Deric Bownds

Chapter 6

Plastic Mind

We now leave general descriptions of the evolution of hominid minds, psychology, and culture to turn to a more detailed account of how those minds develop, the next theme we need in assembling a description of the biology of our minds. The first part of this book has outlined the increasing sophistication and complexity of animal and human minds over evolutionary time, focusing mainly on the adult forms that mate and pass their genes on to future generations. These minds and brains start out as a mere blueprint, or template, provided by their genetic history, and then their adult forms arise only through a complicated sequence of developmental steps. To change the final product, the building blocks and/or the steps in their construction have to be changed.

It is remarkable how labile, or plastic, the construction of brains and minds can be. It varies because the intimate details depend on the actual environment provided by other cells in the developing embryo, on interactions between embryo and mother, and finally on interactions of the brains and bodies of growing children with their physical and social environment. Each of us has a unique brain whose formation is directed by our early surroundings, language, and culture. The structures described below form in a stereotyped and fairly constant way during embryonic development, but after birth they continue to grow and can be molded and shaped by their interaction with the outside world. This ensemble of evolutionary and individual developmental history forms the adult "society of mind" that is the subject of Part III of this book.

This chapter offers some evolutionary arguments for why brain development has become more malleable in more complicated animals, and it also describes the amazing plasticity that we can observe in developing as well as adult brains. This plasticity underlies the effects that both social experience and sexual differentiation have on brain processes, and it is required for the changes in nerve connections that support memory and learning. Our ability to remember facts and procedures---perhaps the most striking example of our brain plasticity---is considered in this chapter. This type of cognitive neuroscience is emphasized later, in Part III, but there are two reasons for discussing it briefly here. First, similar changes in nerve cell synapses seem to occur during nerve development and during the storage of some kinds of memories. Second, the human "selves" whose construction we consider in Chapter 7 are built from memories.

An Outline of Brain Development

Development consists of an intricate sequence of cell divisions and migrations that create local environments whose different chemistries turn different sets of genes on or off. Early development is dominated by axial information systems (dorsal-ventral, head-tail) that span the whole embryo and cause localized gene actions in different parts of the growing embryo. Events at this stage show considerable developmental flexibility and thus are a target for adaptive changes during evolution. Midway through development, an intricate interconnectivity occurs between elements that will later come to represent separate modules such as organ primordia (of brain, heart, lung, liver, and so on). This developmental path restricts the flexibility of the system, because changes in one module can alter many others. By late development, however, more developmental flexibility is seen again because the body has been highly modularized. At this point, adaptive changes in a liver, for example, don't have to cause changes in nervous tissue. Similarly, a genetic change that improves the effectiveness of a brain module doesn't have to affect the kidney.



The enormity of the problem we face us when we try to understand the development of our brain can be illustrated with some simple numbers. Our brains contain at least 1010 neurons with an average of 104 nerve connections or synapses to each, yet the human genome contains less than 105--6 genes. Thus there is no way that the position of each nerve cell, much less each of the many connections made by that cell, could be uniquely specified by a single gene. The problem is compounded by the need for instructions to form all of the other organ systems of our bodies.


We share with all vertebrates an embryo that in its early stages is a flat plate made up of inner, middle, and outer layers: endoderm, mesoderm, and ectoderm. Figure 6-1(a--c) shows cross sections of an early embryo that indicate the way these layers fold to form the neural tube from which the central nervous system develops. Figure 6-1(d) shows a view of the neural tube from above, cut horizontally so that you can see its interior. You can imagine this neural tube as a sausage-shaped balloon filled with water (cerebrospinal fluid, or CSF). Three swellings at the front end of the balloon form forebrain, midbrain, and hindbrain (you might find it useful to refer to Figure 3-2, which offers a side view of these structures). The forebrain further subdivides into a central portion (diencephalon) and two vesicles (telencephalon) that sprout from it to form eventually the two cerebral hemispheres that constitute our cerebral cortex. (Throughout this bulging and folding process, the central CSF-containing channels are maintained and become the ventricular system of the adult brain.) Two smaller vesicles that sprout from the diencephalon become the optic nerve and retina. The diencephalon finally differentiates into the thalamus and hypothalamus. The thalamus is the gateway for information going to and from the cerebral cortex. The telencephalon folds onto itself to form the hippocampus and olfactory cortex on its internal and bottom surfaces. These structures are two cell layers thick. Then, only in mammals, the top external part of the telencephalon forms the "new cortex," or neocortex, which has six layers of nerve cells. When people speak of the cerebral cortex, they usually mean the neocortex, even though "cortex" properly refers to all of the structures derived from the telencephalon. Figure 6-2 provides simple summary drawings of the developing human brain, from the stage of the neural tube shown in Figure 6-1 to the formation of the neocortex.

Figure 6-1

Formation of the neural tube, neural crest, and brain vesicles. (a) The three basic cell types of the early embryo. The endoderm generates many of our internal organs, the mesoderm generates our bones and muscles, and the nervous system and skin come entirely from ectoderm. A part of the ectoderm gives rise to the nervous system by forming a flat sheet of cells called the neural plate. (b, c) A groove forms in this plate, and the walls of the groove join to form the neural tube. The whole central nervous system (CNS), consisting of spinal cord and brain, develops from this tube. As this happens, some neural ectoderm pinches off just to the sides of the neural tube. This is the neural crest that gives rise to the peripheral nervous system (PNS) described in Chapter 3. (d) A view of the neural tube from above, showing forebrain, midbrain, and hindbrain. Adapted from Figures 7.8 and 7.9 in Bear et al. Neuroscience : Exploring the Brain.

Figure 6-2

Changes in the structure of the developing human brain. The first three drawings, which indicate the swellings of the neural tube during the formation of forebrain, midbrain, and hindbrain, are shown enlarged with respect to the later stages. These later stages show the cerebral hemispheres overgrowing the midbrain and hindbrain, as well as partly obscuring the cerebellum.

These brain structures, along with those described in Chapter 3, provide us with a basic hard-wired ensemble of automatic, reflexive, and instinctive mechanisms shaped by genetic trial and error over many generations. If we put our hand on a hot stove, we need not think about escaping as withdrawal reflexes take over, jerking it out of harm's way. We share with other vertebrates hardwiring for fighting, fleeing, sex, aggression, sleeping, and eating. Such hardwiring is centered in the primitive mammalian brain, the limbic system, and in such other structures as the brainstem, thalamus, and hypothalamus. The inferior temporal cortex of 5-week-old monkeys contains neurons that respond to faces and complex geometrical patterns. Our short-term and longer-term behaviors can be under instinctive control. We act out daily, monthly, and annual cycles as we sleep and wake, produce fertile eggs once a month, and undergo seasonal changes in mood and affect.



We are born with brains wired to attend longer to stimuli that look like human faces than to other stimuli, and our own faces register appropriate emotional responses.


There is a problem, however, in relying only on hardwired, or inflexible, behavioral routines that have been crafted over millions of years of evolution. Just as they evolved slowly, they can change only over many generations, as genes and developmental mechanisms are altered, and numerous features of the environment change much faster than that. The solution that evolved as brains became more complicated during evolution was to make their formation more plastic---patterned by the actual surroundings in which they must function. This permits human adaptation to an environment to occur on the time scale of a single life span, rather than the many generations that would be required for genetic selection within the population's gene pool.

Origins of Plasticity

Having a nervous system whose design is completely predetermined by a fixed set of genes and developmental sequences has one major drawback. If the environment changes suddenly, the animal may be left with inflexible behaviors appropriate to the environment that shaped its evolution but quite dysfunctional under the new conditions. A solution, or adaptation, to this problem would be to have genes specify not one fixed nervous system, but rather a starting array of options that can be tested for their appropriateness as developing nerve connections and pathways are used in the real world. 1 Organisms with the ability to select among developmental options should have an advantage over those that cannot easily adapt to changing circumstances. Organisms that have more plasticity should more effectively move toward a desirable adult form---and so reproduce better. Their technique for achieving more plasticity should then be passed on. A Darwin Machine mechanism of the sort described in Chapter 2---preferential duplication of forms that work best---could act on a developmental, rather than a generational, time scale to select and amplify growing nerve pathways that work, at the expense of those that do not work as well.

These options appear to have been taken during the course of evolution. Pathways and competencies that do not prove to be relevant during development do not mature. 2 In the case of feral human infants, raised by animals in the wild, normal language and bipedal locomotion do not develop. 3 The close matching of their behavior with that of their surrogate animal parents shows how little distinctively human behavior is obligatory in the absence of a normal human cultural environment. In the right selective environment, however, the brains of these children could have become the brains of airplane pilots. Similarly, songbirds have a genetically programmed template that specifies a range of sounds that can be learned, but the song actually performed is usually learned from other birds. The rule seems to be "use it or never develop it," a variant on "use it or lose it."



In more complicated animals, interaction between the environment and neural growth leads to a flexible, constructive kind of learning that minimizes the need for detailed prespecification of structures. Selection dictates which of the many possible pathways are chosen and become permanent.


As birds learn to sing, and as human babies learn to walk and produce language, there is room for variation and change, but after this learning period, the design laid down becomes relatively permanent. Thereafter, further learning is much more difficult. Once we have learned to walk or ride a bicycle, we usually don't forget how to do it. But learning to ride a bicycle after adolescence is much more difficult than when we are under 10 years old. Similarly, a young human infant is a polyglot, able to utter the sounds of any human language. But much of this potential range is lost as soon as the language actually learned fixes the design of the infant's language "pathways."

During brain growth there is a constant sorting and juggling of nerve cells and connections. Those that make a match with their environment thrive, and the others wither. Imagine what it would be like to move in a visual world that consisted mainly of vertical lines and contours---to be unable to see the horizon. Experiments on cats and monkeys that we will soon describe suggest that this might happen to us if we were shown only vertical stimuli for a year or so after our birth. Cells that process vertical stimuli would grow to prevail over cells that deal with horizontal lines, because our visual brains wire themselves to be most responsive to the stimuli actually encountered early in life. There are critical periods during which not only our visual brains but also many other areas, including our social brains, are wired to conform to and accommodate the external environment. If that early environment is degraded, so is the development of our mental capacities.

A Hierarchy of Developmental Circuits from Innate to Learned

We would expect innate circuits that shape basic survival behaviors, such as approach/avoidance, to influence circuits that can be modified by experience. Clusters of nerve cells, present deep in our brainstem and common to all mammals, distribute neurotransmitters such as serotonin, acetylcholine, dopamine, and norepinephrine to large areas of our cortex and influence both the growth of the brain during development and its behavior, as well as plasticity, in the adult. 4 These older innate circuits are responsive to what is happening to more modern sectors of the brain and can signal the "goodness" or "badness" of different situations. They are crucial components of the chemistry underlying motivation. They react by influencing how the rest of the brain reacts, and how it is shaped, so that it can become wired in the way that most effectively supports survival. For example, it is most relevant to grow a brain that is predisposed to register associations between illness and taste (rather than between illness and place, sound, or temperature) and then, through learning, determine which foods with which tastes actually lead to poisoning or illness.



A human child cannot be genetically equipped with knowledge of who will be nurturing and who dangerous, but it can be "told" by its genes that facial characteristics are important indicators of this distinction.


What we are describing is a hierarchy of primary (genetic) and secondary (developmental) problem-solving devices. The primary genetic mechanism acts on a generational time scale, and the secondary process comes into play during postnatal development. This is the distinction made by references to ultimate and proximal causes of human form and behavior. The secondary, or proximal, process involves the growth and reinforcement of neuronal pathways and connections that work best, as well as the atrophy of those that are not found useful. This reinforcement comes not only from our learning to move about in the immediate physical world but also from cultural and social influences that guide our behavior and development.

Human genes and environments interact on at least four levels in the construction of brains and selves: (1) Genes instruct, and are instructed by, the complex internal environment of the fertilized egg and by subsequent interactions between groups of dividing cells and tissues. (2) They can be regulated by hormonal and other interactions between the developing fetus and the maternal chemistry of the uterus. (3) Gene expression after birth in infant humans, as in other animals, is patterned by the development of sensing and acting routines appropriate to the species' typical physical environment. (4) Individual social interactions among infant, caretaker, and peers then pattern the learning of more complicated behaviors. Thus the influence of genes is expressed through a long, tortuous, and indirect pathway.

The Wiring of Developing Brains

We now need to consider what is known about brain developmental processes, and the ways in which they are shaped by environmental changes. As we have noted, the brain originates as a swelling at the head end of the spinal cord. Then a series of bulges and folds give rise to the eyes and the lobes of the cortex, as nerve cell bodies migrate to their final positions. Axons then grow out to make interconnections between appropriate groups of nerve cells. A variety of mechanisms guide the growth of axons as they find their targets in the brain. One mechanism is derivative of the chemotaxis described in Chapter 2. A nerve cell body, or rather the growing tip (called a growth cone) of its axon, can be attracted to move toward a target that is releasing a particular growth factor to "call" the growth cone (Figure 6-3). Physical cues, such as those that promote grow along grooves or alongside other nerve cells, are also used. Some cells serve as guideposts for the turnings or migrations of others. Ongoing neural activity---the firing of nerve signals---is crucial in both the growing cells and their potential targets.

Figure 6-3

The growth cone. If we look in the microscope at the growing tip of an axon, we observe what appears to be very intelligent purpose, just as in bacteria exhibiting chemotaxis. Delicate little finger-like projections wave about and literally reach out to touch and taste other cells in the growth path of the axons. The growth cone is repelled by some such cells and attracted by others.

When a final target is encountered, as, for instance, when a motor nerve axon finds the muscle that it will control, the searching "fingers" retract, and both nerve and muscle respond by making the pre- and postsynaptic structures that are required for a synapse. A growing axon may make many temporary connections with other cells, acting as though it is testing them for appropriateness, before moving on to its final position. An important factor appears to be the functional appropriateness of the tentative connections being made and remade. Are they useful in correlating the input and output of the whole animal? Are they appropriate for the environment in which the organism is functioning? 5

Pathways to the Cerebral Cortex

To consider what is happening in more detail, we need a simple description of how sensory cells in the eye, ear, or skin ultimately get their information to the cortex. Virtually all of them connect to cells that pass through the thalamus, which has separate nuclei, or groups of nerve cells, devoted to different kinds of input. The lateral geniculate nucleus carries visual input to the visual cortex, the medial geniculate nucleus carries sound information to the temporal cortex, the ventral posterior lateral nucleus carries somatosensory information from the body to the parietal cortex, and so on (Figure 6-4). The flow of information is bidirectional, for the cortex also sends information back to the thalamus. There is strong evidence that axon pathfinding between major structures such as the eye and the thalamus, and the thalamus and the cortex, is under the control of genes. 6 This pathfinding also appears to require constant nerve activity.

Figure 6-4

Pathways from peripheral sense organs enter lower brain areas and then pass through different nuclei of the thalamus on their way to areas of the neocortex that are specialized for different sensory modalities.

The cortex develops in two directions, radially and tangentially. In the radial direction are six layers of nerve cell bodies; layer 1 is on the top, and layer 6 on the bottom. Information coming into the cortex goes primarily to layer 4, the "in box" of the cortex. Cells of layers 5 and 6 send information back out to subcortical brain regions (functioning as the "out box" of the cortex), and cells in layers 1, 2, and 3 connect different cortical regions to each other, sort of like an interoffice network for sending memoranda back and forth. As we move across different regions of the cortex tangentially, these six layers might be found to be supporting vision, hearing, commanding movements, and so on. Their detailed structure varies in accordance with their different roles.



It is currently thought that the early cortex is not regionally specified and that the specialized areas form as axons from the thalamus grow to them, bringing different kinds of stimulation from the external world. It is almost as though various sensory surfaces---with their specialized receptor cells sensing light, sound, pressure, and so on---impose themselves onto the brainstem, then onto the thalamus, and finally onto the cortex itself. 7


Functional Plasticity in the Formation of Cortical Areas

During the formation of these thalamus-to-cortex pathways, dramatic plasticity at the cortical level can be observed, and specialized cortical regions expand, contract, or take on functions other than their normal ones if unusual circumstances are encountered. A number of experiments of the sort shown in Figure 6-5 demonstrate that areas of cortex can switch from one sensory mode to another. If one sensory modality is blocked at birth, areas of cortex that normally serve it can be invaded by other sensory modes. Thus, in cats deprived of visual input, the parts of the parietal cortex that are usually predominantly visual are taken over by auditory and somatosensory input, and these animals show greater sound and tactile discrimination than normal animals. 8 Another striking observation is that in some congenitally blind humans, the visual cortex is activated by reading Braille. This shows that input from the hand that normally goes primarily to the somatosensory cortex can also be directed to visual areas in the occipital lobe. 9 One model is that at birth the human cortex has synesthesia, mixing of the senses, and that the senses become clearly separate only after about 4 months. Before this, in human babies, electrical responses to spoken language are recorded not just over the auditory regions of the temporal cortex but over the visual cortex as well. In the small number of human adults who have synesthesia, such as those who see colors when they hear sounds, the normal separation of the senses apparently has not occurred. 10

Figure 6-5

Switching cortical function from one sensory mode to another. (a) Most of the input from the eye to the brain goes through the lateral geniculate nucleus of the thalamus to the visual cortex located at the back of the brain, but early in cortical development, a few connections are also made to other parts of the thalamus (dashed line); these connections normally are eliminated later in development. Similarly, sensations at the body surface are sent mainly to the ventral posterior lateral nucleus of the thalamus and then to somatosensory cortex in the parietal lobe located in front of the occipital lobe, but again, some connections are made to other parts of the thalamus and are later eliminated. (b) The experiment, begun early in development, involves placing lesions in the lateral geniculate nucleus and the visual cortex, as well as blocking input from the body surface to the thalamus (as shown by the X's). Now input from the eye has lost its normal targets but still has the few connections made to other areas of the thalamus (dashed line, top). Also, the ventral posterior lateral (VPL) nucleus of the thalamus and the somatosensory cortex are without their normal input. What is observed is that the visual connections to the VPL nucleus (shown as the dashed line in part a) can multiply and become a permanent visual input (solid line) directed via the VPL nucleus to somatosensory cortex. 11 This somatosensory cortex takes on the function of visual cortex. In the same sort of experiment, visual information can also be rerouted to the auditory cortex, which then converts from representing the pitch and frequency of sound to representing the locations of objects in the external world. 12

Other types of experiments further demonstrate cortical plasticity. In the newborn rodent cortex, visual areas can be transplanted to somatosensory areas take on somatosensory function, and the reverse transfer also works. Auditory and visual cortex transplanted onto each other also change their function. However, the transplants never function in exactly the same way as the normal structures, which suggests that they may have already been patterned by some thalamic input. Most of these examples involve primary sensory areas, but a striking plasticity in higher cortical functions is observed when, in newborn monkeys, the portion of the temporal cortex responsible for visual object recognition (see Chapter 8) is removed from both left and right cortexes. This function moves a considerable distance to the part of the parietal lobe that normally deals with the movement and orientation of objects in space. 13



Transplant experiments have demonstrated cortical plasticity by showing that a cortical area that normally serves one sensory modality can be transplanted and take on the functions of another.


Plasticity in Forming the Visual Cortex

Studies on the development of the visual system, particularly the wiring of the cells in the occipital cortex at the back of the brain, have provided several fascinating examples of how labile the wiring-up process is and how susceptible it is to being changed by distortions in the environment. Similar phenomena are found in all parts of the brain, but the visual system has proved so accessible to experimentation that many basic and pioneering studies have used it.

The primary visual cortex in the occipital lobe of one hemisphere receives input from both eyes. If we move an electrode along the surface of this cortex in adult cats or monkeys, we find alternating rows of cells that respond mainly to one eye or to the other. Studies on kittens and monkeys have shown that cells of the primary visual cortex are driven by both eyes at birth, but one eye tends to dominate. During the first 6 weeks of life, as the young animals learn to move about and use their visual world to guide body movement, much clearer domains of cortex are formed that respond mainly to one or the other eye. However, if visual information from one eye is blocked during a critical period after birth (4--6 weeks in the cat), either by putting an opaque contact lens in the eye or by temporarily sewing the eyelids shut, the normal adult pattern is changed. Input from the eye that remains open now expands to occupy a larger area of cortical surface, while the area occupied by input from the closed eye contracts.



There seems to be an active competition between the two eyes for cortical targets. Diminished input from one eye tells the cortex that the active eye is more important and deserves more connections. In other words: Use it or lose it!


The playing out of this competition requires that the cortical targets themselves also be actively communicating. If their nerve signals are silenced by chemical inhibitors, inputs from the closed eye expand. 14 There appears to be a push-pull mechanism of synaptic plasticity: When target cells are working (are giving postsynaptic responses), the active eye has the advantage just described, but without postsynaptic activity, the inactive inputs have an advantage. 15 One possible explanation is that activity from the active eye causes its target cells in the cortex to release molecules (neurotrophins) that stimulate proliferation of the active inputs. 16 There appear to be many ways in which the number and strength of synaptic connections between cells can be altered. Figure 6-6 outlines some of these ways in a very simplified form.

Figure 6-6

Ways in which altering nerve pathways from different sources, such as suppressing input to the visual cortex from the left but not the right eye, might influence the competition of the two eyes for targets in the visual cortex. (a) The normal balance of input from left and right eyes, via the lateral geniculate nucleus. (The example is simplified to show only a few synaptic connections between cells. In reality, the axon from a cell in the lateral geniculate nucleus might branch to make synapses with hundreds to thousands of target cells in the cortex.) After suppression of input from the left eye, increased activity might improve the effectiveness of existing connections (b), disuse might lead to atrophy of the unused pathway (c), and axons of the more active pathway might sprout branches that form new synapses (d).

The story becomes even more interesting as more subtle properties of visual cortical cells are examined. Most of these cells are sensitive to the orientation of a visual stimulus, and cells that respond to all directions of the compass are found at birth. If, however, a young cat is exposed to an environment that contains mainly vertical bars during the critical period when connections are formed, then cells sensitive to vertical bars predominate in the adult visual cortex. 17 As is the case with the ocular dominance areas, a crude, genetically specified pattern present at birth can be profoundly altered by the visual experience actually encountered during the postnatal growth and development of the visual system.

Experience Guides the Formation of Successful Connections

Two processes appear to be central in wiring up our brains: the pruning away of initial, exuberant overgrowth of axons into their target areas, and elaboration of the complexity of the connections that remain. 18 The first process has been likened to a Darwinian competition, including replication of pathways, testing of their appropriateness, and enhanced survival of those copies that work best. This selective process also probably works to test the results of the invention, or construction, of the complex circuits used to solve the problems specific to different kinds of information processing (visual, auditory, somatosensory, and motor) throughout an extended period of brain maturation. An example of a selection process---one that might guide both the formation of new connections and the elimination of unnecessary ones---might be the effectiveness of getting food from hand to mouth, which requires elaborate circuits for hand-eye coordination. We are born with hardwiring set to do it in a primitive and clumsy way. The pathways then grow and are refined through testing. Another example is offered by development of the visual system. We are born with cortical cells that detect stimuli of all orientations, but the ability of human adults and many terrestrial animals to detect vertically or horizontally oriented contours is superior to their ability to perceive oblique angles. The prevalence of vertical and horizontal orientations in natural indoor and outdoor settings apparently leads to increasing stimulation and growth of visual cortical cells that specialize in those orientations. A further example of how nerve cell wiring is influenced by functional testing is outlined in the experiment shown in Figure 6-7.



What a brain is apparently doing during the wiring up of its nerve cells is testing the functional appropriateness of connections that are present or are being formed. Do those connections work in guiding behavior in the real world?


Figure 6-7

The sorting out of appropriate nerve connections. An example of the determination of nerve connections by functional testing comes from experiments in which a bundle of sensory nerves running to the hand of a young monkey is severed. (a) Before the nerve to the hand is cut, nerves in the fingers ultimately report, via the thalamus, to adjacent parts of the somatosensory cortex: finger 1 goes to area 1 of thalamus and then to area 1 of the brain, and so on. (We made these numbers up, just to make our point. They don't refer to existing anatomical designations, used by professional neuroanatomists, in thalamus or cortex.) If the bundle of nerves to the hand is cut (dashed line), its distal end (the end farther from the attachment) degenerates, but then its axons can be regenerated from the ends of the nerves above the cut that are still connected to their cell bodies. (b) These axons grow back to connect to the hand in a disordered fashion. Now the axons reporting from finger 1 to the thalamus might be those that used to report from finger 4, and so on, as shown by the crossed connections on the right. Adjacent areas of the cortex should no longer represent adjacent fingers. Quite to the contrary, however, the representation of this reinnervated hand in the primary somatosensory cortex develops in a orderly fashion. The developing brain is able to sort things out and create order, despite the fact that its sensory inputs were temporarily scrambled. This suggests that the pattern of sensory stimulation from the fingers is the most crucial cue for establishing neighbor relationships in the part of the somatosensory cortex that represents the hand, where adjacent areas of the cortex represent adjacent fingers. 19 The nerves from adjacent fingers will be active together more frequently than the nerves from the first and fifth fingers. Cells that fire together are then more likely to wire together! Synapses that use the main excitatory transmitter of the brain, glutamic acid, appear to play a central role in these examples of plasticity.

It is not clear how much of the circuitry for higher mental functions is determined beforehand by genetic instructions, and how much is discovered by each growing brain as it bootstraps its way to maturity, growing through a series of stages of increasing complexity, each of which makes use of the solutions to problems devised in earlier stages. From what we now know about the molecular details of how nerve cells work, we can see how such factors as the speed of nerve signals and the density of connections between nerve cells might be directly controlled by genes, but we have very little idea how higher functions such as language might be linked to genes. The prevailing view is that the linkages are very indirect, and the examples we have given show that developing cortical areas are very flexible in the functional assignments they will accept.

Perhaps the most amazing example of this comes from clinical observations on infants with intractable epilepsy who have undergone removal of an entire cerebral hemisphere. One might expect that this would completely block normal development, during which competing language-related and other functions displace each other into opposite cerebral hemispheres. (Thus the left hemisphere normally has a skeletal motor bias and is optimized for speed, syntax, and visual details, whereas the right hemisphere has a visceral motor bias and deals with large time domains, the prosody component of language, and spatial relationship.) In defiance of all such predictions, however, in children with one hemisphere removed, all of these operations can collect and develop in the hemisphere that remains.

Adult Brains Can Change Their Nerve Connections

Until recently, it was assumed that brain cells and wiring change very little after puberty, when the final form of the adult brain has taken shape. Thus many studies emphasized the slow decay in cognitive function that frequently occurs in aging: loss of openness to new experiences and so on. We still have to explain, however, why some people can show striking recovery from brain damage 20 and why it is that if we put on eyeglasses that magnify or shrink the visual world we perceive, our ocular reflexes and hand-eye coordination soon adapt as though nothing had happened. Changes in the nerve cells that regulate these reflexes in monkeys have been monitored, and brain imaging experiments in humans have shown the posterior parietal cortex to be involved. 21 Even more striking, we can wear glasses that either turn our visual world upside down or reverse left and right, and within a few days our perceptions change so that we can function almost normally, even throwing and catching a ball. (After we take off the glasses, it takes a day or more for appropriate function to be restored.) It is difficult to reconcile these observations with the assumption of a fixed adult structure.



Consider the intriguing case of the woman who spoke both Italian and English and went to work as a United Nations translator. Before she began her new job, her left hemisphere, the normal locus of language, was used for both languages. After she had been in the job a few months, Italian switched to the right hemisphere! 22


Expansion and Contraction of Cortical Areas

A large number of direct experiments have now demonstrated that the adult cortex can remodel itself within localized areas. 23 There is debate over whether these changes involve the actual growth of new axons and connections over long distances, or whether existing horizontal connections between cortical regions, normally silent, become active. 24 Probably both events occur. The most detailed work has been done in monkeys and entails making direct electrical recordings from the somatosensory cortex. More recently, imaging techniques have permitted similar observations in human brains. 25 When performance of a particular sensory or motor activity is emphasized (such as reading Braille with one finger or developing a repetitive and skilled motor activity with one finger), the portion of sensory and motor cortex devoted to that activity expands. Conversely, when input from a digit is suppressed, for example by taping one digit to the next, the area of cortex that usually represents that digit is appropriated by adjacent digits. A similar phenomenon occurs in the visual system. When a small area of the retina is damaged, the area of visual cortex that represents that area is invaded by adjacent retinal areas, a process called "filling in." 26 This is what normally happens with the blind spot on our retinas, a small area where photoreceptors have been pushed aside by optic nerve fibers gathering to project into the brain.

Brain changes that occur during filling in or shifts in hand-eye coordination happen in only 10--30 seconds---much too fast and far for new connections to grow. Changes that occur over periods of days or weeks as new motor or sensory skills are acquired can involve distances of more than a centimeter of cortex and may involve new growth of axons. The ability of a cortical area that represents one of our fingers to either expand or contract probably reflects a continuous competitive give and take between neurons. If one neuron becomes much more active because it is needed for some skilled task, it can begin to encroach on areas used by other neurons. If it becomes more silent because fewer cells are talking to it, its sphere of influence contracts. These changes might be of the sort shown in Figure 6-6, altering the strength of existing synapses and/or forming new synapses. Such plasticity enables us to adapt to and learn new situations.

A change in somatotopic organization of the hand area in humans has been observed by means of magnetoencephalography after surgical separation of webbed fingers (syndactyly). 27 The presurgical maps displayed shrunken and disordered hand representations. Within weeks after surgery, the cortex had reorganized to reflect the functional status of the newly separated digits. This finding is consistent with the results of the experiment shown in Figure 6-7. The somatosensory cortex appears to use the timing of arriving information to create spatial representations of the fingers. Because input from one finger is usually synchronous and separate from input from the other fingers, the cortical representation of that finger is separate. 28

Blind people who read Braille literally reconfigure sensory cortex for the fingertips. The area devoted to the finger used for reading is several times larger than the corresponding area in sighted people. The expansion of this area as Braille is learned can be followed in imaging experiments. String players exhibit an increased cortical representation of fingers of the left hand. 29 During the training of fingers to tap out a particular sequence, the area of primary motor cortex activated by performing the sequence expands, and the changes persist for several months. 30 In one experiment, one group of adults was given a daily five-finger exercise to play, up and down a keyboard, and another group was told just to press the keys randomly. The brain's representation of hand muscles in the subjects who learned the piano exercise increased threefold compared with subjects who randomly operated the keyboard. The biggest surprise came from subjects who were taught the exercise but then told just to rehearse it mentally, not manually, while looking at the keyboard. This mental exercise achieved the same result as manual practice. 31



Mental rehearsal of a finger exercise can increase the area of cortex that represents the hand just as effectively as actual manual rehearsal can. This may be why imagining movement as a way of learning and refining it has been developed into therapies for rehabilitation and techniques for training of dancers and athletes. 32


Cortical Plasticity and the Phantom Limb Phenomenon.

Cortical plasticity appears to play a central role in "phantom limb" phenomena observed after amputation of an upper limb. 33 Patients experience sensations such as itch or pain in their no-longer-existent limb. The invasion of an area of somatosensory cortex that had been responding to a now-absent limb may underlie the sensations involved. 34 The amount of phantom limb pain correlates with the extent of neuronal reorganization in the somatosensory cortexes. 35 In some patients who have undergone upper limb amputation, the face area invades the area that had been responding to the limb. 36 (The area of cortex representing the face lies just next to the area representing the limb; see Figure 3-6.) Stimulation of the face is then experienced as sensation arising from the hand as well as from the face. A systematic, one-to-one matching is reported between specific regions on the face and individual digits of the hand. A drop of warm water allowed to trickle down the face is felt also as warm water trickling down the phantom limb. In the case of an amputation just above the elbow, stimulating either the face or the remaining upper arm can cause feeling "in the fingers." This is presumably because in the somatosensory body map, the hand area is flanked on one side by the face and on the other by the upper arm and shoulder.

Another striking example of plasticity after limb amputation is seen when a tall mirror is placed vertically on a table and perpendicular to the patient's chest, so that a mirror reflection of the normal hand is seen "superimposed" on the phantom. 37 Moving, touching, or vibrating the normal hand now causes a corresponding sensation in the phantom hand, but heat, cold, and pain are not transferred. This suggests that there is a normally latent link between brain areas in the right and left cerebral cortexes representing the two hands whose activity can be brought to awareness if visual input is provided. In some cases, viewing movement of the mirror image relieves chronic spasms and pain in the phantom limb, perhaps by providing sensory feedback for motor commands that usually register as pain because they lack such feedback.

Social Experience Can Alter Brain Structure

Given the plasticity of our sensory and motor cortex, it would seem reasonable to look for plasticity in parts of the brain that regulate our more complicated social and emotional behaviors. Here there are very little data, apart from studies on behavioral effects of impoverished early environments (these studies are mentioned in Chapter 7). We know much less about the brain pathways underlying these more complex behaviors than about more primary sensory and motor parts of the brain. Imaging studies have shown complex changes in brain activity correlated with schizophrenia and other psychopathologies, but interpretation of the data is in its infancy. We are not yet able, for both ethical and practical reasons, to induce a behavior change in human subjects (such as the change from a dominant to a submissive personality) and then examine the effect on the activity of nerve cells in the hypothalamus.



If the wiring of our adult sensory and motor cortex can be altered by training, what about the parts of our brain that control social and emotional behaviors?


Animal studies, however, provide evidence for a feedback loop between wiring of the adult brain and complicated social behaviors. It is very likely that increasingly sophisticated imaging techniques will yield similar information for humans. We would, in fact, expect that as evolution worked at the group level of animals to shape their social structure, brain plasticity appropriate for switching between alternative social roles would have been generated. This is presumably because role specialization increases the fitness of the whole group.

An example of such role specialization comes from studies of an African cichlid fish species in which aggressive males command large territories and keep other males at bay. These dominant males have enlarged nerve cell bodies in the region of the hypothalamus that regulates mating. If a dominant male is bullied by an even larger male, it becomes submissive and the hypothalamus cells shrink along with the size of the testes, which cease producing sperm. Dominant males have bright coloration unlike that of females and subordinate males, so they are more susceptible to predation. When a dominant male disappears, all the previously meek males rush into the area and start fighting. The winner then grows bigger and gains bright coloration, his hypothalamus cells and gonads enlarge, and sperm production begins. 38

Monitoring exactly how individual nerve cells are changing when social experience alters behavior is very difficult in vertebrate nervous systems, because these cells are extremely small and numerous, and their anatomy is not well understood. It has been possible, however, to take an analysis to the level of single nerve cells in studies on the invertebrate crayfish, whose nerve cells are much larger. A neurotransmitter (serotonin), which is known to be involved in the expression of aggression in many animals, acts differently on a neuron that controls the escape reflex in dominant and subordinate animals. This difference is due to a change in the postsynaptic receptor molecules on which serotonin acts. 39

Most relevant to humans, of course, are experiments on monkeys or apes. Numerous studies have documented neuroendocrine changes that correlate with dominance or submission in these animals. 40 Removal of a dominant male usually results in competition, followed by behavioral changes in the winner. In many mammals, high dominance status correlates with increased adrenocorticosteroid levels. 41 It is likely that similar changes occur in humans: If we change our group role either from "loser" to "winner" or from submissive to dominant, long-term changes in brain wiring probably take place. The increasingly powerful magnetic imaging techniques being developed may soon let us observe localized brain changes that correlate with such events.

Limits of Brain Plasticity

What are the limits of the plasticity of our brains? What can and what can't be changed? This is still unclear. The picture that has emerged so far is of critical periods during development when major pathways that regulate sensing and acting are laid down and can be modulated. Then, in adult brains, there is an ability to fine-tune these pathways if sensory or motor demands change. We don't yet have a link between direct observations made on our brains and more complicated aspects of our personality. Effective therapies exist for altering such things as sexual dysfunctions, panic, and manic or depressed moods, but other attributes (such as sexual orientation) are strongly resistant to change. 42 Perhaps the brain areas that regulate some behaviors are simply less plastic than the areas in charge of others.

Memory Is a Form of Brain Plasticity

An obvious kind of plasticity in our adult brains is revealed by our daily experience of remembering things---facts, faces, contexts, and manual skills. Each memory we can access reflects changes that have taken place in nerve connections in our brains. The remembering of phone numbers, facts, and places---which can be very rapid and may require only one exposure to the relevant item---seems likely to rely more on rapid changes in the strength of existing nerve connections. The slow learning and remembering of skills and procedures, such as your experience when you are trying to improve your tennis game, appear to involve cellular mechanisms very similar to those responsible for the plasticity discussed in previous sections. 43 Nerve cells actually make new connections. An avalanche of exciting new experiments are documenting changes in gene expression and synaptic molecules that correlate with memory formation, and several pharmaceutical companies are rushing to use this information to develop memory-enhancing drugs.

A moment's reflection on our own experience makes it clear that there are several different kinds of memory. These memory types are common to mice and men and have been studied in both species. Working memory is what we access for very recent events, where it is appropriate for recall of these events to lapse after a short time. It is what we use to remember a telephone number we have just looked up until we finish dialing it. There is an important reason why most telephone numbers have seven digits. We can hold in awareness, or short-term memory, approximately seven chunks (meaningful units) of discrete information. About five seconds per chunk is required to commit these units to longer-term memory (as in memorizing a phone number). 44 Regions of the prefrontal cortex appear to be most heavily involved in working memory, and different areas specialize in spatial locations, object identification, and verbal memory. 45



Think about how long you remember a phone number that you have just looked up and will not use again. This is the duration of your short-term memory, or working memory.


Long-Term Memory

The specifics of some of the events held in working memory, if they are perceived as important, might be passed on to one of the two major kinds of long-term memory displayed by animals: procedural memory and episodic memory. Procedural memories are the "recall" components of learned action patterns. They preserve general principles for action and ignore situation-specific details. Episodic memory refers to our ability to recall specific places, facts, number, or instances. Procedural memory and episodic memory involve different neural mechanisms. Birds can lose their songs (a procedural memory system) if lesioned in one part of the brain, and they can lose their ability to hide and relocate food (an episodic memory system) if lesioned in another. Some human amnesiacs can learn new motor skills (procedural memory) with no recall of having learned them (episodic memory). Other names are also used for these two kinds of memory: noncognitive versus cognitive, nondelarative versus declarative, implicit versus explicit. 46

We will deal here with only a few of the different memory systems. At least six such systems can be distinguished and associated with different brain structures: (1) Episodic memory (sometimes referred to as explicit memory), the conscious recall of facts and events, is associated with medial temporal lobe and diencephalic structures. (2) Working memory, which is discussed in Chapter 12, is associated with prefrontal cortex. (3) Priming, which relates perceptual and conceptual representations, is associated with occipital, temporal, and frontal cortex. (4) Motor skill learning is associated with the striatum. (5) Classical conditioning, wherein we learn relationships between perceptual stimuli and skeletal motor responses, is associated with the cerebellum. (6) Emotional conditioning, which relates perceptual stimuli and emotional responses, is associated with the amygdala. 47

Episodic, or explicit, memory consists of our ability to recall specific places, facts, numbers, or instances (as discussed above). 48 Some restrict episodic memory to the recall of events that happened to the individual and use semantic memory for recall of facts such as the location of cities. Several recent studies have suggested that the hippocampus and other parts of the brain play a role in the transfer of information from working, or short-term, memory to this kind of long-term memory. 49 If the hippocampus is damaged (by anoxia or mechanical shock) or is removed, the ability to form new memory is lost. Work with many patients has shown that it is possible for acquisition of long-term memory to be lost completely, while all other intellectual powers remain intact. These subjects can read a newspaper, understand all its contents, and 20 minutes later have no recall of having done so. The kind of memory required to learn both mental and physical skills (procedural memory) remains intact. Furthermore, there is no damage to the working memory used in immediate recall.

Patients with bilateral damage to the amygdala but with intact hippocampus cannot acquire conditioned autonomic responses (such as a skin conductance response to a color that has been paired with an unpleasant noise), but they can recall the pairing of the conditioned and unconditioned stimuli. Conversely, patients with intact amygdala but bilateral damage to the hippocampus learn the autonomic conditioned response without grasping the facts about the conditioning stimuli. 50 The storage of both explicit factual memories and emotional memories involves the medial temporal lobe as well as frontal and other areas. Imaging studies show that activity not just in the hippocampus portion of the medial temporal lobe, but also in a surrounding area called the parahippocampal cortex, correlates with whether novel items are remembered. Imaging studies further suggest that memory storage processes involving words are associated with enhanced activity in the left prefrontal cortex, whereas the right hemisphere becomes more active when visual images are involved. Memory retrieval is accompanied by enhancement of right prefrontal activity. Both imaging and brain lesion studies suggest that memory retrieval is under the executive control of the prefrontal cortex.



Damage to the hippocampus can abolish explicit recall, whereas damage to the amygdala can abolish emotional memory. The idea is that our hippocampus and adjacent structures link the separate parts of a long-term memory as they are formed, making it possible for all to be evoked when the memory is recalled. 51


A famous example of a complex memory occurs in Marcel Proust's novel Remembrance of Things Past. It is the flood of recollections stimulated by eating a small, rich pastry called a madeleine. 52 The event being recalled originally started with perception: the narrator's feeling and seeing his cup of tea, the spoonful he raised to his lips, along with other visual, olfactory, and motor qualities engaging different parts of the cortex. These would have passed their sensations to the hippocampus, and the hippocampus would then have orchestrated the formation of an associative net linking all the perceptions into a multifaceted experience. At a later time, reenacting just one of the events of this scene (eating the madeleine), like pulling a single strand of the net, would draw up behind it all the other strands with which it is connected. Figure 6-8 may help you visualize what may be going on here.

Providing enough cells in the hippocampus to support forming new memories may require the violation of a long-standing "law" about brain cells: that in the adult human brain they do not divide. Division of cells in the hippocampus has recently been observed for the first time, even in aged adults. 53 It has also been discovered that the decay of memory abilities observed in patients with Alzheimer's dementia correlates with MRI measurements showing a decrease in the size and volume of the hippocampus.

Figure 6-8

Model of the role of the hippocampus in forming memories. (a) Features of a scene such as taste, vision, and emotional feeling---processed in different areas of the cortex---all send signals to the hippocampus. (b) Networks of nerve cells in the hippocampus reassociate the separate features and send signals back to the cortex. 54 (c) As a result, the areas of the cortex become linked in such a way as to form a long-term memory that unites the separate features of the scene. 55

Researchers have measured the activity of the hippocampus in lab rats while the rats are exploring and learning a new environment. Simultaneous recordings from 100 or more nerve cells show that their firing patterns correlate with the position of the animal and that these firing patterns change when the animal's position changes. 56 Even more interesting, the correlated activity of the cells during sleep sometimes repeats the activity of those cells during earlier spatial exploration. 57 This may reflect memory consolidation of the exploration. A role for the hippocampus in mice learning their way around mazes is suggested by "knockout" experiments that use genetic manipulations to delete specific synaptic components in the hippocampus. 58 In mice with such deletions, learning is compromised. Brain imaging studies in humans have now shown that locating and navigating accurately in a virtual-reality town presented by a computer display are associated with activation of the right hippocampus.

Why do we not remember events before we were 3 or 4 years old? Studies on children show that from the time they begin to talk, at around age 2, they have both generic memory (recall of repeated episodes) and episodic memories (recall of single episodes). Amygdala and hippocampus have developed sufficiently to support emotional and factual memories. But the kinds of memories that last a lifetime are initiated only when children begin to weave together autobiographical memory---a story line---at around age 3 1/2 59 . It is at this same age that children begin to show strikingly different approaches to their daily experience, different ratios of introversion/extraversion, and patterning versus dramatic behaviors. 60

Memory is not like a fixed filing system. It is not recording information like a camera. What is created is a semblance or approximation of the original events, sometimes with many biases and errors. We know too that the strands of memory are not complete, much information is filtered out. The strands are not always accurate; they can be distorted by personal goals, current knowledge, or social conventions. 61 A given memory changes over time to resemble previous experiences of a similar nature, corresponding to the mental templates into which we fit our experiences. Numerous experiments demonstrate that false memories do indeed occur.



You can perform a demonstration of false memory by simply reading out loud to a friend a list of similar items (such as bed, rest, awake, tired, dream, wake, snooze, doze, slumber). If you then ask your friend whether an unspoken but related word (such as sleep) was on the list, he or she is very likely to report that it was there, even though it was not. 62


Models of Recognition and Memory

Models of interactions between cortex and hippocampus during recognition and memory have been constructed by using computer nerve nets that are very simplified versions of systems of real neurons. 63 A nerve net that models a first-stage attentional subsystem (like the hippocampus) learns and forms short-term memory traces that are relayed to a second stage (like the medial temporal lobe), where long-term traces are formed. These are relayed back to the first stage. Learning occurs when a mismatch between bottom-up and top-down patterns occurs. Thus seeing a familiar face does not trigger learning (that face is already present in the top-down patterns), but seeing an unfamiliar face does. Efforts are being made to correlate such nerve net models with brain imaging experiments. Activation of left dorsal frontal, hippocampal, and other medial temporal brain areas is observed during the encoding of words, whereas the encoding of unfamiliar faces activates corresponding areas in the right hemisphere.

Imaging studies are demonstrating that our learning and memory engage areas of the cortex that deal with sensory input and motor output. 64 Anatomically separable regions in the left hemisphere tend to process words for distinct kinds of items, such as animals, persons, or tools. At least 15-20 different categories for storing knowledge have emerged from brain lesion studies, the lesions being spread mainly over the occipitotemporal axis of the left temporal lobe. 65 Among the categories are plants, animals, body parts, colors, numbers, letters, nouns, verbs, proper names, faces, facial expressions, and categories of foods such as fruits and vegetables, and they all seem to have separate visual and verbal storage areas. The category an object belongs to acutely affects how it registers in the brain.



In thinking about a cow, we apparently must use the circuitry for the category "animal" in order to come up with a color, size, and so on. Objects in a different category, such as "fabricated objects" can pass through a different set of processing channels.


Work with amnesiacs suggests that parallel brain systems exist for remembering specific items and remembering the categories to which they belong. 66 In some amnesiac patients, if a series of related items are displayed, knowledge of the relationship is formed by some sort of procedural memory process, but recall of specific items is impaired. Thus an item might not be remembered, but it can be assigned to a category. This suggests that the neocortex can gradually acquire classification knowledge independently of the hippocampus and related bodies that are damaged in amnesia.

One study on a patient who had sustained cerebral damage shows evidence for (1) a major division between visual and linguistic higher-level representations and (2) processing subsystems within language. The patient could not name animals, regardless of the type of presentation (auditory or visual), but had no difficulty naming other living things and objects. Accuracy of the physical attributes of animals could be distinguished (such as whether animals were correctly colored), but the color of a given animal could not be stated. In this case, knowledge of physical attributes was strictly segregated from knowledge of other properties in the language system. 67 In other cases, patients are unable to generate visual images "in the mind's eye" but have reasonably intact recognition of what they see. The opposite is also observed: Recognition is impaired but internal imagery is intact. This observation suggests that functionally (and perhaps anatomically) distinct routes to a central store of visual templates can be selectively damaged. 68

We have talked about changes in nerve nets, areas becoming linked, and regions where information is stored, but in this book we are not examining the details of what goes on at the level of individual nerve cells and the synapses they use to "talk" with each other. Most investigators assume that memories ultimately have to correspond to long-term changes, either strengthening or weakening, in the information transmitted by individual synapses. They probably also involve both the formation of new synapses and the destruction of existing ones (as in Figure 6-6), processes central to the development and wiring of brain regions described earlier in this chapter. An enormous amount of contemporary research is focused on trying to describe the cellular mechanisms that can cause such changes. There is now evidence that nearly every aspect of synaptic function can be altered in a long-term way: how much neurotransmitter is released when an impulse reaches a presynaptic terminal;the sensitivity or number of postsynaptic receptor molecules that are activated by the neurotransmitter; and the tightness and surface area of the synaptic contact. Formation of new synapses during memory consolidation has also been documented. In many of these cases, the neurotransmitter glutamic acid and its several postsynaptic receptors appear to play a central role.

Procedural Memory

The formation of our procedural, implicit, or nondeclarative memories and habits are associated with regions of the cortex, thalamus, and basal ganglia that are largely distinct from those involved in the formation of explicit memories. 69 This system can sometimes step in to replace functions that a damaged explicit memory system can no longer serve. 70 Procedural memory is used in learning complicated motor sequences associated with manual skills, and as we noted earlier, plasticity in the adult brain is associated with such learning. For example, when a novel finger exercise is taught to the right hand, imaging studies show increased activity in left motor, premotor, supplementary motor, and sensory cortex and in the right anterior lobe of the cerebellum. The idea is that the initial stages of motor learning are related to somatosensory feedback processing and internal language for guidance of the finger movements. As skill is acquired, the areas of activation become less global. 71

PET imaging studies on the processes underlying procedural memory have made use of the phenomenon of priming, a change in the ability to identify or produce an item as a result of a specific prior encounter with it. 72 A standard procedure is to show, on a screen, during the hour before a conversation with the subjects, a word or picture for which there is no conscious recall, and then ask subjects to say the first word that comes to mind when a list of approximately ten word beginnings appears on the screen. For "mot-," for example, the subject might reply "motor" or "mottle" or "motive." Those who have seen the word "motel" on their screen within the past 30 minutes recall it without knowing why. When they are asked just to finish the word, the visual cortex lights up, so the process seems more like perception than thinking. This suggests little understanding of the word, only a perception of its letter shapes. Patients with a damaged hippocampus perform this task normally. By contrast, asking subjects to try to recall, from a previously presented list, words that match the stem currently being presented causes the frontal lobe to light up, which implies that conscious, higher-level thinking is needed in the memory search. The simplest tasks that can be devised light up several parts of the brain at once, which suggests many little "processors" or agents that are linked together.

Olfactory Memories

Smell is one of our most powerful and primitive memory systems. 73 You may have noticed this when a smell distinctive of your childhood unleashed a flood of memories. Olfactory input links directly to primitive brain centers in the hypothalamus, and manipulation of this system is the goal of a new aromatherapy industry that aspires to influence mood through smells, with different odors facilitating alert, calm, or refreshed moods. 74 In addition to the primary olfactory system that responds to smells, higher vertebrates, including humans, contain a second system, the vomeronasal system, that responds to sexual attractant molecules (pheromones). Although this system is very important in the mating of many mammals, it has been assumed to be vestigial in humans. But perhaps this is not so. The vomeronasal system does project to regions of the hypothalamus that are involved in sexual and affective behavior. And only recently it has been found that odorless compounds from the armpits of women in the late follicular phase of their menstrual cycles can shorten the menstrual cycles of women exposed to the compounds. Some biotechnology companies are now working on what they think are pheromones extracted from human skin that can regulate affect. 75 Another random and curious point about what may be a genetically encoded smell memory: A woman, when asked to sniff T-shirts that men have slept in, tends to prefer the smells emanating from men who have histocompatibility complexes (genes involved in generating our immune systems) that differ from her own. Mating with such men would confer a wider array of immunity options on the offspring. 76

The Sexual Brain---Plasticity Induced by Hormones

This chapter has emphasized how malleable the brain is to shaping by the external environment. A final example of this plasticity is a topic that has inspired much debate: the sexual differentiation of our brains. A crucial environment during the brain's plastic development is the uterus, where the developing embryo is exposed not only to its own hormones but also to those of adjacent embryos and its mother. This environment shapes the development of sexually differentiated brains and behaviors. There is intense interest in this area, and few topics excite livelier debate than the issue of nature versus nurture in determining sexual behaviors.

After the internal and external genitalia, the brain is the most significant area of sexual differentiation. 77 The hypothalamus contains separate centers for male-typical and female-typical sexual behavior and feelings. The medial preoptic area (just in front of where our optic nerves cross each other) plays a vital role in male-typical sexual behavior. It has a high level of receptors for testosterone (released by the gonads). Damage to this region in male animals of many species can cause a reduction in or cessation of heterosexual copulatory behavior and sometimes causes those males to perform more female-typical behaviors in approaches to a stud male. If this area in normal male animals is electrically stimulated, they begin mounting and pelvic-thrusting behaviors. At least one cluster of nerve cells in the medial preoptic region is sexually dimorphic---that is; it differs between the sexes. The third interstitial nucleus of the anterior hypothalamus is threefold larger, on average, in males than in females. This is the area that has been suggested to vary with sexual orientation in males, being smaller in gay men. 78 The formation of the sexually dimorphic nucleus depends crucially on the circulating levels of testosterone during a critical period just prior to and just after birth. During this period in females, neurons of the nucleus die, and this can be prevented by adding testosterone. It takes a specific external signal, the presence of sufficient circulating levels of androgens, to reprogram development in the male direction.



Androgen receptors are found throughout the brain, and their stimulation by testosterone during development appears to play a large part in superimposing male features on the brain's intrinsic female developmental program.


Two further brain regions that show sexual dimorphism are the corpus callosum and the anterior commissure. The anterior commissure is a bundle of nerve fibers connecting the two hemispheres that appeared early in vertebrate evolution; the larger corpus callosum that connects the hemispheres is unique to placental mammals. Both of these structures have been reported to be a larger part of the total brain in women than in men, which suggests richer connectivity between the hemispheres in women. This difference may account for women's greater verbal fluency and for the fact that language seems to be restricted to the left hemisphere to a lesser extent in women than in men. A number of other differences between men and women in brain organization have been suggested. 79 Brain damage studies show that for language and control of hand movements, women depend more than men on anterior regions of the brain.

Hormonal Influences on Behavior

The organizing effects of testosterone have major consequences for behavior. Prenatal testosterone levels in male monkeys influence their levels of rough-and-tumble play. Human girls who have congenital adrenal hyperplasia, and thus are exposed to unusually high androgen levels prenatally, can show play and toy preferences more typical of males. There is considerable variability in sexual behavior even among animals of the same sex, and this can sometimes be traced to natural processes during development. Female rats sometimes mount other females and perform pelvic thrusting. The propensity for this male-typical behavior is strongly influenced by the position of the female when it was a fetus in the uterus. Female fetuses that happen to lie between male fetuses are more likely to display mounting behavior as adults than those that lie between female fetuses. This appears to be because they take up testosterone from the adjacent male fetuses. In human female-to-male transsexual operations with testosterone administration, aggressive tendencies, sexual motivation, arousability, and visuospatial ability increase and verbal fluency declines. The results for male-to-female subjects after androgen deprivation are changes in the opposite direction. 80

A masculinizing effect can be seen on the auditory systems of women who have a male twin. 81 Normal human cochleas (inner ears) generate very weak sounds that propagate back through the middle ear, where they can be recorded. Women exhibit more of these weak sounds than men, a sex difference that exists from birth. Females who have a male twin generate about half the average number of sounds observed in same-sex female twins and in female non-twins, about the same number as seen in men. Prenatal exposure to high levels of androgens has apparently had a masculinizing effect on the auditory systems of these women.

The trait of male aggressiveness in mammals correlates with greater levels of testosterone from puberty onward. Males castrated before puberty (including human eunuchs) are more docile. The relationship between testosterone and behavior is not a simple one, for testosterone and its related group of androgens can be both causes and effects of aggressive behavior. Nothing is ever simple when we are dealing with hormones, and the testosterone = aggression equation is no exception. Deficiency of testosterone can also be linked to aggression. 82 (Targeted at men over age 50 who worry about their declining energy and libido, an "Androderm patch" sold by a pharmaceutical company releases 5 mg a day of testosterone into the bloodstream. It is reported to boost libido, increase muscle mass, and lower the voice! There are concerns, however, about the long-term effects of such hormone therapy.)



The greater aggressiveness of human males is perhaps the most striking example of a sexually differentiated trait. The trait is observed in most mammals and is presumably the result of sexual selection during evolution.


The amygdala is important in sexual behavior and aggressive behavior, which involve the corticomedial and basolateral nuclei, respectively. The corticomedial nucleus links to the medial preoptic area, and a corticomedial lesion can interfere with sexual behavior. The basolateral nucleus connects to regions farther back in the hypothalamus that play a role in aggressive behavior, and a basolateral lesion can reduce aggressive behavior and rough-and-tumble play. Bilateral ablation of the amygdala has been performed on very aggressive adults and children who have not responded to other forms of treatment, with the result of reducing their aggressive behavior.

Hormonal Influences on Visuospatial Skills

The other most widely studied difference between male and female brains involves spatial or visuospatial skills. On average, males (rat or human) are better at mazes than females, and human males are better at visualizing rotating objects. (We don't know how to test rats for this latter capability!) Conversely, women tend to outperform men on some tests of verbal ability and fluency. Human women (and female rats) have a greater tendency to use landmarks in spatial learning tasks, whereas males use geometric cues. Are these cognitive differences inborn or learned? Probably both, but it is striking that women who were exposed to high levels of androgenizing hormones while they were fetuses (either as a result of congenital adrenal hyperplasia or because their mothers were given the synthetic steroid diethylstilbestrol) score better than other women on spatial tests. Similarly, the maze-running performance of female rats given androgens around the time of birth is better than that of untreated females. In androgen-insensitivity syndrome, chromosomal males have testes that secrete testosterone, but these males have a mutation in the gene for the androgen receptor. The body acts as though no androgens were present and develops as female; individuals usually are raised as girls. They do worse on visuospatial tasks than their male relatives and have verbal skills similar to their female relatives. During the menstrual cycle in women, as estrogen levels increase, spatial abilities decrease, and articulatory abilities increase. In men, a lowering of testosterone levels in the spring correlates with an increase in spatial abilities. In all of these examples, hormones apparently modulate the synaptic pathways that underlie our higher cognitive faculties.



Cognitive patterns may remain sensitive to hormonal fluctuations throughout life. Recent animal studies have shown that steroid hormones can cause cellular changes and synaptic reorganization in the adult hypothalamus. 83 Estrogen administration stimulates memory and the growth of axons and dendrites in the hippocampus. 84



The examples of brain plasticity that have emerged from studies on the sexual differentiation of the brain and on the mechanisms of learning and memory paint a picture of our developing and adult brains as plastic devices that are constantly re-forming themselves. The evolutionary rationale for this plasticity is that it permits us to wire up our brain circuits in response to the world in which we actually grow up and function, rather than the world of our ancestors. This description represents a revolutionary change in perspective, for as recently as 30 years ago, it was widely assumed that development was rigidly programmed and that adult brains remained constant through life. At the level of individual cells and complex systems of cells, what we see instead is a constant testing of what works best and an ability to overcome even extraordinary perturbations or lesions in normal development. A most dramatic example is provided by the near-normal development of human infants that have had an entire cerebral hemisphere removed. We will consider some further examples of brain plasticity when, in Chapter 11, we discuss the effects of genetic mutations and brain lesions on the development of areas of the brain crucial to language.

Each of our systems, from fundamental visual and motor mechanisms to the higher functions of socialization and personality formation, appear to have a critical period, or window of time, during which their habits are laid down. Once the structures are formed, we are pretty much stuck with them. Even so, our adult brains have abilities to rewire themselves when new tasks of discrimination are required or when areas of brain tissue have been damaged. Changes in the brains of some animals have been observed to accompany changes in social status, and it is not unreasonable to suppose that analogous processes might occur in humans. Subtle changes in nerve connections underlie the memory systems that permit us to recall places and events and to remember skilled action sequences. Interactions with the internal hormonal environment also pattern brain plasticity, as in the sexual differentiation of the brain. The immersion of each of our developing brains in a particular physical, linguistic, and cultural environment generates in each case a physically unique structure. This structure, broadly similar in all humans but differing for each individual in the details, supports the generation and definition of the human selves that are the subject of Chapter 7.

Questions for Thought

1. Permitting a nervous system to wire up in a flexible way, depending on the environment that the organism actually encounters, seems like such a good trick that one might expect animals less able to do this, such as insects, to have lost out in the evolutionary competition. Why, then, are beetles perhaps the most varied and numerous animal species on this planet?

2. There is debate about whether complicated nerve circuits---such as those that coordinate hand and eye movements---are formed by processes akin to Darwinian evolution or by an active, constructive learning process. Are the nerve pathways selected from existing options or invented? Can you design some thought experiments that would determine whether either of these alternatives (or both) is (or are) correct?

3. Suppose you were curious about observing the plasticity of your own adult cerebral cortex, and suppose you had a friend who was willing to let you play with her MRI or magnetoencephalography equipment. This machinery allows you to observe what brain areas become most active during a given task. What experiment would you design, and what result might you expect to see?

4. Storage in memory of the objective factual components and the emotional components of an experience appears to require the activities of two different brain structures. What are these structures, and what kinds of observations have demonstrated their importance?

Suggestions for Further General Reading

Deacon, T.W. 1997. The Symbolic Species: The Co-Evolution of Language and the Brain. New York: Norton. Chapters 6 and 7 discuss brain development and its plasticity.

Bear, M., Connors, B., & Paradiso, M. 1996. Neuroscience: Exploring the Brain. Baltimore, MD: Williams & Wilkins. Chapter 18 of this book discusses the wiring of the brain.

Schacter, D.L. 1996. Searching for Memory: The Brain, the Mind and the Past. New York: Basic Books. An engaging description of memory mechanisms and how they can be distorted.

LeVay, S. 1993. The Sexual Brain. Cambridge, MA: M.I.T. Press. This book is one of the sources for the section on the sexual differentiation of the brain.

Reading on More Advanced or Specialized Topics

Deacon, T.W. 1990. Rethinking mammalian brain evolution. American Zoologist 30:629--705. This article summarizes the experiments, described in Figure 6-5, that demonstrated that cortical function can be switched from one sensory mode to another.

Florence, S.L., Jain, N., Pospichal, M.W., Beck, P.D., Sly, D.L., &Kaas J.H. 1996. Central reorganization of sensory pathways following peripheral nerve regeneration in fetal monkeys. Nature 381:69--71. This is the article on which Figure 6-7 is based.

Yeh, S., Fricke R., & Edwards D. 1996. The effect of social experience on serotonergic modulation of the escape circuit of crayfish. Science 271:366--369. These experiments demonstrate a link between social experience and the chemistry of specific synapses.

Milner, B., Squire, L.R., & Kandel E.R. 1998. Cognitive neuroscience and the study of memory. Neuron 20:445--468. A description of the history and current status of studies on the mechanisms of memory.

Ungerleider, L.G. 1995. Functional brain imaging studies of cortical mechanisms for memory. Science 270:769--775.

1. Material in this section is derived from Plotkin, 1994,Ch. 5; Dennett, 1991, Ch.7; Wright, 1994.

2. Gazzaniga, 1993; Sporns and Tonini, eds., 1996.

3. Famous examples are the Wild Boy of Aveyron (1800-1809), and the wolf girls Kamala and Amala (India, 1920) (see Canland, 1993; Maturana and Varella, 1992)

4. Damasio, 1994, pp. 108-112.

5. Even before they are connected into a behavioral pathway, immature circuits generate activity patterns that drive developmental processes. See Feller, 1999.

6. A review of recent work in developmental neurobiology can be found in Science 274:1099-1138, 1996.

7. Elman et al., 1996, contains a useful summary of brain development from which some of this material is taken.

8. Rauschecker, 1995.

9. Sadato et al., 1996.

10. Paulesu et al., 1995.

11. For a review of this work see Deacon, 1990.

12. Roe et al., 1990.

13. Elman et al., 1966 pg. 278.

14. Hata and Stryker, 1994.

15. There is a further twist to this story. The correlation of the firing from the two eyes influences the periodicity of the ocular dominance columns. When the correlation is weakened, which happens when one eye is pointing in a slightly different direction from the other, the spacing of the columns becomes wider. See Goodhill and Lowel, 1995.

16. Ghosh, 1996.

17. Tieman and Hirsch, 1982. Weliky and Katz, 1997, show that another way of disrupting the orientation tuning of visual cortical cells during development is to provide artificially correlated activity into the visual pathway through synchronous activation of retinal ganglion cell axons in the optic nerve. This is another indication that activity has an instructive role in shaping cortical neuron wiring.

18. For discussions of the relative importance of selection versus instruction in developing neural pathways, see Purves, 1996; Quartz, 1999, and Crair, 1999.

19. Florence et al, 1996.

20. See Lowenstein and Parent, 1999; and Kempermann and Gage, 1999, for brief reviews on how the brain can repair itself.

21. Clower et al., 1996.

22. Blakeslee, 1991

23. described in Ramachandran ( 1993), Kandel et al., 1991, pg 1024 ff.; J.Barinaga, 1992), Garraghty and Kaas, 1992.

24. Donoghue, 1995.

25. A list of references to plasticity in human cortex is given in Pascual-Leone (1992 )

26. Gilbert, 1994

27. Mogilner et al, 1993

28. Wang et al., 1995.

29. Elbert et al., 1995.

30. Karni et al., 1995.

31. Chase, M. 1993

32. Feldenkrais, 1972.

33. See Melzack, 1992, for a general discussion of phantom limbs.

34. Ramachandran, 1993.

35. Taub et al, 1995. See Merznich, 1998, for a brief review of changes in the cortex and the thalamus observed after limb amputation.

36. Yau et al., 1994.

37. Ramachandran et al., 1996.

38. Angier, N. 1991.

39. Yeh et al., 1996.

40. Reynolds, 1980, pg. 149; also Sapolsky, 1994, Ch. 13 discusses rank, stress and glucocorticoid levels in Baboons.

41. Morell, 1996

42. Seligman, 1994.

43. Bear et al., 1996, Chapters 18-20.

44. Miller, 1956; and see Ornstein, 1991, Ch. 18 for a discussion on memory

45. Beardsley, 1997.

46. I am not giving a complete list of memory systems here. There are at least six: 1. Explicit memory, which is conscious memories of facts and events, associated with medial temporal lobe and diencephalic structures; 2. Working memory, which maintains activity of other representations and is associated with prefrontal cortex; 3. Priming, which tunes perceptual and conceptual representations, associated with occipital, temporal, and frontal cortex; 4. Motor skill learning associated with the striatum; 5. Classical conditioning which learns relationship between perceptual stimuli and skeletal motor responses, associated with the cerebellum; and, 6. Emotional conditions, which relates perceptual stimuli and emotional responses and is associated with the amygdala. See Willingham, 1997.

47. For discussion of the contributions of different cortical areas to memory processes see Owen, 1998; Murray and Bussey, 1999; and Buckner, Kelley and Petersen, 1999.

48. The term episodic memory is restricted by some to refer just to recall of events that happened to you, and the term semantic memory refers to recall of facts like the location of cities. Vargha-Khadem et al., 1997, have examined several patients with hippocampal lesions occurring at an early age who can not remember episodes of daily life but maintain low to average levels of semantic memory, or factual knowledge.

49. Squire and Zola-Morgan, 1991. Petri and Mishkin, 1994, is a review of the brain systems participating in explicit memories. The hippocampus appears to also be part of the circuits involved in novelty detection (Knight, 1996).

50. Bechara et al., 1995.

51. Squire and Zola-Morgan,1991. For a review, see Parkin, 1996.

52. Hilts, 1991a.

53. See Gould et al, 1999, for discussion of a possible role in learning attributed to the production of new neurons in the hippocampus, and Engert and Bonhoeffer, 1999, for description of how dendritie spines change during long term changes in hippocampal synapses.

54. Experiments by Miyashita et al. (1996) with monkeys suggest that if this step is blocked, associative memories can not be formed.

55. Fletcher et al., 1997. See Turrigiano, 1999, for a discussion of how experience might modify the properties of neuronal networks.

56. For a review of place cells, see Muller, 1996.

57. Skaggs and McNaughton, 1996.

58. Roush, 1997;Wilson and Tonegawa, 1997.

59. Goleman, 1993

60. Storr, 1989, pg. 90 ff

61. Schacter, 1996.

62. Beardsley, 1997a.

63. Carpenter and Grossberg, 1993;Eichenbaum, 1993, Eichenbaum et al., 1994.

64. Ungerleider, 1995.

65. Damasio et al, 1996. See also Hilts, 1992; 1995.

66. Knowlton and Squire, 1993

67. Hart and Gordon, 1992

68. Gurd and Marshall, 1992b

69. Graybiel, 1995.

70. Stipp, D. 1993.

71. Seitz and Roland, 1992

72. Squire et al, 1992; Hilts, 1991.

73. The book by R.Joseph, 1993, has an extended discussion of this topic.

74. O'Neill,1991

75. Blakeslee, S. 1993. Human nose may hold an additional organ for a real sixth sense. N.Y.Times 9/7/93.

76. Mirsky, 1995. Richardson, 1996.

77. Much of this material is taken from LeVay, 1993.

78. LeVay, 1991. While there is intense debate about the genetic basis of sexual orientation and sexual behavior in humans, it is clear that sex specific behaviors in fruitflies are primarily determined by genes (O'Dell and Kaiser, 1997).

79. Kimura, 1992. For a more recent review, see Kimura, 1996.

80. Holden, 1995.

81. McFadden ( 1993)

82. Sapolsky, 1997.

83. Mainard-Demotes et al, 1993

84. Wickelgren, 1997b.

Home   |   Mind Blog   |   Lectures & Writings   |   Music   |   About Me   |   Contact

©Deric Bownds • All Rights Reserved
Counter 1