The development of memory by the human brain from conception to adulthood is an extraordinary story. The physical differences among the infants shown in Figure 3-1, who are arranged in order of increasing age from 2 to 18 months, are easy enough to see. Harder to see—but suggested by the differences in the sizes of their heads—are their brains, which have been growing, processing information, and creating memories, starting several months before birth.

The human brain has upward of a trillion nerve cells. A number like this is much too large for most of us to grasp (nevertheless, politicians seem to have no difficulty tossing such numbers around when they talk about federal budgets and deficits). A better idea of how large this number is comes from looking at how fast new neurons multiply over the nine months of the brain’s development. From conception to birth the growing human brain adds new neurons at the astonishing rate of 250,000 every minute!

Actually, a newborn baby’s brain has more neurons than at

FIGURE 3-1 Babies from 2 to 18 months of age.

later ages. Some neurons die as the brain is shaped and sculpted by experience. We used to think that the total number of neurons present in the brain at birth was it. Some would die, but no new neurons would form. We now know that this is not the case. The brain forms neurons throughout life. We will look more closely at these surprising facts and how they relate to learning and memory later.

Have you ever encountered a young baby you didn’t know? Perhaps a chance encounter in an elevator with an infant lying in her carriage. She looks at you, at your face, with an intense blank stare. One has the impression that behind that stare is an incredibly powerful computer clicking away at high speed. Beginning well before birth, experience shapes the brain.

At what point in the development of the brain does learning first occur? The growing brain of the fetus shows a simple form of learning a month or more before birth. If a loud sound is presented close to the mother, her fetus will show changes in heart rate and also make kicking movements. If the sound is given repeatedly, these responses will gradually cease or habituate. If the nature of the sound is markedly changed, the fetus will again respond. This must mean that the fetus has learned something about the properties of the initial sound.

Habituation is the most basic form of learning. It is simply a decrease in response to a repeated stimulus and occurs in a wide

range of animals, in addition to humans. Habituation of such responses as looking at an object is widely used to study learning in newborns and young babies. You might be interested to learn that the rate of habituation of responses in very young infants is correlated with later measures of IQ in the growing child—faster habituation is associated with higher IQ. We know a good deal about how habituation occurs in the nervous system too, as we will see later.

What kinds of stimuli does the developing fetus experience? In heroic studies in which mothers have swallowed microphones, it is clear that sounds from the outside penetrate the womb. Not just loud sounds, but even conversations close to the mother can be detected. The voice that comes through loud and clear is the mother’s own voice. At birth an infant can hear more or less like a partially deaf adult, with thresholds considerably higher than those of normal adults (20 to 40 decibels higher), but by about six months of age an infant’s hearing is virtually normal.

Although the eyes develop early, as the fetus grows there is little light in the womb. A newborn baby has visual acuity resembling that of a rather nearsighted adult, somewhere between 20/200 and 20/600, but the direction of gaze can be somewhat controlled by the newborn. The newborn appears to be largely colorblind, but by two to three months color vision is normal. The reasons for the poor vision of the newborn lie not in the eyes but in the brain. The development of the visual part of the brain, and the role of experience in this development, is an extraordinary chapter in neuroscience (see Box 3-1).

It used to be thought that the mind of the newborn is a blank slate and that the world is a vast and meaningless sea of sensations. Part of the reason for this is that a newborn has little ability to control her movements. A young baby cannot lift her head, roll over, or even point at objects with her arms. How do you ask such a helpless little creature what she sees or knows? As it happens,

babies do one thing extremely well: they suck. They can also move their eyes and head from side to side to look at objects, and they can kick. We can use these simple behaviors and their habituation to ask the infant what she sees, hears, and can learn.

Newborn babies will learn to suck on an artificial nipple to turn on certain types of sounds. They particularly like the sound of their mother’s voice, much more than the speech of another mother. They have learned enough about their mother’s voice in utero to recognize it after birth. An infant can also display one response to its mother’s voice and another response to a stranger’s voice prior to birth. This was shown in a study of near-full-term fetuses in which each fetus was exposed to the recorded sound of its mother’s voice or the voice of a stranger. Fetal heart rate increased in response to the mother’s voice and decreased in response to the stranger’s voice. Newborn babies also prefer their native language to foreign languages.

Similar results have been obtained for more complex speech, as shown by the remarkable study dubbed the “cat in the hat” experiment. Pregnant women read aloud twice a day during the last six weeks of pregnancy. Each mother was assigned one of three children’s books, one of which was Dr. Seuss’s The Cat in the Hat. The mothers-to-be who were assigned this story read it aloud faithfully twice a day for the entire six-week period. Recordings were made of the readings of all three stories. Three days after birth the infants were tested using the sucking response on a pacifier on all three stories. Infants that had, for example, been read The Cat in the Hat in utero sucked so as to produce this story over the other two, even if the story was read by another mother. (Never underestimate the power of Dr. Seuss!) Exactly

how the babies learned to recognize this particular story from experience in utero is a mystery, but learn it they did.

What do newborns see? Or, more specifically, what do they like to look at? Faces are very potent stimuli for humans and other primates. If shown her mother’s face versus other women’s faces, a baby just a few hours after birth will look more at her mother’s face. What exactly tells the infant it is her mother? If the women wore scarves over their hair and part of their heads, the babies no longer showed any preference for their mothers. They apparently had learned a kind of global perception of the mother’s head rather than attending to smaller details of the face.

Shown drawings of faces versus drawings of mixed-up faces, newborns prefer to look at faces. Newborns show an extraordinary ability to learn about individual faces. When shown pictures of different individual women’s faces and then shown composite faces made up from individual faces that they had either seen or not seen before, babies show a clear preference for composites made up from faces they had seen before. This learning occurs in less than a minute in human infants as young as eight hours old!

Equally remarkable is the sophistication of the kind of learning that human infants are capable of, even very early in postnatal development. Cross-modality matching refers to a process by which information experienced exclusively in one sensory modality (such as tactile stimulation) can “transfer” for use in connection with stimuli or events in another modality (such as vision). This was shown in a study in which infants first experienced a pacifier of a given shape (such as example A in Figure 3-2) solely by sucking on it—the infants never saw it. When the infant sucked hard enough on the pacifier, a computer monitor would present a visual display to the infant. The display showed a picture of the pacifier being sucked on or a picture of a differently shaped pacifier (such as example B) in a random order. Infants as young as 12 hours old spent more time looking at the picture of

FIGURE 3-2 Infants can visually recognize the shape of the pacifier simply by sucking it.

the pacifier they had in their mouth during the testing. (As testing progressed, there was a shift to spend more time looking at the new pacifier; this is the novelty preference effect again.) It is not known how the infant brain is able to accomplish this kind of “going beyond what’s given”—that is, of abstracting information. It has become clear in recent years that many psychologists—including the renowned child cognitive psychologist Jean Piaget—have severely underestimated the learning and cognitive abilities of human infants.

Adults recognize individual human faces far more accurately than individual monkey faces, and the opposite is true for monkeys (see Figure 3-3). These differential abilities seem to be due to early learning. Using the looking-preference procedure, studies have shown that six-month old infants are equally good at discriminating between individual human and individual monkey faces they have or have not seen before, but by nine months the infants are like adults: They readily distinguish between old (previously seen) and new human faces, but they no longer discriminate between old and new individual monkey faces.

There is a face “area” in the human brain, a localized region of cerebral cortex termed the fusiform gyrus. Patients with damage to this area can still identify faces as faces, but they can no

FIGURE 3-3 Human and monkey faces.

longer recognize individual faces (a condition called prosopagnosia). They cannot recognize their own relatives or even spouses by face, only by voice. Indeed, they cannot even recognize themselves in the mirror—if they happen to bump into a mirror, they might apologize to the face in the mirror. If recording electrodes are placed in this brain area, nerve cells show face-selective responses.

There is a similar area in the monkey brain. This discovery, made by Charles Gross then at Harvard, is a dramatic case of serendipity. Gross and his associates were studying the responses of single neurons with a recording electrode in a region of the monkey cerebral cortex thought to be a higher visual area. The monkeys were anesthetized and then presented with various simple visual stimuli—spots of light, edges, and bars—the standard elementary visual stimuli used in such studies. The neurons in this area responded a little to those simple stimuli but not to

sounds or touches, so it seemed to be a visual area. But the neurons were not very interested in their stimuli. After studying one neuron for a long time with few results, they decided to move on to another. As a gesture, the experimenter said farewell to the cell by raising his hand in front of the monkey’s eye and waving goodbye. The cell responded wildly to this gesture. Needless to say, the experimenters stayed with the cell. This particular cell liked the upright shape of a monkey’s hand best. Other neurons in this region responded best to monkey faces. This region of cerebral cortex in the monkey corresponds more or less to the face area in the human cerebral cortex.

Is the face area formed by experience and learning or is it built into the brain? The answer seems to be both. A newborn infant prefers to look at faces rather than nonfaces but must learn to identify its mother’s face or other faces. This learning can be very rapid, so there must be a “face area” in the brain that is formed by the organization of the neuronal circuits in this region of the cerebral cortex. It is innate. But individual faces must be learned. Further, it seems that the memories for individual faces may be stored in this localized face area, given the devastating loss of ability to recognize individual faces when the area is damaged.

Studies of infant memories for faces have been performed using tests that do not require verbal responses or complex motor movements. One such procedure is based on the measurement of “looking” preferences. In the response-to-novelty task, an infant first looks at a picture of a face; then at varying intervals the infant is shown two pictures—the previously seen one and a new one—and the experimenter records the amount of time the infant spends looking at each (see Figure 3-4). The typical finding is that the infant spends more time looking at the new picture, which implies that the infant has retained information about the original picture.

An ingenious method of studying memory in young infants was introduced by Carolyn Rovee-Collier at Rutgers University. She

FIGURE 3-4 A toddler looking at two pictures.

hung a mobile over an infant’s crib and attached a ribbon to the infant’s foot. When the infant saw an object she had seen before, she kicked more, showing that she remembered having seen it. This kind of task is termed operant training or conditioning. The infant is in control of her behavior. This procedure was initially developed by B. F. Skinner many years ago to probe the learning abilities of animals.

Two- or three-month-old infants were trained in the task and showed excellent retention at 24 hours. When tested with a different item, they did not respond at all. They showed both memory and discrimination at 24 hours. In a series of studies the duration of memory was determined to be a function of the infant’s age and to be virtually a straight-line function, reaching many weeks at 18 months. So memory improves dramatically with age.

Even longer-lasting memories have been found for babies using imitation. Babies love to imitate. If you stick out your tongue at a young infant, she will do the same to you (see Figure 3-5). Imitation is one of the most powerful learning tools available to

FIGURE 3-5 An infant imitating an adult.

infants and young children. Using imitation children learn to mouth their first words and to master the nonverbal body language of facial expressions and posture. Infants are also capable of deferred or delayed imitation. When shown a sequence of actions modeled by an adult, such as removing a mitten from a puppet’s hand, infants repeat the behavior. At six months of age infants are capable of imitating the action 24 hours after seeing it. At nine months, babies can remember for a month, but by the age of 20 months most toddlers remember this task for a year! This kind of memory is called recall memory. The baby has to remember how to remove the mitten. Simply recognizing the puppet is not enough. In general, recall memory is more difficult than recognition memory for adults as well as children and is especially difficult when it involves information about the order of events or actions.

Babies clearly develop long-term memories for past experiences beginning at the age of about six months or even earlier. By the end of the first year of life they are beginning to learn and remember words. Obviously, adults retain skills and information acquired during infancy and early childhood, yet adults seem unable to consciously remember and describe anything of their own experiences before about age 3. This is called infantile amnesia.

Sigmund Freud first described infantile amnesia. Freud thought the period extended back to age 6, but more recent studies indicate that it may extend to age 3 or 4. Freud initially thought this might be due to repression of existing memories but apparently later abandoned this idea.

As we have seen, infants learn and remember very well indeed, beginning at birth and continually improving with age. Why adults cannot remember much if anything about their own experience before age 3 or 4 is a mystery. We are referring here to what is called autobiographical memory. Such memories are described in words (that is, declarative memory, deliberate conscious recollection). Such memories are usually time tagged; we remember more or less when and where important events occurred in our lives. This is particularly true for very emotional experiences. Some researchers argue for autobiographical memory going back to as young as age 2, but the claim is controversial and the relevant research is hard to do. For example, in some studies researchers report having found memories in young adults of independently verified events such as injuries and hospitalizations that occurred around age 2, but a major and obvious problem here is the difficulty of ruling out the argument that such “recollection” may be based on later family discussions and reminders of the events.

Infantile amnesia has important relevance in real-world situations. Someone who claims to remember having been sexually molested at the age of six months is likely to have a false memory. Recent work suggests that infantile amnesia has much to do with language. It appears that we cannot describe in words events that

occurred in our lives before we knew the words to describe them. Sound confusing? Enter the “magic shrinking machine.” Investigators at the University of Otaga in New Zealand exposed children to a unique event at an age when they could just barely talk. They visited the homes of toddlers two to three years of age, bringing with them the magic shrinking machine. The child would place a teddy bear in the machine, close the door, and pull on levers. The machine would make odd sounds. The child would then open the door and—behold—the teddy bear was now much smaller. The toddlers greatly enjoyed this. Their language skills were also evaluated at that time.

A year later the investigators returned with the machine. The children, now a year older and with much improved language skills, recognized the machine. However, when asked to remember what it did, they used only the few words they knew a year earlier when they first saw the machine. People cannot reach back in time to non–verbally coded memories and describe them with words. These observations may provide at least part of an explanation of infantile amnesia. A complementary explanation concerns the fact that certain regions of the cerebral cortex, where we think language is learned and stored, require several years after birth to mature fully.

You may find it surprising that with all of the available evidence, there is still a great deal of controversy about the nature of infant memory. The debate centers on the vexing concept of consciousness. Recall that one of the defining features of episodic memory is that it consists of conscious recollection of events from one’s personal past. Some theorists claim that human infants lack this form of memory. We think the issue is irrelevant. As Carolyn Rovee-Collier points out, many of the basic processes of learning, retention, and forgetting can be found in infants in much the same way as in adults. What is different is language. Infants cannot store memories verbally because they haven’t yet learned to speak. As with the magic shrinking machine, preverbal memories cannot be expressed verbally, and words are the only tool we have to study consciousness, as when people describe their experiences.

Looking at objects in a room, we see chairs, tables, and so on. For a long time it was believed that young infants did not actually see objects as separate entities. Again, the problem is that young infants are incapable of interacting with objects. Elizabeth Spelke and her associates at the Massachusetts Institute of Technology completed an ingenious series of studies which indicated that by three months of age infants do perceive objects as distinct from backgrounds. They do not have to manipulate objects to know this, as earlier thought. Visual experience of objects and the way that experience shapes the brain are enough.

Infants also seem to grasp the basic notion of cause and effect. In one study a film showed a red brick moving across a screen until it encountered a green brick at which point the red brick stopped and the green brick moved off. If the two bricks collided and there was a pause before the green brick moved, the infant responded as though surprised. Using variations on this scenario, infants as young as six months of age showed some appreciation of cause and effect.

Some scientists have argued that Pavlovian conditioning may be the basis for the understanding of causality. Due to experiences with the world, the brain has come to be so wired over the course of evolution. Remember Pavlov’s basic experiment: A bell rings and a few seconds later meat powder is put in a dog’s mouth, which causes the dog to salivate. After a few such experiences, the bell alone elicits salivation. It doesn’t work the other way around: Put meat powder in the dog’s mouth and then sound the bell. That bell would be of no help in predicting food. Since Pavlovian conditioning occurs in a wide range of animals, from relatively simple invertebrates to humans, we argue that cause and effect is “understood” by virtually all animals possessing brains. It forms the basis of adaptive behavior, behavior based on the ability to predict consequences.

Young infants seem to think that if an object is suddenly hidden and they can no longer see it, it has ceased to exist. Dangle an enticing object such as a toy insect in front of an infant and he will happily reach for it. Before he reaches it, cover it with a towel. The infant suddenly looks blank and shows no interest in the towel. Instead, cover the toy with transparent plastic and he will pull the plastic off to grab the toy. It appears that when the object is covered such that it cannot be seen, the infant does not remember it is there.

Between about 7 months and 12 months of age, there is dramatic growth in an infant’s ability to remember that hidden objects still exist. The same is true for young monkeys. The task used to study the development of this memory ability is called delayed response, and it is very simple. The infant, human or monkey, is shown a board with two wells in it. A desired object, a small toy or raisin, is placed in one well, and the two wells are covered with identical blocks. A screen is then interposed so that the tiny subject cannot see the board for a brief time. The screen is then raised, and the infant reaches the blocks and pushes one aside (he is allowed to push only one). With a delay of a second or less while the screen is down, a seven-month-old human infant quickly learns to go to the baited well. If the delay period while the screen is down is increased beyond a second or so, the task quickly becomes very difficult for the seven-month-old. However, there is a dramatic increase with age in the infant’s ability to remember where the object is hidden. By the age of 12 months the infant can easily remember for 10 seconds or more where the object is.

Monkeys show exactly the same progression in memory for hidden objects. But monkeys are born a good deal less helpless than humans and mature much more rapidly. At 50 days of age the infant monkey can remember for about one second the location of an object, and by 100 days can easily remember delays of 10 seconds or more. These are examples of the early development of a raw memory ability.

This simple test was developed more than 50 years ago to chart the growth of memory ability in monkeys but has only more recently been used with human infants. Instead, a seemingly similar task was developed years ago for human infants by the pioneering Swiss child psychologist Jean Piaget. He called his “task A and not B.” Actually, the task is the same as delayed response but is done a little differently. What is dramatically different is the performance of young infants.

Suppose only the left well is baited with a toy repeatedly, so it is always correct. The screen is lowered for only a second or so each time. The seven-month-old (human) learns this very well. Then, in full view of the infant, the other well is baited instead. The screen is lowered and raised after one to two seconds. The infant immediately goes to the wrong well, the one that had earlier been baited repeatedly, even though she saw the other well being baited. As the infant grows older, she can perform correctly at longer and longer delay intervals. In more recent work, infant monkeys showed exactly the same results as infant humans on A, not B, but over a shorter period of development.

It is somewhat bemusing that these two tasks existed separately and independently for many years—delayed response for monkeys and A, not B, for humans—perhaps because in earlier years the two disciplines of child psychology and animal behavior did not interact.

These two tasks, particularly the A, not B, seem to involve two kinds of processes. On the one hand, in both tasks the infant has to remember where the missing object is over a brief period of time. The real surprise comes when young infants always choose the incorrect well, even though they see the other well being baited. If they had simply forgotten which well was baited, they should choose each well equally. But in fact they do remember that one well was repeatedly baited in the past and are unable to prevent or inhibit their response to this well. Learning to inhibit responses or behaviors that are incorrect or inappropriate is a key aspect of memory.

As it happens, a good deal is known about the brain system that underlies our ability to remember where objects are and to

inhibit the wrong responses. The critical region of the brain is in the cerebral cortex in the frontal lobe—that is, the prefrontal cortex. It was discovered many years ago that small lesions in this region on both hemispheres in adult monkeys completely abolished the animal’s ability to perform the delayed response task at any delay. The lesion changed an adult monkey into an infant monkey, at least in this task. When adult monkeys with these small brain lesions were tested in the A, not B, task, they behaved just like young humans and monkeys. At delays of as little as two seconds, they chose the well that had previously been baited repeatedly, even though they saw the other well being baited. Growth in the ability of infant humans and monkeys to perform these tasks correspond closely to the maturation of this critical region of the frontal cortex. The number of synapses (neuronal connections) grows rapidly from birth to one year of age.

What about adult humans who have damage to this region of the frontal lobe? The delayed response and A, not B, tasks are much too simple for adult humans, even if they are much impaired in their behavior. They can use words to hold the memory of where the object is during the delay period, using what is called working memory. But if a more sophisticated, analogous task is used, they show dramatic impairment in this kind of memory. The key task, famous in neuropsychology, is the Wisconsin card-sorting task. A deck of cards with various symbols is shuffled, and the person being tested is given the cards and told to sort them into categories. The investigator does not tell the subject the rules but only says “right” or “wrong” each time a card is sorted. Once the subject has mastered a category, the investigator, unbeknownst to the subject, changes the category. Normal people catch on quickly to the change in category, even though they are only told “right” or “wrong.”

On the other hand, people with damage to the critical region of their frontal lobe find it almost impossible to correctly perform on the task. The problem comes when they have to change to another category. They can’t. They might even say they need to change categories but can’t do it. More precisely, they cannot in-

hibit responding to the earlier and now incorrect category. Shades of A, not B. They behave like human and monkey infants on this more complex A-B task. Incidentally, patients with the mental disorder schizophrenia also have great difficulty with the card-sorting task.

Rachel Gelman, now at Rutgers University, and her associates completed an interesting series of studies on the basic numerical abilities of children. It appears that the ability to count, at least a few items, is present as early as two years of age, well before language mastery. Indeed, infants as young as six months can distinguish one, two, or three objects. Toddlers do seem to understand the concept of counting, but their verbal behavior is not perfect. If there are two items, the child might count: “one,” “two,” but if three items, might count: “one,” “two,” “six.” Yet the number of counts she states equals the number of objects.

In addition, human infants have a sense of “numerosity” similar to that of adults. This really means a number sense, a feeling about larger versus smaller numbers of items. In elegant studies, Elizabeth Spelke and her associates, then at the Massachusetts Institute of Technology, tested numerosity ability in six-month-old infants. They were able to discriminate groups of visual objects when one group was twice as large as the other but not when it was only one and a half times as large. The same thing happened when the infants were given auditory sequences. Charles Gallistel, now at Rutgers University, has summarized evidence indicating that animals also may possess a sense of numerosity.

Interestingly, there are neurons in the posterior association areas of the cerebral cortex in cats and monkeys that behave like counters. One of the present authors (RFT) conducted the study in cats (Figure 3-6). A given neuron would respond to each stimulus (sound or light) or every second stimulus, up to about seven stimuli. Similar results were obtained for the monkey. A frontal area becomes involved as well. Two closely analogous areas in the

FIGURE 3-6 Regions in the cerebral cortex in cat and monkey where neurons respond to the number of objects and corresponding areas in the human brain.

human brain also are activated during numerical computations. A basic ability to count may be common among higher mammals and might be innate.

Although a sophisticated theory of mind does not develop until about age 5, even by age 1 babies have an amazing understanding of people. Suppose an adult is playing with a one-year-old and looks into two different boxes. When the adult looks into one box she has an expression of happiness, but when she looks in the other box she shows an expression of disgust. She then pushes the two boxes toward the one-year-old. What do you think happens? The baby happily reaches for the box that made the adult happy but won’t touch the other box. The baby not only recognizes when a person is happy or disgusted but also understands that things can make someone feel happy or disgusted.

The development of a theory of mind in children is a striking example of social learning. Over the period from about three to five years of age, children learn by experience with others that people have ideas, thoughts, and emotions and that they act on the basis of these internal representations. They learn that the mind can represent objects and events accurately or inaccurately. As each child’s theory of mind (actually theories about the minds of others) develops and becomes more sophisticated, the child is better able to predict the behavior of others, particularly as this behavior relates to the child.

The mind of a five- or six-year-old child is extraordinary. Not only does she have ideas about the minds of others, she is also aware that she has these ideas about ideas; she has self-awareness—so much so that by six to eight years of age children

find it very hard not to think. It is easy to see how and why the child develops her theory of mind. The better her theories are about other individuals’ minds and what their intentions are, the better she can respond to and even influence or control their behavior. The better her theory of mind, the better it is rewarded.

Some researchers believe that the beginnings of a theory of mind develop by way of imitation. As we saw earlier, children will imitate expressions. Partly as a result of imitation children learn that drooping shoulders in another person mean sadness. Facial expressions can mean happiness, sadness, anger, and so on. A primary aspect of the disorder autism is thought to be due to the lack of development of a theory of mind in autistic children.

Autism is one of the most common serious disorders of childhood. Earlier thought to be rare, it is now known to affect 1 in every 150 children age 10 and younger. If adults with autism are included, more than a million people in the United States have the disorder. It is much more common in boys than girls and the symptoms range from very mild to devastating. Individuals with severe cases can be profoundly retarded and unable to dress or even go to the bathroom without help. They often engage in meaningless repetitive behaviors, including self-injury, and may have temper tantrums and violent outbursts.

Children with the more severe forms of autism show a regular progression of symptoms that counter the normal development of learning in infants and children. At one year of age they still do not babble or point and by age 2 still cannot speak two-word phrases. A striking difference is that autistic infants and children do not imitate adults. If an adult pounds a pair of blocks on the floor, a normal 18-month-child will do the same, but young children with autism will not. Later, autistic children do not engage in pretend play, do not make friends, and do not make eye contact. They do make repetitive body movements and may have intense tantrums.

At the other extreme are people with Asperger’s syndrome, discovered in 1944 by Hans Asperger, an Austrian pediatrician. It wasn’t until 1994 that the American Psychiatric Association officially recognized Asperger’s syndrome as a form of autism. Children with Asperger’s syndrome are generally bright, sometimes gifted, and particularly adept at solitary repetitive activities like transformer toys or computer programming (see Box 3-2).

Asperger’s syndrome is much more subtle and usually not diagnosed until about age 6 or older. Such children have difficulty making friends, cannot communicate with facial and body expressions, and have an obsessive focus on very narrow interests. All autistic children have one characteristic in common. They seem unable to recognize that other people have minds that differ from their own. They think that what is in their mind is in everyone else’s mind and that how they feel is how everyone else feels. In short, they have never learned to develop a theory of mind.

Autism runs in families. If one identical twin has the disorder, the chances are 60 percent the other will too and a better than 75 percent chance that the twin without autism will exhibit one or more traits. It has been estimated that between 5 and 20 genes are involved in autism. One discovery of particular interest is the involvement of a gene on chromosome 7. A putative “language” gene maps to this same area. We will say more about this language gene in our treatment of language.

Several brain abnormalities have been associated with autism, but it is too early to draw firm conclusions. Some areas of the brain are smaller, particularly those involved in emotional behavior, and some areas are larger. There appear to be abnormalities associated with the cerebellum, an older brain structure earlier thought to be involved only with movement and motor control but now known to be involved in learning, memory, and cognitive processes as well. One particularly interesting finding from brain imaging work concerns the face area. Unlike normal people, this area does not become active at all in the brains of autistic individuals when they look at stranger’s faces. However, when looking at the faces of loved ones, an autistic’s face area and

related brain areas become much more active than is true for normal people. Why autistic children do not develop a theory of mind and what brain abnormalities are responsible remain mysteries. The lack of imitative behavior in autistic children may prove to be a most important clue.

Some very basic psychological principles of learning have been applied with great success in the treatment of autism. For over 30 years, Ivor Lovaas has run a clinic at the University of California at Los Angeles that uses basic operant conditioning techniques to teach communication to autistic children. Many of these children lack even the rudiments of social speech, and the training process proceeds in small steps with the goal of developing normal social speech. For example, suppose the therapist’s goal is to get an autistic child to respond appropriately to the question “What is your name?” The child does not respond, does not look at the therapist, and generally acts as if the therapist doesn’t exist. The training proceeds by rewarding the child with snacks for paying attention, first, and then for making eye contact (which might have to first be “prompted” by the therapist). Then the therapist focuses on manually shaping the child’s lips to prompt pronounciation of the first letter of his or her name and from there on rewards any of the child’s behavior that represents a step (sometimes very small ones) toward the goal of getting the child to pronounce his or her first name when asked. This training can last for weeks before the goal behavior is achieved. The program then continues to more complex speech behavior, always building on what has been learned and always based on the principle of rewarding desired behavior, never unwanted behavior. It is quite amazing to see the results of the training: a formerly uncommunicative child who now uses normal, spontaneous social speech. Lovaas’s success in treating autistic children lends further support to a statement made earlier: The limits of human abilities and their modifiability by learning are not yet known.

Perhaps the most dramatic example of a critical period in development is imprinting in birds. Birds like chickens, ducks, and geese that can walk and feed immediately after hatching, imprint on their mothers within the first day. The chick will stay close to its mother and follow her everywhere. The reader may have seen pictures of a mother duck leading her ducklings across a road, marching in single file.

If the mother is not present, the newborn chick will imprint on whatever object is present if it moves and makes noise. Imprinting works best for objects that look and sound like mother. In the laboratory, chicks have been imprinted on all kinds of objects, even little robots that move and make sounds. The classic example concerns the pioneering ethologist Konrad Lorenz, who imprinted a gaggle of goslings who were under the unfortunate impression he was their mother to follow him everywhere.

The critical period for imprinting is very brief. Once the chick has imprinted on the initial object, hopefully the mother, it will no longer imprint on other objects. Perhaps the closest example to imprinting immediately after birth in mammals occurs with olfactory learning. The human infant learns to identify and prefer the odor of his mother within the first day, especially if the mother is breast-feeding.

The now-classic example of a critical period in development for humans and other mammals is the visual system. Donald Hebb, a pioneer in the study of memory and the brain, has described clinical cases of people who were born and grew up with severe cataracts. They were essentially blind. Several had their cataracts removed as adults, after which their visual function, although improved, was still greatly impaired. They could see and quickly learned to identify colors, which are coded in the retina of the eye, but could not identify the shapes of objects. If presented with a wooden triangle, they could not identify it until they could touch and feel it. Object vision is coded in the brain, not the eye.

Being blind in one eye is almost as bad. The classic work on

the critical period in vision development was done by two scientists then at Harvard, David Hubel and Torsten Wiesel, who worked initially with cats. If they kept one eye of a kitten closed from the time its eyes opened after birth until about two months of age, the kitten was permanently blind in that eye. If the same procedure was done on an adult cat, there was no vision impairment in the closed eye. This same massive impairment in vision occurs in monkeys and humans, the only difference is the duration of the critical period: For humans it extends to about five years of age. Children born with severe impairments in one eye (for example, a cataract or squint), if not treated, will become permanently blind in that eye. Indeed, even keeping one eye closed for a period of several weeks in a one-year-old child—for example, for a medical procedure—can cause damage. Better to keep both eyes closed. In all these cases the eyes themselves are fine; it is the visual area of the cerebral cortex that is damaged.

Hubel and Wiesel were able to determine the processes in neurons in the visual cortex that led to this devastating loss of vision after one eye had been closed. The basics of the normal human visual system are summarized in Box 3-1. As noted there, both eyes project to each side of the visual cortex. In adults the neurons in the visual cortex are the first to be activated when an object is seen (activity from the eye is relayed to the visual thalamus and on to the cortex). They receive this input from either the left eye or the right eye but not from both (see Figure). They in turn converge on other neurons in the cortex that are activated by input from both eyes, permitting binocular vision.

At birth the situation is quite different. These first-activated cortical neurons receive virtually identical input from both eyes. Over the critical period, if vision is normal in both eyes, a kind of competition occurs so that each neuron ultimately loses input from one eye or the other to yield the adult pattern. But if one eye is closed during the critical period, activation from the seeing eye gradually takes over all the connections in each of these neurons and the input from the shut eye dies away. We think a reason for this is that activation—the occurrence of neuron action poten-

tials—is necessary to maintain synaptic connections in the critical period. Consequently, all input from the shut eye to the visual cortex vanishes.

This provides a particularly clear example of how genes and experience interact in human brain development. The basic connections, the neural circuits, are originally programmed by the genes and their interactions with the developing brain tissue. But the fine details of connections among neurons are guided by experience and learning. If one eye of a kitten is covered by an opaque goggle that transmits light but not detail, vision with that eye is still much impaired. Experience fine-tunes the organization of connections among neurons in the brain. These connections are termed synapses.

If any one process is key to the organization of the brain and to the formation of memory it is the formation of synapses, the functional connections from one neuron to another. Each nerve cell has one fiber extending out to connect with another neuron, called the axon (Greek for “axis”) and many other fibers extending out from the body of the cell that receive input from other neurons. These fibers are called dendrites (Greek for “tree,” the dendrites on a neuron resemble the branches of a tree). Each dendrite is covered by small bumps or spines. Each spine is a synapse receiving an axon connection from another neuron (see Figure 3-7). Physically, synapses consist of a small synaptic terminal (presynaptic—before the synapse) in close opposition to a spine on the postsynaptic (after the synapse) neuron dendrite. There is a very small space between the two. When this presynaptic terminal is activated it releases a tiny amount of a chemical neurotransmitter that diffuses across to the postsynaptic membrane to activate the neuron. The number of synapses on a neuron is astonishing. A major type of neuron in the cerebral cortex may receive up to 10,000 synaptic connections from other neurons. The champions are a type of neuron in an older brain structure, the Purkinje neu-

FIGURE 3-7 A typical neuron in the brain. The dendrites are covered with little bumps or spines. Each spine is a synapse, a connection from another neuron.

rons in the cerebellum, named for a famous anatomist. Each Purkinje neuron (there are millions) receives upward of 200,000 synaptic connections from other neurons!

Synapses grow, form, and die from well before birth through adulthood to death. The process of synaptic formation and growth is called synaptogenesis. Thanks to the work of Peter Huttchenlocker at the University of Chicago and many other scientists who have studied the brains of cadavers, we now have

solid data on the rates of synaptogenesis in various regions of the cerebral cortex in humans of all ages.

At birth the human brain is immature. New neurons are still forming and growing, synapses are multiplying, and myelin formation is far from complete (myelin is the insulating covering on nerve fibers that increases the speed of the electrical conduction of the neurons). The newborn brain is about one-fourth the size it will eventually reach, and most of its growth occurs in the first year of life. Most synapses connect to other neurons at terminals on dendrites. So the growth and elaboration of dendrites are closely associated with increased numbers of synapses.

At birth most of the connections, including dendritic growth and branching and synapse formation, are missing. Much of the growth occurs during the first year or so of life. At the end of this period the total number of synapses in the infant brain is twice that of an adult’s. Following this early period of exuberant growth, there is a slow and prolonged period of elimination and pruning of excess synaptic contacts, which is not completed until late adolescence in many cortical areas. But new synapses, and indeed new neurons, form throughout life in some regions of the brain.

The growth of and decline in the numbers of synapses in various regions of the cerebral cortex are closely associated with critical periods in development. This is strikingly evident in the visual area of the human cortex. The total number of synapses is very low at birth, skyrockets to a peak at about 6 to 12 months of age, and then gradually declines until about age 10, after which it is relatively stable until old age. The period of maximum synaptogenesis closely corresponds with the period of maximum plasticity and maximum sensitivity to impairment in the visual system.

Hearing in infants is poor at birth but reaches virtual adult competence by about six months. This is closely paralleled by the growth of synapses in the human primary auditory cortex (Heschl’s gyrus). By six months of age there has been a massive increase in the number of synapses, which approach adult levels by about three years of age and then gradually decline until about age 10.

Some musicians have perfect pitch; they can immediately identify the pitch of any note. The results of one study showed that all musicians who had perfect pitch began musical training before age 6. Those who began training after the age of 6 did not have perfect pitch, corresponding to the decline in synaptogenesis in the auditory cortex from age 3 to age 10. It is not known whether there are individuals with perfect pitch who did not begin musical training until after age 6 or so. But, of course, we must be careful about cause and effect in findings like this. Perhaps people with perfect pitch are so musically gifted that they naturally seek out musical training early.

The development of hearing is a clear example of a critical period. Many cases of early deafness can be helped by a device implanted in the inner ear, the cochlea. This device codes sounds into electrical impulses delivered to the cochlea to activate auditory nerve fibers.The development of cochlear implants has made it possible for some individuals who had been deaf from birth to learn to hear and understand speech. Interestingly, the implants are most effective if implanted before the age of 3. They decrease in effectiveness when implanted at older ages. Whether this is due to decreased plasticity in the auditory cortex or in the language areas of the cortex is not known. The growth of synapses in the language areas of the cortex also corresponds closely to the development of language functions.

As discussed earlier, infants show a dramatic increase in their ability to solve delayed response and A, not B, tasks from 7 to about 12 months of age and beyond. The prefrontal cortex, the region critical for these tasks, shows a rapid increase in the number of synapses from birth to one year of age and then remains high until at least age 5, declining in adolescence.

Parents have always been interested in ways to improve their child’s learning abilities and performance in school. A recent

popular fad is the so-called Mozart effect. Stores now sell Mozart CDs for young children to “enhance” their intelligence. This fad derived from studies on college students in which it was reported that listening to a Mozart sonata enhanced their subsequent performance on spatial-temporal IQ-like tasks. What was also reported but widely overlooked in the popular press is that the Mozart effect lasted only about 10 minutes. There is, in fact, no evidence to support such an effect (see Box 3-3).

There is some evidence that musical training may enhance performance on some tests of mental abilities, but the effects are not great. To some extent, this is another chicken-and-egg problem. Does musical training enhance performance on the tests or do children who take musical training exhibit enhanced performance on the tests because of their particular interests and abilities? But early musical training perhaps increases the possibility the child will have perfect pitch, as we noted, presumably enhancing later musical capabilities. Actually, one recent and carefully done study found a greater increase in the IQ scores of children after taking music lessons than after taking drama or no lessons.

Raising animals in a rich environment can result in increased brain tissue and improved performance on memory tests. Much of this work has been done with rats. The “rich” rat environment involved raising rats in social groups in large cages with exercise wheels, toys, and climbing terrain. Control “poor” rats were raised individually in standard laboratory cages without the stimulating objects the rich rats had. Both the rich and the poor rats were kept clean and given sufficient food and water. Results of these studies were striking: Rich rats had a substantially thicker cerebral cortex, the highest region of the brain and the substrate of cognition, with many more synaptic connections, than the poor rats. They also learned to run mazes better.

The popular press made much of the enriched-environment

effect, and commercial devices were developed to “enrich” the environment of babies’ cribs with bells, whistles, and moving objects. It turns out that the wrong conclusion was drawn from the animal literature. The data were clear; the rich rats had more developed brains than the poor rats. But when wild rats were examined, their brains were like those of the rich laboratory rats. Indeed, in one study a large area was fenced off outside the psychology building at the University of California at Berkeley, and rats were raised in this semiwild environment. They also developed rich rat brains. It seems that it was the poor rats whose brains developed abnormally, from being raised in isolation without the stimulation of normal rat life.

There are parallels in tragic cases where children have been raised in isolation. A California girl spent most of her first 14 years of life tied to a chair in a small room. She was never able to acquire normal language. There are also reports of impaired infant development in orphanages in Eastern Europe where the infants were left alone day and night except for feedings and changings.

What are the best things that concerned parents can do to enhance the cognitive development of their infants? The experts tell us to do what comes naturally. Talk to them, play with them, make funny faces, pay attention to them—spend time interacting with them. Mozart and noisy mobiles don’t really help. It is people they want to interact with.