side of the screen. He knows that the other person can't see it even though he can himself. He can predict quite explicitly when you will see the object and he won't; he'll tell you that you can't see it but he can. Three-year-olds can even tell you about what an object looks like from different perspectives. If you put a yellow toy duck behind a piece of blue plastic, it will look green. You can show this trick to three-year-olds and let them see that the duck really is yellow. Three-year-olds will say that the duck looks green to the person on one side of the plastic but looks yellow to the person on the other side. Contrary to much conventional wisdom, these very young children are already beginning to go beyond an egocentric understanding of other people.
Alison has done other experiments that point in a similar direction. For example, three-year-olds seem to be unable to remember how they learned about something, even when the events took place only a few moments before. In one study the experimenter hid a cup under a cloth "tunnel," a wire arch covered with cloth, with an opening at either end. Children found out what was underneath the tunnel in one of three ways: they picked up the tunnel and saw the cup, they put their hands in the tunnel through the openings and felt it, or the experimenter simply told them, "There's a cup inside." Then she asked the children what was under the tunnel. They always got that answer right. But the next question was harder. She asked, "How do you know there's a cup in the tunnel? Did you feel it, or did you see it, or did we tell you about it?" Children were confused about how they had found out about the object. They said, for example, that they had seen the cup when actually she had told them about it. (These experiments have obvious implications for very young children's eyewitness testimony. Children aren't any more likely to lie than adults, and they don't confuse fantasy and reality. but they may well confuse what they saw and what a well-meaning lawyer or social worker told them.)
Children's discoveries about belief also have consequences for other aspects of their relations to people. To deceive peopie, or to recognize that they are deceiving you, you need to be able to understand the differences between what they believe and what you believe. Doing that depends on understanding the way beliefs work. It depends on knowing what you have to do to make someone believe something that isn't actually true. Two- and three-year-olds are such terrible liars. they hardly qualify as liars at all. A three-year-old will stand on the other side of the street and yell back to you that he didn't cross it by himself. They are terrible liars just because they don't seem to understand what it takes to make someone have a false belief. We can show systematically that "real" lies only begin to appear at about four, at the same time that children start to understand "false-belief" problems like the deceptive candy box. Similarly, children only begin to understand that they can be deceived at about that age.
While learning to lie may not, at first, seem like a terrifically desirable skill, some kinds of deception are essential to civilized life. Children don't even seem to understand the necessary lies we call politeness until about four or five years old. They are baffled by a scenario in which someone pretends pleasure at an unwelcome birthday gift or hides the pain of a skinned knee under a show of stoicism. The idea that you could feel one emotion and yet express another seems contradictory to them. This may be why, in their everyday life. young children also have such a hard time masking their own emotions, another reason that life with a three-year-old can be like a twelve-hour-a-day performance of Tosca.
But we also have some more direct evidence for the idea that children learn like scientists. Alison and Virginia Slaughter, one of her students, looked at three-year-old children who didn't yet fully understand belief—children who still said they had always thought that there were pencils in the candy box. Then, over the course of a few weeks, Virginia gave the children systematic evidence that their predictions were false. She told them firmly that they hadn't said pencils at all, they had said candies. When the children predicted that Nicky would think there were pencils in the box, she dragged Nicky in and asked him. Another group of children got very similar training about number problems—problems that had nothing to do with the children's understanding of the mind. At the end of the two weeks she asked the children a new question about false beliefs (about a set of soaps that looked like golf balls) The children who had received counter-evidence to their mistaken ideas about the candy box did much better on questions about the golf-ball soap than the children who had learned about numbers. But she also asked the children new questions about all sorts of other aspects of belief, questions such as where beliefs come from and how appearances and realities differ. The children who had gotten the counter-evidence not only did better on the questions about the trick objects, they also did better on lots of other questions about belief. By providing just the right kind of evidence at just the right time, we seem to have provoked a big, sweeping change, a sort of theoretical revolution, in the way these children thought about the mind.
We think that children learn about other people, and that they learn the same way scientists learn about the world. At first they may just ignore counter-evidence that contradicts their theory. In fact, three-year-olds will tell you that they said there were pencils in the box when they first saw it and will even maintain that Nicky said there were pencils in the box, when he actually said just the opposite. Gradually, though, as enough different kinds of contrary evidence accumulate, it's no longer possible to just ignore or reinterpret the facts. When the new theory finally replaces the old one, there are far-reaching implications. The new theory doesn't just let us deal with the contrary evidence; it also lets us understand many other phenomena in a new way. And the new theory lets us create a whole set of new predictions about what will happen in the future.
1. If you show very young babies a video of a static Big Bird that then explodes into its separately defined parts, they won't be perturbed. Because all the parts had separate edges anyway, they may, for all the babies know. have been separate objects to begin with. But if you show them Big Bird moving first, so that they see that all the parts of the object move together, and then show them the exploding Big Bird, they'll look much longer and more attentively, as if they recognize that something is wrong. Seeing the parts move together, seeing their common fate, seems to tell the babies that this is just one object and that its parts are eternally joined together. So babies already have some principles they can use to impose order on a chaotic world.
Movement seems to be important for babies in other ways, too. Very young babies already know a surprising amount about how objects characteristically move. Young babies not only can follow the movements of an object in front of them, they seem to be able to predict how an object will move in the future. Suppose you show the babies an object following a particular trajectory—that is, moving in a particular path at a particular speed—say, a ball rolling on the table. Now the ball rolls behind a screen. They will look ahead to the far edge of the screen, to the place where the object ought to appear if it keeps moving at the same rate and on the same path. If the object does appear there, the babies are unperturbed and keep following the object. But if the object doesn't appear there, or if it appears at the wrong spot or too quickly or too slowly, they look intently at the edge of the screen for much longer. Sometimes, in fact, they look back to the other edge of the screen, or look farther ahead along the path the object should have taken. They seem able to predict where the object should be and when it should get there.
Another great English philosopher, John Locke, posed another classical epistemological problem. What would happen if you miraculously restored the sight of someone who had been blind from birth? Would that person recognize all the objects he had known so intimately through touch, or would he have to painstakingly learn that the smooth, hard, curved surface looked like a porcelain teacup, or that the familiar, soft, yielding swells and silky hairs translated into a visual wife? Locke thought that the blind man would have to learn to make connections between the two types of experience.
Babies are a more common miracle than suddenly cured blind men, and it turns out you can ask them Locke's question, too. They think Locke, like Berkeley, got it wrong. Andy gave one-month-old babies one of two pacifiers to suck on. either a bumpy one or a smooth one. The babies never saw the pacifiers. They just felt them. Then he let the babies look at bumpy and smooth objects, without letting them feel them. The babies looked longer at the object that was the same shape as the one they had just been sucking on. Somehow, they could relate the feel of the pacifier in their mouths with its visual image.
You can ask the same question about the relationship between sound and vision. Even newborns will turn their heads and look toward an interesting noise, suggesting that they already expect to see something in the direction of the noise. You can do more systematic experiments to test this, too. For instance, you can show babies two objects bouncing at different times and play an audiotape of a boing, boing, boing sound that is synchronous with only one of them. Babies can tell which visual display matches what they hear; they look longer at the one that bounces in sync with the audiotape.
Even more startling, Andy and Pat showed babies a silent video of a face saying either ahhh or eeee, and then they played the babies audiotapes of each vowel sound. Five-month-olds could tell which face went with which sound. They looked at the face with the wide-open mouth when they heard the ahhh sound and at the face with pulled-back lips when they heard the eeee sound. Babies evidently have a primitive ability to lip-read, at least for simple vowels. (This was a provocative experiment—all those wide-open mouths and ahhhs. Soon after they finished doing the study together, Andy and Pat got married.)
So in the first few months of life, babies already seem to have solved a number of deep philosophical conundrums. They know how to use edges and patterns of movement to segregate the world into separate objects. They know something about how those objects characteristically move. They know that those objects are part of a three-dimensional space. And they know the relationship between information that comes from their different senses—they can link the feel of a nipple and its pink protuberance, the sound of a voice and the moving lips they see, the ball's exuberant bounce and its accompanying boing.
As scientists we think that everything is mediated by physical causality of some sort, including our interactions with other people. There are, in fact, light and sound waves that go from one person to another even if we can't see them with the naked eye. But from our everyday point of view, it appears we are able to influence people without any direct physical contact at all. (It's probably that fact that makes telepathy seem plausible to so many people.) After all, just looking at someone across a crowaea room can set quite a dramatic chain of events in motion. We influence people psychologically by communicating, talking, gesturing, and making faces—we don't have to touch them. In fact, trying to physically manipulate other people to get them to do what we want is usually quite counterproductive, if not actually illegal. Psychological causality is often our most powerful tool.
Psychological causality is particularly important for babies, not only because they can't push things around as much as we can, but because they have to get other people to satisfy most of their needs. Wnen very young babies first try to influence the external world, they may not differentiate between physical and psychological causality, and this may lead to the apparently magical and irrational quality of many of their actions. They make the mistake of using psychological means to try to influence the physical world. Smiling and cooing can get a reaction from Mom even though you're not physically attached to her. It's as if they think maybe they'll have the same effect on the mobile.
In fact, much of what we think of as magical, irrational thinking in adult life may really reflect the same sort of confusion between physical and psychological causality. Shamans and magicians say special words, wave their hands in particular ways, and take care in choosing particular garments in order to influence events in their world. This may seem odd and irrational, but when you think about it, all of us do this when we're trying to influence other people (well, two out of three of us for the garments). If you can use words to get someone into a white-hot rage or into bed with you, why not try to use words to give someone a disease or make her pregnant? "Magical procedures" of this type, whether in children or in adults, are, in fact, ineffective, but believing in them may not really be irrational—just mistaken. They may be based on a confusion about where psychological causality leaves s off and ordinary physical causality begins.
By the time babies are about a year old, there seems to be an important change in their understanding of causes. They seem to have learned something about the differences between psychological and physical causality, and they understand more about how physical causation works. They also know something about how events or objects can influence each other. Younger babies can learn to produce an action that has an effect in the world. For example, they can pull a cloth that has a toy on top of it toward them. The peculiarities and limitations of that understanding become clear, though, when you present the babies with a new, slightly different problem by putting the toy to one side of the cloth. The babies pull on the cloth just as intently and are startled to see that nothing happens, just as they keep kicking even when the ribbon is disconnected. By the end of the first year, though, babies no longer make this mistake; they seem to know right away that the object has to be on top of the cloth. They won't pull the cloth if the object is to one side of it. (In fact, they may give the experimenter a definite "Are you kidding?" look.) This greater understanding of physical causality means their actions look much less magical and are much more effective. This allows them to really plan and scheme and use physical objects as tools.
However, there is some surprising evidence that young babies are actually not particularly interested if a blue toy car goes in one edge of the screen and a yellow toy duck emerges at the far edge on the same trajectory! A grown-up would assume the duck that came out was brand-new and the other toy was still there behind the screen. But young babies seem content to think the toy somehow magically became a new kind of thing behind the screen. The particular kind of category-crossing magic trick in which the scarf turns into a dove wouldn't be surprising to them. Although young babies can discriminate between yellow and blue, and between the duck shape and car shape, they don't seem to rely on these features to determine which object this is. By the time babies are a year old, however, it is easy to show that across a wide range of situations they are surprised when the car turns into a duck, which suggests they have developed a new view of categorization.
Babies do other things that suggest they have a new view of categories. Alison and Andy gave babies a mixed-up bunch of objects: four different toy horses and four different pencils. Alison would put her hands palm up on the table and watch what the babies did with the objects. Nine- and ten-month-olds picked up the horses and pencils, played with them, and often put them in her hands, but they did so pretty much at random. But twelve-month-olds would sometimes pick all the objects of one group, all the horses or all the pencils, and put them in a hand or in a single pile on the table. By the time they were eighteen months old, babies would quite systematically and tidily sort the objects into two separate groups, carefully placing a horse in one hand and then a pencil in the other. In one experiment a particularly fastidious and precise little girl (there actually are fastidious eighteen-month-olds) noticed that one of the pencils had lost its point. She looked carefully at both hands and then reached for her mother's hand to make a separate spot for this peculiar and defective object.
By the time they are two or three years old, children already seem to have a deeper conception of what it means for an object to belong to a category. They can go beyond the superficial appearance of an object and comprehend something about its essential nature. And they begin to understand that knowing an object's category lets you predict specific new things about the object. For instance, you can tell three-year-olds some new fact about a particular object, you can point to a rhinoceros and say, "This rhinoceros has warm blood." If you then tell them that another animal is a rhinoceros, they will say that it has warm blood, too. But they won't extend their new discovery to a triceratops, which looks like a rhinoceros, if you describe it as a dinosaur.
It turns out that, just by the nature of the grammar of their languages, Korean- and English-speaking parents talk about the world quite differently. Korean (like Latin or French) uses an elaborate system of different verb endings to convey different meanings. As a consequence, Korean-speaking parents can, and often do, omit nouns altogether when they talk to their children. A Korean mother can say the equivalent of "moving in" when she sees the baby put a block in a cup, without saying anything about who or what is doing the moving or what it's moving into. In English, on the other hand. we must include at least one noun in almost every intelligible sentence. Moreover, English-speaking parents spend a lot of time pointing to objects and giving them names: "There's a dog! Look at the bird! Car! Airplane!"
Alison and a Korean colleague, Sonja Choi, looked at the kinds of things English-speaking mothers and Korean-speaking mothers said to their eighteen-month-old babies and found that this was indeed true: English-speaking mothers used more nouns and fewer verbs than Korean-speaking mothers. English-speaking mothers tended to name objects a lot, while Korean-speaking mothers were more likely to talk about actions.
When Alison and Soonja looked at what the eighteen-month-old children understood about the world, they found there were consistent differences between the Korean and English speakers. Like their parents, the Korean children used more verbs than the English-speaking kids, while the English-speaking kids used more nouns. But in addition, the Korean-speaking children learned how to solve problems like using the rake to get the out-of-reach toy well before the English-speaking children. English speakers, though, started categorizing objects earlier than the Korean speakers. For instance. they were more likely to put the toy horses and the pencils into two separate piles. It was as if the Korean-speaking children paid more attention to how their actions influenced the world, while the English-speaking children paid more attention to how objects fit into different categories. The likeliest explanation for this is that the children were influenced by what the grown-ups around them said, which in turn was shaped by the grown-ups' language.
We mentioned that part of what makes learning language difficult is that languages carve up sounds and different Ianguages carve them up differently. A wide variety of different sounds, with very different spectrograms, will all seem like the same sound to us, and, in turn, that sound will seem sharply different from other sounds that are actually quite similar to it physically. Suppose you use a speech synthesizer to gradually and continuously change one particular feature of a sound, such as the consonant sound r, and play that gradually changing sound for people. You very gradually and continuously change the r sound to /. What is actually coming into the listeners' ears is a sequence of sounds, each of which is just slightly different from the last. But what they perceive is someone saying the same sound, r, over and over, and then suddenly switching to a new sound, /, over and over. The listeners have divided up the continuous signal into two sharply defined categories: either it's an r or an I, not anything in between. They can't distinguish between all the different r's. even though the sounds themselves are quite different. Scientists call this categorical perception, because a continuously changing set of sounds is perceived categorically as being either black or white, r or /, with nothing in between.
The way we categorically perceive speech is unique to each language. In English we make a sharp categorical distinction between rand Z sounds. Japanese speakers don't. In fact,Japanese speakers can't hear the distinction between American r and /, even when they are listening very hard. (Hence all the dubious jokes about Japanese speakers ordering what sounds like "flied lice" instead of "fried rice.") Pat was in Japan to test Japanese adults and their babies on the r-/distinction. She had carefully carried the computer disk with the rand /sounds to Japan, and when she arrived in the laboratory in Tokyo, she played them on an expensive Yamaha loudspeaker. She thought that such clearly produced sounds would surely be distinguished by her Japanese colleagues, who were quite good English speakers as well as being professional speech scientists. As the words rake, rake, rake began to play out of the loudspeaker, Pat was relieved to know that the disk worked and the sound was perfect. Then the train of words changed to an equally clear lake, lake, lake, and Pat and her American assistant smiled, looking expectantly at her Japanese colleagues. They were still anxiously straining to hear when the sound would change. The shift from rake to lake had completely passed them by. Pat tried it over and over again, to no avail.
Why do the speakers of different languages hear and produce sounds so differently? Ears and mouths are the same the world over. What differs is our brains. Exposure to a particular language has altered our brains and shaped our minds, so that we perceive sounds differently. This in turn leads speakers of different languages to produce sounds differently. When and how do babies start to do this? Do they start out listening like a computer, with no categorical distinctions? Or do they start out with the categorical distinctions of one particular language, say English or Japanese or Russian?
We can't ask babies directly whether they think two sounds are the same or different, but we can still find out. Very young babies can tell us what they hear by sucking on a special nipple connected to a computer. Instead of producing milk, sucking on this special device produces sounds from a loudspeaker. one sound for each hard suck. Babies love the sounds almost as much as they love milk: they may suck up to eighty times a minute to keep the sounds turned on. Eventually, though. they slow down; they get bored hearing the same thing over and over again. When the sound is changed, however, infants immediately perk up and suck very fast again to hear the new sound. That change in their sucking shows that they can hear a difference between the new sound and the sound they heard before. Using this technique we can do the same r and l experiment we just described with adults. We can use a speech synthesizer to present the babies with a slowly and continuously changing consonant sound. Then we can test the babies to see which sounds they think are the same and which sounds they think are different.
Scientists anticipated that these tests would show that very young babies initially can't hear the subtle differences between speech sounds and only slowly learn to distinguish those that are important in their particular language, such as r and l in English. In fact, the results were just the reverse. In the very first tests of American infants listening to English, babies one month old discriminated every English sound contrast we threw at them. Moreover, the babies demonstrated the categorical perception phenomenon. They thought all the r's were the same and different from all the l's, just as adult English speakers do.
But then shortly afterward speech scientists discovered something even more remarkable. Kikuyu babies in Africa and Spanish babies in Mexico were also excellent at discriminating American English sounds as well as the sounds of Kikuyu and Spanish, and American babies were just as good at discriminating Spanish sounds—much better than American adults. The sophisticated Japanese scientists who strained to hear the difference between rake and lake would not have had any trouble doing so when they were forty or fifty years younger. Very young babies discriminated the sounds not only of their own language but of every language, including languages they had never heard. Infants were as good at listening to American English as they were at listening to African Kikuyu, Russian, French, or Chinese regardless of the country they were raised in. Pat also discovered that babies, unlike computers, make these distinctions no matter who is talking—a man or a woman, a person with a high squeaky voice or one with a deep resonant voice.
So babies start out knowing much more about language than we would ever have thought. Newborn babies already go well beyond the actual physical sounds they hear, dividing them into more abstract categories. And they can make all the distinctions that are used in all the world's languages. Babies are "citizens of the world." Perhaps we grown-up scientists failed to predict this because our skills are so much more limited. Our citizen-of-the-world babies clearly outperform their culture-bound parents.
Initially children use just a few names, mostly for familiar things and people. But when they are still just beginning to talk, many babies will suddenly start naming everything and asking for the names of everything they see. In fact, what'sat? is itself often one of the earliest words. An eighteen-monthold baby will go into a triumphant frenzy of pointing and naming: "What'sat! Dog! What'sat! Clock! What'sat juice, spoon. orange, high chair, clock! Clock! Clock!" Often this is the point at which even fondly attentive parents lose track of how many new words the baby has learned. It's as if the baby discovers that everything has a name, and this discovery triggers a kind of naming explosion.
It turns out you can show experimentally that babies at this stage have a new approach to learning words. You can give a baby just one example of a new nonsense word naming a new kind of thing ("Look, a dax!" you say, pointing to an automatic apple corer), and it will become a permanent part of the baby's vocabulary. Weeks or even months later, he'll correctly identify the "dax." Just one salient instance and babies will internalize a word forever (sometimes, of course, with rather embarrassing consequences). The process is called fast mapping. The babies seem to assume at once that the new name they hear names the new object they've just seen. Babies start to fast-map at about the time they have their naming explosion.
The tests show that babies' preferences have nothing to do with the actual words mothers use. Babies choose motherese (or "parentese" or "caretakerese") even when the speaker is talking in a foreign language so infants can't understand the words, or when the words have been filtered out using computer techniques and only the pitch of the voice remains. Apparently they choose motherese not just because it's how their mother talks but because they like the way it sounds. Motherese is a sort of comfort language; it's like aural macaroni and cheese. Even grown-ups like it. Pat's graduate students discovered that listening to the lab tapes of motherese in a foreign language was a wonderful therapy for end-of-term stress. The mother's voice is an acoustic hook for the babies. It captures babies' attention and focuses it on the person who is talking to them.
The elaborate techniques of computer voice analysis reveal exactly what it is we do when we talk to an infant. The pitch of our voice rises dramatically, sometimes by more than an octave; our intonation becomes very melodic and singsongy; and our speech slows down and has exaggerated, lengthened vowels.
Motherese is a universal language. People across all cul:ures do it when they talk to their infants, even though they usually aren't aware of doing it at all. When mothers listen to recordings of themselves producing motherese, the reaction is: That can't be me. I sound really stupid. Should I be doing that? But they do it intuitively, without conscious awareness.
Why do we do it? Do we produce motherese simply to get the babies' attention? (It certainly does that.) Do we do it just to convey affection and comfort? Or does motherese have a more focused purpose? It turns out that motherese is more than just a sweet siren song we use to draw our babies to us. Motherese seems to actually help babies solve the Language problem.
Motherese sentences are shorter and simpler than sentences directed at adults. Moreover, grown-ups speaking to babies often repeat the same thing over and over with slight variations. ("You are a pretty girl, aren't you? Aren't you a pretty girl? Pretty, pretty girl.") These characteristics of motherese may help children to figure out the words and grammar of their language.
But the clearest evidence that motherese helps babies learn comes from studies of the sounds of motherese. Recent studies show that the well-formed, elongated consonants and vowels of motherese are particularly clear examples of speech sounds. Mothers and other caregivers are teachers as well as lovers. Completely unconsciously they produce sounds more clearly and pronounce them more accurately when they talk to babies than when they talk to other adults. When mothers say the word bead to an adult, it's produced in a fraction of a second and it's a bit sloppy. But when mothers say that same word to their infants, it becomes beeeeeed, a well-produced, clearly articulated word. This makes it easier for infants to map the sounds we use in language.
The job of computer scientists, of course, is to design the programs that let electronic computers accomplish those impressive feats of thinking and knowing. The computer scientists have to figure out how to make programs that get to the right kind of output from the right kind of input. But our job as cognitive psychologists is rather different and even harder. We are more like archaeologists than engineers.
Actually, it's a familiar Star Trek story. We have landed on a planet that already contains amazing biological computational devices. They were designed eons ago over millions of years by a force far more powerful than any we possess. The one thing we know about them for sure is that they employ incredibly advanced technology. There are no operating manuals, no wiring diagrams, no Homo Sapiens for Dummies. We can't even hope that sometime in the last few minutes of the show the all-powerful designing intelligence will take over one of our crew and explain its intentions in a suitably resonant and spooky voice (the usual Star Trek resource in these situations). We're on our own.
When a three-month-old, a one-year-old, and a four-year-old look at the same event, they seem to have very different thoughts about it. They seem to transform the light waves and sound waves into different representations, and they use different rules to manipulate those representations. Children don't have just a single, fixed program that gets from input to output. Instead, they seem to switch spontaneously from using one program to using another, more powerful program. That makes babies and children look very different from the computers we have now...
How can we explain these changes? One idea might be that the changes are simply a result of the fact that babies grow, the way caterpillars change into butterflies as they grow. or the way we develop breasts and beards as we grow and reach puberty. The changes might just involve a genetic blueprint that unfolds on a particular maturational timetable. The child's program for understanding false beliefs might appear when she's four the same way her breasts appear when she's twelve. After all, we don't think that the caterpillar learns how to be a butterfly. Similarly, we might not think that the child learns about false belief any more than she learns how to have breasts.
Another very different possibility is that we change our ideas about the world just by taking in more and more information about it. We simply accumulate more and more input. Then we associate some pieces of that input with other pieces. We hear the dinner bell and food comes, and after a while we link the bell with the food. We give a particular answer to the experimenter's question and we get praise, and after a while we try to give answers like that. Babies could end up linking particular inputs to each other and to particular outputs in this sort of specific, piecemeal way.
Babies start out believing that there are profound similarities between their own mind and the minds of others. That belief gives them a jump start in solving the Other Minds problem. But during the first three years they also observe the differences in what people do and say. Those differences stem from the fact that all minds aren't actually entirely alike. Babies and young children watch and listen with careful focused interest as their mother refuses to let them touch the lamp cord or as their older brother tells them they are completely wrong. This new evidence makes babies revise the beliefs they started out with.
Similarly, babies start out knowing that space is threedimensional and that objects move in predictable ways. They even reach out to objects and shrink away from them. By the time they are eighteen months old, as they watch and manipulate the things around them, as they play peekaboo and sort things into piles, they see those objects act in new ways and they look for ways to explain what they see. They learn that three-dimensional moving objects continue to exist no matter how they appear or disappear, and they learn that all those objects belong in categories. By three or four they have transformed those first categories into biological species and "natural kinds," as they begin to understand that kittens become cats and that tigers have guts inside and rocks don't.
Finally, babies start out making all the possible distinctions between the sounds of languages. Like citizens of the world. mierican newborns can distinguish African Kikuyu sounds as well as English sounds. By twelve months, as they repeatedly hear the sounds of their own language, babies create new representations that reflect the sound categories of their particular language. One-year-old American babies can't discriminate Kikuyu categories anymore, but they can discriminate the English categories better, and they have even become "English-sounding" babblers.
In each case the things babies already think influence where they will go next. They determine which events will engage them, which problems they will tackle, which experiments they will do, even which words they will listen to. Then babies change what they think in the light of what they learn.
Babies have another ability that man-made computers lack. They can do things. They can actively intervene in the world as well as passively learn about it. A one-year-old can reach for a new rubber duck, put it in his mouth, bang it against the side of the tub, splash it in the water, and watch his father's reactions to all of this. A key aspect of our developmental picture is that babies are actively engaged in looking for patterns in what is going on around them, in testing hypotheses, and in seeking explanations. They aren't just amorphous blobs that are stamped by evolution or shaped by their environment or molded by adults.
In Chapter Three we described how children need to figure out what's going on around them—they have a kind of explanatory drive. This drive pushes them to act in ways that will get them the information they need; it leads them to explore and experiment. The apparently pointless activities we call play often seem to be the result of this drive. Babies who are who are figuring out how we see objects play hide-and-seek; babies who are figuring out the sounds of language babble. It's all very serious fun.
The baby computers start out with a specific program for translating the input they get into accurate representations of the world and then into predictions and actions. But the interesting thing about these computers is that they don't stop there. Instead, they reprogram themselves. They actively intervene in the world to gather more input and check their predictions against that new input. The things they find out lead them to construct new and quite different representations and new and quite different rules for getting from inputs to representations. If we wanted to make a new computer as powerful as the biological computers, this is what it would have to be like.
Other, less tragic kinds of evidence also support this idea. Most people have a much more difficult time learning a second language late in life than they do in childhood. Immigrants may try to learn the language of their new country, only to be outdone by their own children. When we visit a foreign country for a while, our kids seem to be happily chatting with the other kids in the playground, while we are still painfully looking through the phrase books. When we learn a second language past puberty, we speak with a foreign accent—in other words, with phonetics, intonation, and stress patterns that are not appropriate for the new language. We also have more difficult)^ understanding spoken speech and more difficulty with the grammar of the language. Puberty seems to be an important time. An immigrant who speaks nothing but English from the age of eighteen on may still have a heavy accent in his old age; another immigrant who arrives at four years of age may have no trace of one.
Raising children is an intrinsically difficult and uncertain job in ways that science can't really address. For most of us parents there is literally nothing more important than the well-being of our children. There are not many things we could imagine giving our lives for, but we could give our lives for them. And, in a less melodramatic way, of course, we do give our lives for them. For fifteen or twenty years our everyday energy, our individual liberty, our income, our attention, our concern are all devoted to our children. There is nothing else in human experience to match it.
And yet all this seriousness and commitment, this moral purpose, is combined with a deep, even necessary, lack of control. A British prime minister once intoned that the press wanted "power without responsibility, the prerogative of the harlot throughout the ages." Perhaps it's fitting that the prerogative of the mother is the opposite of the prerogative of the harlot: we parents have responsibility without power. Mothers and fathers are at the mercy of innumerable accidents—accidents about the random genetic mix of temperament and ability, accidents about how that mix interacts with our own temperaments and abilities, accidents about what our lives happen to be like in the few years that constitute a childhood, accidents about what the rest of the world has to offer our children.
There is also a deeper sense in which we have less control than responsibility. The whole point of the enterprise, after all, is to end up creating an autonomous agent, a person who can leave us, who can choose to make grave mistakes and decide to be thoroughly miserable. It's like falling utterly, madly. deeply in love and yet knowing that in twenty years the object of your affections will leave you for other lovers and, in fact, that your job is to make your beloved leave you for other lovers. The very best outcome is that our children will end up as decent, independent adults who will regard us with bemused and tolerant affection; for them to continue to treat us with the passionate attachment of infancy would be pathological. Almost every hard decision of child-rearing, each tiny step--Should I let her cross the street? Can he walk to school yet? Should I look in her dresser drawer?—is about how to give up control, not how to increase it; how to cede power, not how to gain it.
We want certainty, and that leaves us open to fraud. Mothers used to lie awake listening to their babies scream because the experts said not to pick the baby up or to feed "off schedule." They might well have felt that bloodletting would have been preferable.
One benefit of knowing the science is a kind of protective skepticism. It should make us deeply suspicious of any enterprise that offers a formula for making babies smarter or teaching them more, from flash cards to Mozart tapes to Better Baby Institutes. Everything we know about babies suggests that these artificial interventions are at best useless and at worst distractions from the normal interaction between grown-ups and babies. Babies are already as smart as they can be, they know what they need to know, and they are very effective and selective in getting the kinds of information they need. They are designed to learn about the real world that surrounds them, and they learn by playing with the things in that world, most of all by playing with the people who love them. Not the least advantage of knowing about science is that it immunizes us from pseudoscience.
We could also immediately change workplaces to allow for part-time work that has similar benefits and pay to full-time work and to allow for flexible hours and career paths. Our own workplaces, the universities, provide both very good and very bad examples. For years professors have worked at home and determined their own schedules with no loss of productivity. On the other hand, the career structure of universities is deeply in conflict with the imperatives of evolution—the years when we expect academics to work the hardest and longest hours are exactly the years when women can have children.
The very automaticity of our response to babies suggests that it can be combined with doing other things, as it surely was in the Pleistocene. Perhaps the telecommuting home office with the crib next to the fax machine will turn out to be the contemporary equivalent of the baby in the sling on his mother's back or the father plowing next to his children. Perhaps the circle of fellow workers and friends will help replace the extended family group. Grandparents and uncles and aunts have also disappeared from children's lives just when they are most needed, and grandchildren and nieces and nephews have sadly disappeared from our lives. Perhaps we will construct institutions that allow people whose own children have grown up, or who don't have children, to be involved with other people's children.
Until very recently doctors didn't use analgesia when they operated on small babies, because they thought their minds were too primitive to really feel pain or to remember it if they did. This is a dramatic example, but it often seems as if we discount children's pain compared with adult pain. Child abuse isn't evil because it may produce neurotic adults but because it abuses children. Divorce doesn't have a cost because it may produce adults who have difficulty with relationships but because it causes emotional pain to children. Parents aren't important because they may shape their children's adult personalities but because they are the most profound influence on children's lives while they are children. Looking at babies attentively makes us treat them differently.
We think there are very strong similarities between some particular types of early learning—learning about objects and about the mind, in particular—and scientific theory change. In fact, we think they are not just similar but identical. We don't just think that the baby computers have the same general structure as the adult-scientist computers, in the way that perceptual learning and artistic learning and political learning may all have the same general structure. We think that children and scientists actually use some of the same machinery. Scientists are big children. Scientists are such successful learners because they use cognitive abilities that evolution designed for the use of children.
Parents with college graduates still living in the spare room may occasionally envy the mother cats and father birds who ruthlessly throw their young out after a couple of months. But we know that we couldn't summon up a similar ruthlessness, nor would our babies survive if we did (of course, the college graduates may be a different story). No creature spends more time dependent on others for its very existence than a human baby, and no creature takes on the burden of that dependence so long and so readily as a human adult.
These features of our evolutionary design are consistent with the idea that human beings have unusually powerf-ful and flexible learning abilities. We deploy those abilities during that protected and protracted Eden we call childhood. During our immaturity we don't have to commit ourselves to act t in any particular way in order to survive; grown-ups take care of us. That leaves us free to explore many possibilities and to learn just what to do in our particular world. Childhood is a time when we can safely devote ourselves to learning about our specific physical and social environment. We can do pure, basic research while the grown-ups provide the funding and the technology.
For most grown-ups, for most of history, that learnirgg may have largely stopped when we reached maturity and turrmed to the more central evolutionary business of the four /'s (feeding, fleeing, fighting, and engaging in sexual reproduction) We learned most of what we need to know a long time before kindergarten. As adults we can survive in our particular world because as children we figured out how it works.
All the same, the continued existence of these learning abilities allows some of us, some of the time, to continue to learn new things about the world around us. When we give grownups leisure and money and interesting problems to solve, they can be almost as smart as babies. We think that, throughout history, some adults continued to learn new things about the world, especially when they were relevant to particular problems of survival. This might explain, for example, the achievements of hunter-gatherer "folk botany" or of Australian aboriginal geography. But the contingencies of history some five hundred years ago gave many more adults the chance to learn about the world. We invented institutions that re-created the conditions of childhood—protected leisure and the right toys. We call those institutions science
Five hundred years ago a natural activity of children was transformed into an institutionally organized activity of adults. Of course, this transformation led to many differences between what children do and what scientists do. Perhaps the most important difference is that children typically make up theories about close, middle-sized, common objects, including people. As a result they are positively immersed in evidence that is relevant to their theories. Everything they need to know is easily available to them. Scientists, in contrast, often make up theories about objects that are very small or very big, hidden or rare or far away, and the relevant evidence is often very thin on the ground. They make up theories about things such as distant stars and elusive diseases. This relatively small difference has big cognitive and social consequences.
What are the babies' representations and rules like?
First, the babies' representations are rich and complex. As we've seen, they include ideas about how their face resembles the faces of others, how objects move, and how the sounds of a language are divided. The young babies' world is not simple. Babies translate the input at their eyes and ears into a world full of people with animated, expressive faces and captivating. intricate, rhythmic voices. It's also a world full of objects with complex multidimensional structure that move in a dizzying variety of ways.
Babies' representations are also abstract. They go beyond the data of immediate sensation. Most obviously these early representations link information from different senses: they link the way the tongue feels and the way it looks, the bounce of a ball and the boing sound it makes, the look of an open mouth and the sound of an aah.
But the representations go beyond sensation in other, more profound ways. They turn facial expressions into emotions. They convert two-dimensional images into three-dimensional objects. They take a continuous stream of noise and divide it up into discrete speech sounds. Even newborn babies end up with representations that are radically different from the input at their eyes and ears. The babies' world isn't concrete any more than it's simple. Babies already see the soul beneath the skin and hear the feeling behind the words.
These representations and rules lead young babies to interpret what happens to them in particular ways—to pay attention to some things and ignore others. At first they are particularly captivated by faces and voices; within a few days they pay special attention to familiar faces and voices. At first they pay special attention to the way things move and less attention to their shape or color or texture; later they will start to pay more attention to these properties of objects. At first and not others; later they will no longer attend to sound changes that once intrigued them.
Finally, the babies' representations and rules allow babies to form expectations, and even to make predictions, about new things that will happen in the future. When the babies' program gets information about a current event, it can generate a representation of a future event. When young babies see a toy car go behind the screen, they look ahead to the far edge babies flirt, they expect that their coos will be answered by adult goos. When they see an open mouth, they expect that they will hear an aah sound. They react in characteristic ways when their predictions turn out to be wrong and their expectations are dashed. They show conflict when the toy car doesn't appear to behave as it should, and they are distressed when their flirtatious advances are met with an impassive stony face. Just as the babies' world isn't simple or concrete, it also isn't limited to the here and now. Even very young babies can remember what happened in the past and predict what will happen in the future.
The significance of this inborn program goes beyond just the simple fact that there is a lot there to begin with. The baffling problem for philosophers and psychologists was always how we get from the raw, undigested matter of sensation—the "blooming, buzzing confusion"—to an understanding of the world. How do we even know which kinds of sensations to pay attention to? The answer the babies give is that we are never dealing with raw matter. There never is a blooming, buzzing confusion. From the very beginning we can understand the world, pick and choose what's important. know what to expect. From the time we're born, we run a program that translates the light and sound waves into people, objects, and language.
By the time babies are about one-and-a-half yearsrs old, they start to understand the nature of these differences between people and to be fascinated by them. Again we can demonstrate this systematically. Alison and one of her students, Betty Repacholi, showed babies two bowls of food, one full of delicious Goldfish crackers and one full of raw broccoli. All the babies, even in Berkeley, preferred the crackers. Then Betty tasted each bowl of food. She made a delighted face and said. 'Yum," to one food and made a disgusted face and said. ''Yuck," to the other. Then she put both bowls of food near the babies, held out her hand, and said, "Could you give me some?"
When Betty indicated that she loved the crackers and hated the broccoli, the babies, of course, gave her the crackers. But what if she did the opposite and said that the broccoli was yummy and the crackers were yucky? This presented the babies with one of those cases where our attitude toward the object is different from theirs, where we want one thing and they want something else. Fourteen-month-olds, still with their innocent assumption that we all want the same thing, give us the crackers. But the wiser (though, as we will see, sadder) eighteen-month-olds give us the broccoli, even though they themselves despise it. These tiny children, barely able to talk. have already learned an extremely important thing about peopie. They've learned that people have desires and that those desires may be different and may even conflict.
We can demonstrate this discovery in the laboratory, but it is also dramatically apparent in ordinary life. Parents all know. and dread, the notorious "terrible twos," when the adorable if somewhat out-of-hand one-year-old rogue becomes a steely-eyed two-year-old monster out of melodrama. What makes the terrible twos so terrible is not that the babies do things you don't want them to do—one-year-olds are plenty good at that—but that they do things because yon don't want them to. While one-year-olds seem irresistibly seduced by the charms of forbidden objects (the lamp cord made me do it), the two-year-olds are deliberately perverse, what the British call bloody-minded. A two-year-old doesn't even look at the lamp cord. Instead his hand goes out to touch it as he looks. steadily, gravely, and with great deliberation, at you.
But this perverse behavior actually turns out to be quite rational. Just as experiments with very young babies explain our parental intuition that we have a special kind of rapport with our newborns, experiments with toddlers explain our intuition that that rapport sometimes breaks down when they get older. Two-year-olds have just begun to realize that people have different desires. Our broccoli experiment shows ths childreren oDnly begin to understand differences in desires when they are about eighteen months old. Fourteen-month-olds seem to think that their desires and ours will be the same. The terrible twos seem to involve a systematic exploration of that idea, almost a kind of experimental research program. Toddlers are systematically testing the dimensions on which their desires and the desires of others may be in conflict. The grave look is directed at you because you and your reaction, rather than the lamp cord itself, are the really interesting thing. If the child is a budding psychologist, we parents are the laboratory rats.
It may be some comfort to know that these toddlers don't really want to drive us crazy, they just want to understand how>w we work. The tears that follow the blowup at the end of a terrible-twos confrontation are genuine. The terrible twos refleets a genuine clash between children's need to understand other people and their need to live happily with them. Experimenting with conflict may be necessary if you want to understand what people will do, but it's also dangerous. The terrible twos show how powerful and deep-seated the learning drive is in these young children. With these two-year-olds, as with scientists, finding the truth is more than a profession— it's a passion. And, as with scientists, that passion may sometimes make them sacrifice domestic happiness.
As they hear us talk, babies are busily grouping the sounds they hear into the right categories, the categories their particular language uses. By one year of age, babies' speech categories begin to resemble those of the adults in their culture. Pat conducted some even more complicated experiments with Swedish babies using simple vowels to see how early they start organizing the sounds of their language in an adult-like way. She showed that at six months the process has already begun. The six- to twelve-month time span appears to be the critical time for sound organization.
What might be happening to the babies between six and twelve months? One way of thinking about it is in terms of what Pat calls prototypical sounds. After listening to many r sounds in English, for example, babies develop an abstract representation of r—a prototypical r—that is stored in memory. When we want to identify a new sound, we seem to do it by unconsciously comparing the new sound to all of the prototypes stored for our language and picking the one that's the best overall fit. Once we've unconsciously done this, we distort the way we hear a sound to make it more like the prototype stored in memory than like the sound that actually hits our ears.
It's similar to what happens when you show people a drawing of something they've seen very often, a house, for example, and then ask them to copy it from memory. If the house you show them doesn't have a chimney, many people will add one to their drawing anyway, even though it wasn't in the original drawing they saw. Once they coded the picture as a house, they distorted their memory of it to make it more like what they think of as the prototypical house. We can do complicated analyses to show just what the prototypes of our speech sounds are and just how we distort what we hear to suit them. Our language prototypes "filter" sound uniquely for our language, making us unable to hear some of the distinctions of other languages. Pat's tests suggest that babies' language prototypes begin to be formed between six and twelve months of age.
It isn't just that younger babies have a skill they lose later on. Rather, the whole structure of the way babies organize sounds changes in the first few months of life. Before they are a year old, babies have begun to organize the chaotic world of sound into a complicated but coherent structure that is unique to their particular language. We used to think that babies learned words first and that words helped them sort out which sounds were critical to their language. But this research turned the argument around. Babies master the sounds of their language first, and that makes the words easier to learn.
When babies are around a year old, they move from sounds to words. Words are embedded in the constant stream of sounds we hear, and it is actually difficult to find them. One problem computers haven't yet solved is how to identify the items that are words without knowing ahead of time what they are. Try to find the words in a string of letters like theredonateakettleoftenchips. The string contains many different words: The red on a teakettle often chips or There, Don ate a kettle of ten chips and so on. Of course, in written language there are normally spaces between words. But in spoken language there aren't actually any pauses between words. That's why foreign speech sounds so fast and continuous, and that makes the Language problem very hard for computers to solve.