Chapter 2: Neuroscience of Learning and Memory PDF

Chapter 2 The Neuroscience of Learning and Memory Structural Plasticity in Nervous Systems Learning and Memory in Everyday Life: Top Five Tips for Faster Forgetting  Brains How Experience Changes Brain Structure Functional Properties of Learning and Memory Systems What Brains Do...

Chapter 2 The Neuroscience of Learning and Memory Structural Plasticity in Nervous Systems Learning and Memory in Everyday Life: Top Five Tips for Faster Forgetting  Brains How Experience Changes Brain Structure Functional Properties of Learning and Memory Systems What Brains Do How Remembering Changes Brain Activity Finding and Manipulating Memories Looking for Memories How Researchers Change Brain Activity Learning and Memory in Everyday Life: Can a Pill Improve Your Memory? IN THE MIDST OF A NEIGHBORHOOD BASEBALL GAME, a ball hit Orlando Serrell in the head so hard that it knocked him to the ground. At the time, he was only 10 years old, and like most boys who take a hit while playing with peers, he walked it off and eventually went back to the game. This seemingly innocuous incident proved anything but typical, however. Sometime after the hit, Orlando discovered an amazing ability: for any date after that fateful game, he could remember the day of the week on which that date fell, as well as what the weather was like on most of those days— without making any conscious effort to memorize these details or perform calculations with dates. Orlando’s case is not unique but is an instance of a rare condition called acquired savant syndrome (Treffert, 2014 ). Chapter 1 describes Clive Wearing, who lost many of his memory abilities after part of his brain was destroyed by a rare virus. Unlike Clive, individuals with acquired savant syndrome actually can gain prodigious memory capacities as the result of brain injury. The startling implication of this phenomenon is that at least some (and perhaps all) human brains appear to have a much greater capacity for storing and recalling memories than people typically exhibit. If humans have hidden learning and memory capacities, might other animals also possess capacities of which we are currently unaware? If brains have such capacities, then why can’t all individuals take full advantage of them? Might it be possible to develop neural technologies that improve a person’s encoding and recollection of specific facts or erase memories of episodes one would prefer to forget? The story of how scientists explore such questions, and identify the biological factors that determine what an individual remembers or forgets, is the story of the neuroscience of learning and memory. Although scientists still have a long way to go in understanding how nervous systems work, they are compiling fascinating information about the brain’s structure and functioning, as well as the ways in which it contributes to learning and memory. New imaging and sensing technologies allow researchers to observe healthy human brains as they form and retrieve memories, while new techniques for animal research allow researchers to measure and manipulate neural changes during learning. Insights into the neural mechanisms of learning and memory can help you understand how your actions may impact your own attempts to learn, remember, and in some cases forget the materials you study (for some ideas, see “Learning and Memory in Everyday Life ”). LEARNING AND MEMORY IN EVERYDAY LIFE Top Five Tips for Faster Forgetting Chapter 1 provides 10 tips for how you can change your behavior to improve your memory. In case you are more interested in erasing memories than retaining them (wasting time is fun!), you can also modify your brain function to improve your forgetting. Here is what you do. 1. Don’t sleep. People who do not get enough sleep are less able to concentrate during the day, which makes it harder for them to encode new memories and retrieve old ones. Sleep is also important for transforming the learning of the day into long-term memories. Sleepy brains work worse. (More on this in Chapter 7.) 2. Stress out. Stress generally interferes with recall. So, if you want to make retrieving memories particularly troublesome, just keep fixating on trying to remember things you can’t, until frustration overcomes you. (More on this in Chapter 7.) 3. Overextend yourself. The more things you try to keep in mind simultaneously, the greater the chance you will forget a bunch of them. So put aside all your note-taking devices—your pens, pads, computers, and iPhones—if you really want to maximize your loss. (More on this in Chapter 9.) 4. Deprive your senses. The more impoverished your sensory inputs, the less likely you are to encode facts, events, and skills well enough to recall them later. Wear headphones, shades, and oven mitts. Minimize your brain activity. (More on this in Chapter 3.) 5. Be apathetic. Nothing is more forgettable than something you couldn’t care less about. Just keep chanting inside your head, “Whatever … whatever … whatever … ,” and you can easily avoid the kinds of emotionally triggered brain states that make memories stick. (More on this in Chapter 7.) 2.1 Structural Plasticity in Nervous Systems Researchers in the field of neuroscience — the study of the brain and the rest of the nervous system—overwhelmingly believe that the brain is the seat of learning and memory. This was not always the prevailing opinion, however. When ancient Egyptians mummified a body, they first removed the organs they considered important and preserved them in special airtight jars —but they discarded the brain. Many centuries later, Aristotle, one of the most empirically oriented philosophers in history (whose ideas on memory are noted in Chapter 1 ), argued that the brain serves primarily to cool the blood. However, observations over the centuries since Aristotle’s time have convinced scientists that brain activity controls behavior and, by extension, the changes in behavior associated with learning and memory. neuroscience The study of the brain and the rest of the nervous system. Historically, most early studies of learning and memory focused on observable behavior rather than on the brain and how it functions (Chapter 1 ). This is not because early learning and memory researchers were oblivious to the importance of the brain. Ivan Pavlov designed all of his behavioral experiments to answer questions about how the brain works. John Watson, the originator of behaviorism, started out studying how developmental changes in neural structures correlate with developmental changes in learning abilities. B. F. Skinner, perhaps the most famous behaviorist of the twentieth century, began his career as a physiologist. Why, then, did these researchers place so much emphasis on behavior and so little emphasis on the role of the brain? Part of the answer is that brains are among the most complex structures in nature. Even as recently as 60 years ago, the complexity of the neural functions required for most learning tasks seemed incomprehensible. As new technologies became available, however, the study of brain function became more manageable. Today, aspects of brain function that previously were inaccessible are being measured daily in laboratories and medical institutions around the world. These new technologies have dramatically increased the number and productivity of studies exploring the neural substrates of learning and memory. Brains The brain is just one component—albeit a very important one—of a collection of body organs called the nervous system , the organ system devoted to the distribution and processing of signals that affect biological functions throughout the body. The tissues that are specialized for accomplishing these tasks include cells called neurons , which perform a variety of tasks, such as collecting incoming signals from sensory systems (leading to sight, taste, smell, touch, and sound) and from the rest of the body (indicating such conditions as hunger and sleepiness), processing these signals, and reacting to them by coordinating the body’s responses (such as muscle movement and activity of internal organs). nervous system An organism’s system of tissues specialized for monitoring sensations, generating movements, and maintaining the function of most internal organs. neuron A special type of cell that is one of the main components of the nervous system. In vertebrates, the nervous system can be divided into two parts: the central nervous system and the peripheral nervous system. The central nervous system (CNS) , where most of the events responsible for learning and memory take place, is made up of the brain and the spinal cord ( Figure 2.1 ). The peripheral nervous system (PNS) consists of nerve fibers that connect sensory receptors (for example, visual receptors in the eye or touch receptors in the skin) to the CNS and of other fibers that carry signals from the CNS back out to the muscles and organs. Most of these fibers pass through the spinal cord, but a few—such as those from the light receptors in your eyes and those that activate the muscles controlling eye movements— travel directly to the brain without first making connections in the spinal cord. FIGURE 2.1 Nervous system components Every vertebrate has a central nervous system (CNS) and a peripheral nervous system (PNS). The CNS consists of the brain and spinal cord. The PNS consists of motor and sensory neurons that transmit signals between the CNS and the rest of the body: (1) sensory receptors in the skin, eyes, ears, and so on provide sensory inputs to the CNS; (2) motor fibers deliver directives from the CNS to muscles; (3) PNS fibers from the CNS regulate organs and glands. Description The parts are as follows. Central nervous system (CNS) comprising the brain and the spinal cord, Peripheral nervous system (PNS) comprising motor and sensory neurons that connect  the brain and the spinal cord to the rest of the body: 1. Sensory organs (skin, eyes, ears, etc.), 2. Muscles, and 3. Body organs. central nervous system (CNS) The part of the vertebrate nervous system consisting of the brain and spinal cord. peripheral nervous system (PNS) The part of the nervous system that transmits signals from sensory receptors to the central nervous system and carries commands from the CNS to muscles. Although every vertebrate possesses a CNS and PNS, there are big differences between the nervous systems of different species. Let’s look at the vertebrate you are probably most familiar with: the human. The Human Brain The cerebral cortex , the tissue covering the top and sides of the brain in most vertebrates, is by far the largest structure of the human brain ( Figure 2.2A ). The word cortex is Latin for “bark” or “rind,” reflecting that the cortex, although about the size of the front page of a newspaper if spread out flat, is only about 2 millimeters thick. To fit inside the skull, the cerebral cortex is extensively folded, much like a piece of paper crumpled into a ball.  FIGURE 2.2 The visible surface of a human brain (A) A photograph of a human brain. (B) In each brain hemisphere, the visible cerebral cortex is divided into four principal areas: frontal lobe, parietal lobe, occipital lobe, and temporal lobe. Below the cerebral cortex are the cerebellum and brainstem. The brainstem connects the brain to the spinal cord. Description Section A shows the external view of the human brain. Section B shows a labeled schematic of the human brain. The parts are as follows. The cerebral cortex is divided into four major areas: frontal lobe, parietal lobe, occipital lobe, and temporal lobe. The anterior part of the brain is labeled, frontal lobe, it is the largest part of the brain. The adjacent part of the frontal lobe is labeled, parietal lobe, it is located at the top of the brain. The rearmost lobe of the cerebral hemisphere is labeled Occipital lobe. It is located behind the parietal and temporal lobes. The second largest part of the brain is labeled, temporal lobes, it is located anterior to the occipital lobes and inferior to the parietal and frontal lobes. The cerebellum and brainstem are labeled inferior to the four lobes. cerebral cortex The brain tissue covering the top and sides of the brain in most vertebrates; involved in storage and processing of sensory inputs and motor outputs. In humans, as in all vertebrates, the brain consists of two sides, or hemispheres , that are roughly mirror images of each other, so brain scientists, when talking about the cortex, may specify the “left hemisphere” or the “right hemisphere.” In each hemisphere, the cortex is divided further into the frontal lobe at the front of the head, the parietal lobe at the top of the head, the temporal lobe at the side of the head, and the occipital lobe at the back of the head (Figure 2.2B ). The term lobe refers to the fact that these regions are anatomically distinct. The individual lobes got their somewhat odd names from the names of the skull bones that cover them. If you have trouble memorizing these four terms, remember: “F rontal is F ront, P arietal is at the P eak, T emporal is behind the T emples, and the O ccipital lobe is O ut back.” frontal lobe The part of the cerebral cortex lying at the front of the human brain; enables a person to plan and perform actions. parietal lobe The part of the cerebral cortex lying at the top of the human brain; important for processing somatosensory (touch) inputs. temporal lobe The part of the cerebral cortex lying at the sides of the human brain; important for language and auditory processing and for learning new facts and forming new memories of events. occipital lobe The part of the cerebral cortex lying at the rear of the human brain; important for visual processing. Subregions within each lobe are associated with a wide variety of perceptual and cognitive processes. For example, the frontal lobe helps you retrieve memories from your past, and enables you to keep track of multiple recent experiences; the occipital lobe allows you to see and recognize the world; the parietal lobe enables you to feel the differences between silk and sandpaper and to associate those differences with the look of silk and of sand; and the temporal lobe makes it possible for you to hear and to remember what you have done. We will discuss the functional roles of cortical subregions in greater detail throughout this book. Knowing the names and locations of the different lobes will help you keep track of what is happening where in your brain. Sitting behind and slightly below the cerebral cortex is the cerebellum (Figure 2.2B ), which contributes to the coordination of sensation and movements and is thus especially important for learning that involves physical action. At the base of the brain is the aptly named brainstem (Figure 2.2B ), a collection of structures that connects the brain to the spinal cord and plays a key role in the regulation of automatic functions, such as breathing and regulating body temperature. cerebellum A brain region lying below the cerebral cortex in the back of the head. It is responsible for the regulation and coordination of complex voluntary muscular movement, including classical conditioning of motor- reflex responses. brainstem A group of structures that connects the rest of the brain to the spinal cord and plays key roles in regulating automatic functions such as breathing and body temperature. Other brain structures, buried under the cerebral cortex, are not visible in photographs such as that shown in Figure 2.2A. You will learn about many of these structures later in the book. For now, we will just introduce a few that are especially important for learning and memory ( Figure 2.3 ). FIGURE 2.3 Brain regions known to contribute to learning and memory Located near the center of the human brain, the basal ganglia, thalamus, hippocampus, and amygdala all contribute to learning and memory in different ways. Description The labeled parts are as follows. Basal ganglia, Thalamus, Amygdala, and Hippocampus. The basal ganglia is located deep in the cerebral hemispheres of the brain. The Thalamus is located in the rostral end of the brainstem. The amygdala is located deep in the medial  temporal lobe. It has an almond-shaped structure. The hippocampus is located in the temporal lobes of the brain. It has a horseshoe-shaped structure. First, near the center of the brain lies the thalamus , a structure that receives various sensory signals (associated with sight, sound, touch, and so forth) and connects to many cortical and subcortical regions. You can think of the thalamus as a gateway through which almost all sensory signals pass in order to affect brain activity. Sitting near the thalamus are the basal ganglia , a group of structures important for learning, planning, and producing skilled movements such as throwing a football or juggling. The hippocampus lies a little farther away, inside the temporal lobe; it is thought to be important for learning how to get places and for remembering autobiographical events (such as what you did last summer). Sitting at the tip of the hippocampus is a group of cells called the amygdala ; this little brain region is important for emotional memories. If you remember the happiest—or saddest—day of your life, it is probably because your amygdala was particularly active at the time, adding emotional strength to those memories. Because you have two hemispheres, you actually have duplicates of each of these structures. For example, you have a left hippocampus and a right hippocampus and a left amygdala and a right amygdala. Scientists are only beginning to understand what these brain areas do and how they relate to learning and memory, but it is becoming increasingly clear that it is a mistake to think of the brain as a single organ, like a liver or a kidney. Instead, the brain is a “society of experts,” with each region making its own specialized contribution to what we do and think. Imaging Brain Structure In the late 1800s, Franz Joseph Gall (1758–1828), a German anatomist and physiologist, pioneered the idea that different areas of the cerebral cortex are specialized for different functions. He also proposed that differences in character or ability are reflected in differences in the size of the corresponding parts of the cerebral cortex: that is, people with a special skill for languages would have a larger-than-average language area, and people with superior memories would have an overgrown memory area. Gall assumed that these differences in cortical size would be evident from bumps in a person’s skull. He developed a technique called phrenology , in which he used skull measurements to predict an individual’s personality and abilities (Gall & Spurzheim, 1810 ). phrenology An obsolete field of study that attempted to determine mental abilities by measuring head shape and size. Phrenology was quickly taken over by quacks who found various ways of making the idea pay. The main problem with phrenology, however, was that bumps on the skull are not caused by bulges in the underlying brain tissue. Gall did not know this because he had no way of examining the brain of a living person. Nearly two centuries later, technology finally advanced to the point where scientists could see inside the skulls of people who are still alive and begin to identify and measure brain structures that determine what individuals can learn and remember. Modern techniques for creating pictures of anatomical structures within the brain are described collectively as structural neuroimaging , brain imaging, or “brain scanning.” The images produced using these methods can show details of brain tissue and also brain lesions , areas of damage caused by injury or illness. structural neuroimaging Techniques (such as MRI) for creating images of anatomical structures within the living brain. lesion Damage caused by injury or illness. Structural neuroimaging provides a way not only to directly observe physical properties of a person’s brain but also to track changes in those properties over time. These include changes that might occur as a function of aging, injury, disease, or learning experiences. Structural images of human brains are also critical for analyzing and interpreting changes in brain function that occur with learning, a topic we discuss in greater detail below. Currently, brain images are most often produced through magnetic resonance imaging (MRI) , in which changes in magnetic fields are used to generate images of internal structure. MRI employs an extremely powerful magnet, usually constructed like a giant tube. The person lies on a pallet that slides into the tube, and magnetic changes are induced in the brain tissues. The brain tissues are then allowed to return to normal. During this latter phase, a computer collects the different signals emitted by different tissues and uses them to generate images that look like photographs of a sliced brain. For example, Figure 2.4A shows an image comparable to what you would see if someone’s head were sliced in half (minus the spewing blood), revealing a cross section of cerebral cortex, cerebellum, and brainstem, as well as some facial structures. An image measured at the level of the eyeballs, as in Figure 2.4B , shows a different cross section. A different type of MRI, called diffusion tensor imaging (DTI) , measures connections between brain regions, enabling researchers to study how major pathways between different brain regions change over time. FIGURE 2.4 MRI images (A) This brain image measured near the center of the head shows a cross section through cortex, cerebellum, brainstem, and an upper portion of spinal cord, as well as nose and mouth cavities. (B) An image measured at the level of the eyeballs (visible at the top of the image) contains little cortex (since the position is so far down in the person’s head) but captures the low-hanging cerebellum. Description The imagery in section A shows a lateral sectional view of the human head with the following parts labeled: The cortex on the upper part of the brain, the cerebellum adjoining the brain stem, the brainstem at the center part of the brain connected to the spinal cord. The nose, mouth, and the spinal cord are also labeled. The imagery in section B shows a sectional view of the human head, from the level of eyeballs. The parts labeled in section B are as follows. eyes, ears, brainstem, left cerebellum, and right cerebellum. magnetic resonance imaging (MRI) A method of structural neuroimaging based on recording changes in magnetic fields. diffusion tensor imaging (DTI) A type of MRI that measures connections between brain regions. It is easy to confuse structural neuroimaging, used to visualize what brains are physically like, with a different kind of imaging, known as functional neuroimaging , which is described in Section 2.2 and shows what brains are doing at the time of imaging. Both types of neuroimaging can reveal changes associated with learning, and examples of both are presented throughout this book. Whenever you come across a figure in which patches of color are superimposed on a picture of a brain, the first thing you should ask yourself is, do these colored regions show changes in brain structure (structural imaging), or do they show changes in brain activity (functional imaging)? Neurons and Glia Neurons are the building blocks of the nervous system. Some act as sensory receptors (such as those in the eyes, ears, and tongue that respond to visual, auditory, and taste stimuli), and some transmit signals from the spinal cord to the muscles. In vertebrates, many of the neurons are in the brain. Neurons are capable of changing their function and modifying the way they respond to incoming signals. These changes are thought to make learning and memory possible. The prototypical neuron has three main components: (1) dendrites , which are input areas that receive signals from other neurons; (2) the cell body , or soma , which integrates signals from the dendrites; and (3) one or more axons , which transmit signals to other neurons ( Figure 2.5 ). For the most part, neural activity flows in one direction: from dendrites to axons. FIGURE 2.5 Neurons, the building blocks of brains (A) Brain tissue, stained to make neurons evident and photographed through a powerful microscope. The pyramid-shaped cell bodies and interconnecting branches of several neurons are visible. (B) The prototypical neuron has three main components: dendrites for monitoring the activity of other neurons, a cell body (soma) that integrates incoming signals, and one or more axons that transmit signals to other neurons. Neural activity flows mainly from dendrites to axon(s). Description The micrograph shows a network of neurons with the cell body, thread-like dendrites and axons labeled. Dendrites and axons are at the opposite ends of the cell body.  The schematic on the right shows the structure of a neuron. The flow of activity is from the top dendrites, down through the cell body, and out the axons at the bottom. Text corresponding to different parts of the neuron read as follows. 1. Dendrites: synapses with axons from other neurons that modulate activity are labeled. Dendrites monitor activity of other neurons. 2. Cell body: combines effects of other neurons. 3. Axons: transmit electrical signals to dendrites that monitor ongoing activity. A synapse is labeled at the end of a dendrite at the bottom of the neuron. dendrite On a neuron, an extension that is specialized to receive signals from other neurons. cell body The central part of the prototypical neuron; it contains the nucleus and integrates signals from all the dendrites. Also known as the soma. axon The output extension of a neuron, specialized for transmitting signals to other neurons or to muscles. When the incoming signals arriving at the cell body have a total electrical charge that exceeds a certain threshold, the neuron “fires,” propagating an electrical charge called an action potential down its axon. This propagation is an all-or-nothing event: either the neuron fires or it does not; there is no in- between stage. It is convenient to talk about a “prototypical neuron.” But in reality, neurons, like brains, come in a wide array of shapes and sizes. For example, pyramidal cells are neurons with pyramid-shaped cell bodies (shown in Figure 2.5A ); stellate cells have star-shaped cell bodies. A neuron may have a single axon, two axons, or many axons. Neurons known as interneurons , which connect two or more neurons, have short axons or no axons at all. The neurons that carry signals from the spinal cord to the feet have axons that stretch a meter or more in humans. The various shapes and sizes of different neurons undoubtedly contribute to their function. But, in many cases, neuroscientists do not know the specific advantages that a particular shape or size provides. Neurons are not the only kind of cell in the brain; they are matched in number by various species of cells known as glia. Astrocytes are glia that line the outer surface of blood vessels in the brain and may help in the transfer of oxygen and nutrients from the blood to neurons. Glia called oligodendrocytes wrap the axons of nearby neurons in myelin , a fatty substance that insulates electrical signals transmitted by neurons, thus speeding the transmission of signals down the axon. Glia are as important as neurons for normal brain (and overall central nervous system) function. For example, multiple sclerosis is a disease in which the myelin coating of axons degenerates; this interferes with neural function, leading to jerky muscle movements and impaired coordination, as well as problems with vision and speech. glia Cells of various types that provide functional or structural support to neurons; some contribute to changes in connections between neurons. Traditionally, researchers thought that glia mainly provide functional and structural support to neurons. New research shows, however, that glia directly contribute to learning and memory by actively controlling changes in connections between neurons (Alberini et al., 2018 ; Fields et al., 2014 ). Even so, most neuroscientists who study the neural bases of learning and memory focus their efforts on understanding neurons: how they control behavior and how they change during learning. How Experience Changes Brain Structure Chapter 1 describes some early theories of associationism , in which memory depends on forming links between experienced events, and of conditioning , in which repeated pairing of stimuli or responses with behaviorally relevant events brings about changes in behavior. William James (1890) proposed that the links created during learning and memory formation are not just conceptual connections but actual, physical properties of brains. According to James, the ability to learn and remember shows that brain structure can be changed by experience. He believed that brain structures gradually change when activity repeatedly flows through specific pathways; like Ivan Pavlov, James thought that it is this plasticity of brain tissue that makes learning possible.  Techniques for imaging neurons in brains make it possible to visualize structural changes that occur during learning. Do you think the changes shown here in green occurred in the dendrites, soma, or axons of these neurons? plasticity The capacity of brain structure to change over time. There are many ways in which experience might change the brain. Networks of neurons can be rewired or replaced. Functional features of neurons might be strengthened or weakened. New functions or even new neurons might be introduced. In the case of learning and memory formation, there is evidence that all these kinds of structural plasticity and more occur. Imprinting Much of what is known about the neural bases of learning and memory comes from studies of animals other than humans. Many aspects of a mouse brain, a monkey brain, a bird brain, and even an insect brain are similar enough to a human brain to have made this possible (as predicted by Darwin’s theory of natural selection, described in Chapter 1 ). Comparative studies provide a foundation for understanding how structural changes in brains relate to learning and memory abilities. One unique case of learning that has proven particularly informative is imprinting , in which a newborn animal forms a lifelong attachment to whatever movements it views early on—typically, those of the animal’s mother. Imprinting is more complicated than it might seem at first. It involves elements of classical conditioning, perceptual learning (discussed in Chapter 3 ), recognition memory (also discussed in Chapter 3 ), and operant conditioning (McCabe, 2013 ). In addition, imprinting occurs only within a short window of time after birth. We will return to the topic of imprinting when we discuss developmental aspects of learning and memory (Chapter 12 ). For now, you just need to know that the memory produced by imprinting is one of the first memories some young animals form and that it is specific enough to enable youngsters to spot their mother in a lineup. Special chemical solutions have been developed which make it possible to locate neurons that have changed as a function of specific learning experiences. Usually, the brain tissue has to be bathed in the chemicals, and the dyed neurons are visible only through a microscope, so the brain must be removed from the animal soon after the learning occurs. Using such methods, researchers identified a specific area in young chicks’ brains that showed extensive structural changes after the chicks imprinted on visual stimuli (Horn, 2004 ). As Gall’s ideas about division of labor in the brain suggest, there appears to be a specialized “mother memory module” that enables chicks to rapidly learn what their mothers look like. Neuroscientists have discovered that neurons in this region change in many ways, depending on the specific experiences chicks have, how much the chicks learn from those experiences, how well the chicks sleep afterward, and whether the neurons are located in the left or right side of the brain (McCabe, 2013 ). Learning who your mom is can require a bit of brain rewiring if you are a bird. Environmental Enrichment Early studies of brain structure in rats found that simply providing young rats with more opportunities for learning, social interactions, and exercise could lead to visible changes in their neurons. Researchers housed one group of rats in an enriched environment , where there was plenty of sensory stimulation and opportunity to explore and learn. For the rats, this meant a large cage filled with toys to play with and other rats with whom to socialize. A second group of rats—the control , or comparison, group— lived in standard laboratory housing, each rat isolated in a small chamber that contained nothing but a drinking spout and food cup. The results? The rats housed in the enriched environment showed better maze learning than the rats kept in standard laboratory housing (Renner & Rosenzweig, 1987 ; Rosenzweig, 1984 ), and as mentioned, these increased learning capacities were associated with structural changes in their neurons. enriched environment An environment that provides sensory stimulation and opportunities to explore and learn; for a rat, this may mean housing in a large cage with many toys to play with and other rats to socialize with. Rats raised in an enriched environment have cortical neurons with more and longer dendrites than experience-impoverished control groups ( Figure 2.6 ). The dendrites of rats in the enriched environment also have more connections with other neurons (Globus, Rosenzweig, Bennet, & Diamond, 1973 ; Greenough, West, & DeVoogd, 1978 ). These neural changes occur quickly: as few as 60 days of housing in an enriched environment can result in a 7 to 10% increase in brain weight of young rats and a 20% increase in the number of connections in the visual cortex. Similar changes are seen in the brains of monkeys and cats raised in enriched environments. Even the brains of fruit flies housed in large communal cages with visual and odor cues show similar changes, compared with flies housed alone in small plastic vials (Technau, 1984 ).  FIGURE 2.6 Deprived environment versus enriched environment Representations of neurons from the cortex of (A) a rat raised in standard laboratory housing and (B) a rat raised in an enriched laboratory environment. Neurons from rats raised in enriched environments typically have more and longer dendrites than experience-impoverished controls. Description The first schematic shows a neuron in a standard laboratory housing, with less dendritic growth. The second schematic shows a neuron in an enriched laboratory environment with greater dendritic growth. Do similar effects occur in humans? Preschool children placed in “high- quality” day care (with lots of toys, educational experiences, and teacher interaction) often fare better in elementary school than children whose day care offers fewer opportunities for learning (Peisner-Feinberg, Burchinal, & Clifford, 2001 ). There isn’t yet definitive evidence that human brains undergo enlargement similar to that of rats after environmental enrichment because the current structural neuroimaging approaches used on children do not have the resolution necessary to detect changes in individual neurons. However, suggestive data come from a study of London taxi drivers. On-the-Job Training London is a sprawling city with hundreds of small, crooked streets. To receive an official license, London taxi drivers must study for up to 3 years and pass a grueling exam that, for example, requires them to indicate the shortest path between random London addresses. This means that licensed London taxi drivers are a group of people sharing an extensive fund of spatial knowledge. Researcher Eleanor Maguire and her colleagues used MRI to compare brain volumes in a group of London taxi drivers with those of age-matched Londoners who had not studied the geography of their city so extensively (Maguire et al., 2000 ). The only part of the brain that differed significantly between the groups was the hippocampus: the taxi drivers had slightly larger hippocampal volumes than non–taxi drivers. Furthermore, the size of the hippocampus differed even among individual taxi drivers: those who had been driving for more than a decade had a larger volume than those who had been driving for only a few years. One possible interpretation of these volume differences is that the intensive spatial learning in taxi drivers causes an increase in dendritic branching in hippocampal neurons—making those neurons take up more room, just like the rat neurons shown in Figure 2.6. Hebbian Learning Any physical change in neurons, or in the systems that support them (such as glia and blood vessels), can affect neural connectivity — the connections between neurons through which brain systems interact. The idea that connections between neurons change during learning was first popularized by Santiago Ramón y Cajal (1852–1934), a Spanish physiologist and anatomist whose work in neuroanatomy won him the Nobel Prize in Physiology or Medicine in 1906. Specifically, Ramón y Cajal theorized that learning involves strengthening or weakening connections between individual neurons (Ramón y Cajal, 1990 ). This same basic idea was also proposed by William James, who, as mentioned earlier, believed that changes in physical connections within the brain determine how memories are linked together. neural connectivity The connections between neurons through which brain systems interact. But how exactly can the right connections between neurons be weakened or strengthened by learning experiences? One of neuroscience’s enduring insights regarding the neural substrates of learning came from Donald Hebb, a Canadian neuroscientist. In one of the most often quoted passages in neuroscience, Hebb wrote, “When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place such that A’s efficiency, as one of the cells firing B, is increased” (Hebb, 1949, p. 62 ). In other words, if two neurons—we will call them neuron A and neuron B—are connected, and if they often fire at nearly the same time, then the connections between them should be strengthened, “wiring” the two neurons together. This would increase the probability that whenever neuron A became active, it would cause neuron B to become active, too. A shorthand version of this “rule” that neuroscientists often use is neurons that fire together, wire together. Learning that involves strengthening connections between neurons that work together is called Hebbian learning. Figure 2.7 shows a simple model of Hebbian learning. Eight hypothetical cortical neurons are shown, each with weak connections to surrounding neurons (Figure 2.7A ). Now let’s assume that some sensory stimulus evokes activation in a subset of these neurons that are tuned to features of this stimulus (solid circles in Figure 2.7A ). As those neurons become active, they produce outputs that are transmitted to other nearby neurons. According to Hebb’s rule—neurons that fire together wire together—the connections between coactive neurons are strengthened as a result. Repeated coactivity of the same subset of neurons, in response to the same stimulus, has a cumulative effect, resulting in the strong connections (heavy lines) shown in Figure 2.7B. Thus, repeated exposure to a stimulus can strengthen connections within a distinctive subset of cortical neurons; this subset can then provide an increasingly reliable basis for identifying the stimulus that is activating them. Changing the connections between cortical neurons creates a pattern that makes a repeated stimulus more likely to be recognized and distinguished from other stimuli.  FIGURE 2.7 A simple model of Hebbian learning Circles correspond to cortical neurons, and lines denote connections between them. (A) Stimulus inputs activate a subset of the neurons (solid circles). (B) Connections between coactive neurons are strengthened (heavy lines). (C) After connections between coactive neurons have been established, an incomplete version of a familiar stimulus may activate just some of the neurons (solid circles) in the subset that represents the stimulus. Activation flows along the strengthened connections and ultimately retrieves the complete stimulus, resulting in the representation shown in part B. Description The first illustration shows stimulus inputs activating a subset of the neurons, represented by 4 solid circles and 4 empty circles connected through lines. The second illustration shows a strengthened connection between coactive neurons. All the lines between the solid circles become heavy lines. The third illustration shows 2 solid circles and 2 empty circles. The two solid circles are connected with two empty circles through heavy lines. Hebbian learning The principle that learning involves strengthening the connections of coactive neurons; often stated as, “Neurons that fire together, wire together.” Hebbian learning also explains how repeated experiences can enhance the ability to recognize familiar stimuli (discussed further in Chapter 3 ). Suppose that once connections have been established between cortical neurons, the organism encounters an incomplete version of a familiar stimulus (Figure 2.7C ). Only some of the neurons that represent that familiar stimulus are activated at first (solid circles in Figure 2.7C ), but the connections already established through repeated experiences will produce outputs that complete the familiar pattern, reconstructing Figure 2.7B. Similarly, recognition of distorted versions of a familiar stimulus, such as might occur when you meet an old friend who has dyed her hair, could also be facilitated by stored patterns encoded as connections between neurons that on previous occasions were simultaneously active. According to Hebb, learning-related changes in connections between neurons are an automatic result of the neurons’ mutual activity and the brain’s capacity for structural plasticity. We now know that Hebb was on the right track. But it was several more decades before technology advanced to the point where researchers could directly monitor such experience-related changes in neural activity. The ability to observe functional changes in brains as individuals learn and perform memory-based tasks has led to a host of new discoveries about how learning and memory work, samples of which we review throughout this book. Interim Summary The brain and spinal cord make up the vertebrate central nervous system (CNS). The brain controls behavior through connections with the peripheral nervous system (PNS), which consists of sensory neurons coming from sensory receptors and motor neurons going to body muscles. The vertebrate brain is made up of several different regions that contribute to learning and memory, including the cerebral cortex, cerebellum, hippocampus, basal ganglia, and amygdala. Neurons, the building blocks of the nervous system, are capable of changing their function and modifying the way they process inputs. Modern structural brain-imaging techniques (including MRI and DTI) provide ways to measure variations in the brain structure of living humans without causing harm. Techniques for imaging neural structures make it possible to collect detailed information about neural changes that occur during learning. Enriched environment studies show that learning experiences can have a profound impact on brain structure, as well as on an individual’s learning and memory abilities. The ability of brains to change with experience is called structural plasticity. Strengthening or weakening the connections between neurons is thought to be a primary mechanism of memory formation. 2.2 Functional Properties of Learning and Memory Systems Modern brain scientists assume that brains are composed of multiple systems that specialize in collecting, processing, and storing particular kinds of information. But there is not always the one-to-one relationship supposed by Gall and other phrenologists, in which each individual function or ability is performed in a dedicated corner of the brain. Instead, one brain area may play a role in many functions, and one function may rely on contributions from many brain areas. What determines how brain regions contribute to learning and memory processes? Two major factors are the kinds of input a region receives and the kinds of output it produces. These inputs and outputs are closely related to the stimuli and responses that behaviorists emphasized in their theories of learning (reviewed in Chapter 1 ). What Brains Do Chapter 1 defines learning as a process by which changes in behavior arise as a result of experience. Thus, when Pavlov’s dogs began to salivate after hearing a sound that predicted food, this change in behavior—salivation in response to a sound—provided evidence that the dogs had learned about the relationship between the sound and the food. But even before Pavlov began using the dogs in his experiments, they would salivate in response to food. Salivation during eating is a reflexive behavior that dogs (and other mammals) develop early in life; it helps the digestive system get ready to process incoming food. A reflex is an involuntary and automatic response that is “hardwired” into an organism; in other words, it is present in all normal members of a given species and does not have to be learned. Just like Pavlov’s dogs, humans salivate when eating food. This is only one of several reflexes that humans are biologically prepared to perform: newborns suck when they encounter a nipple (sucking reflex), hold their breath when submerged underwater (the diving reflex), and grasp a finger so tightly that they can support their own weight by hanging on to it (the palmar grasp reflex). Adults have reflexes, too, such as the knee-jerk reflex when the doctor hits your knee with a rubber mallet and the eyeblink reflex when someone blows air at your eye. Why isn’t this infant drowning? reflex An involuntary and automatic (unlearned) response. Chapter 1 describes René Descartes and his belief that reflexes are hydraulic movements caused by spirits flowing from the brain into the muscles. For many years, scientists accepted this explanation, assuming that there must be some kind of fluid carrying instructions from the brain to the muscles. It wasn’t until the early twentieth century that researchers discovered, first, that there is no such fluid and, second, that the brain is not in absolute control of the muscles at all. Instead of a hydraulic fluid, there are two distinct types of nerve fibers (bundles of axons) connecting the muscles to the spinal cord—one set of fibers carrying sensory signals from the peripheral nervous system into the spinal cord, and a second set carrying motor signals back from the spinal cord to the muscles (Bell, 1811 ; Magendie, 1822 ). If a pinprick or other painful stimulus is applied to a dog’s leg, its leg jerks reflexively (just as you would automatically pull your leg away if someone pricked you). If the sensory fibers are cut, the dog’s sensation of pain disappears, and the reflex fails to occur—although the dog can still move its leg voluntarily. On the other hand, if the motor fibers are cut, the animal can still feel pain but again does not make reflexive leg movements (or voluntary ones). From such observations, scientists deduced that in the spinal cord, sensory fibers are separate from motor fibers. They run in two parallel nerve pathways: one devoted to sensing and the other to responding. This finding, called the Bell– Magendie law of neural specialization , represents the historical first step toward understanding the neural mechanisms of learning. Specifically, it shed light on how the nervous system responds to stimuli and how it controls responses evoked by those stimuli. Following up on the discovery of neural specialization, English physiologist Charles Sherrington (1857–1952) conducted many studies on dogs whose spinal cords had been surgically disconnected from their brains so that the spinal cord no longer received any brain signals. Such surgically altered dogs show many basic reflexes, such as jerking their leg away from a painful stimulus. Because the brain cannot contribute to these reflexes, they must be generated by the spinal cord alone. In fact, we now know that sensory inputs can activate motor fibers traveling out of the spinal cord, without requiring signals from the brain. (The sensory pathways in the spinal cord are largely separate from the motor pathways there, as noted above; yet at the same time, sensory and motor neurons are closely interconnected throughout the nervous system.) If you have ever stuck your hand into dangerously hot or cold water and jerked it away almost before realizing what you have done, or watched your knee jerk in response to the tap of a doctor’s rubber mallet, then you have experienced your spinal cord responding without receiving any help from your brain. Sherrington concluded that such simple “spinal reflexes” can be combined into complex sequences of movements, and these reflexes are the building blocks of all behavior (Sherrington, 1906 ). Sherrington’s description of reflexes differed from that of Descartes in assuming that spinal reflexes did not depend on the brain and did not involve the pumping of spirits or fluids into the muscles. Sherrington received a Nobel Prize in 1932 for his work in this area, and he is now considered to be one of the founding fathers of neuroscience. His ideas provided the groundwork and motivation for Pavlov’s early investigations of reflex conditioning in dogs (Pavlov, 1927 ) and have continued to influence learning and memory researchers ever since. If the spinal cord controls reflexes and if complex actions can be described as combinations of these reflexes, then where does the brain come in? Sensory fibers enter the spinal cord and connect to motor fibers there, but some fibers also travel up to the brain. The brain processes these inputs and produces its own outputs, some of which may travel back down the spinal cord and out to the muscles. The parallel sensory and motor pathways traveling up and down the spinal cord to and from the brain are similar to the parallel sensory and motor pathways that were identified traveling into and out of the spinal cord.  Traditionally, reflexes are presumed to function the same way in men and in women. But what kinds of reflexes might be sex specific? (Chapter 12 considers how genetic differences between sexes can affect learning abilities.) Incoming Stimuli: Sensory Pathways into the Brain We’ll focus first on the sensory pathways that provide inputs to the brain. As noted in Section 2.1 , most sensory inputs enter the brain through the thalamus. The thalamus in turn distributes these inputs to cortical regions specialized for processing particular sensory stimuli, such as the primary auditory cortex (A1), for sound; the primary somatosensory cortex (S1), for sensations from skin and internal organs (such as touch and pain); and the primary visual cortex (V1), for sight. A1 is located in the temporal lobe, S1 in the parietal lobe, and V1 in the occipital lobe ( Figure 2.8 ). Such areas are collectively called primary sensory cortices , as they are involved in the first stage of cortical processing for each type of sensation. Each primary sensory cortex can then transmit outputs to surrounding cortical regions for further processing. For example, the primary visual cortex may start the processing of stimuli from the eye by extracting simple features—say, lines and shading—from a visual scene; later stages of cortical processing elaborate by detecting motion or shape in the scene and, finally, by responding to features of individual objects and their meaning. Damage to primary sensory cortices can eliminate particular perceptual abilities. For instance, people with damage to V1 can become blind, even though their eyes are in perfect working order, and damage to A1 can cause deafness.  FIGURE 2.8 Cortical regions for processing inputs and outputs Specific regions of the cerebral cortex are specialized for processing light (primary visual cortex), sound (primary auditory cortex), and sensation (primary somatosensory cortex). Other regions are specialized for generating coordinated movements (primary motor cortex). Description Primary motor cortex (M1), the primary somatosensory cortex (S1), primary auditory cortex (A) and Primary visual cortex (V1). The primary motor cortex (M1) is located in the dorsal portion of the frontal lobe. The primary somatosensory cortex (S1) is located in the postcentral part of the human brain. The primary auditory cortex (A1) is located bilaterally at the upper side of the temporal lobes on the human brain. The primary visual cortex (V1) is located above the brain stem. Outgoing Responses: Motor Control Just as various brain regions are specialized for processing sensory inputs, other brain regions are specialized for processing the outputs that control movements. Chief of these is the primary motor cortex (M1), which generates coordinated movements. M1 is located in the frontal lobe, adjacent to S1 in the parietal lobe (Figure 2.8 ), and it sends output to the brainstem, which in turn sends instructions down the spinal cord to activate motor fibers that control the muscles. M1 gets much of its input from the frontal lobes, which are responsible for making high-level plans based on the present situation, past experience, and future goals. (Should you pick up that hot coffee cup? Should you try to catch that ball with one hand or two?) Other important inputs come from the basal ganglia and cerebellum, which help to translate the high-level plans into specific sets of movements. All these inputs help determine the outputs that M1 sends to the brainstem. Other motor areas—including the cerebellum, basal ganglia, frontal cortex, and the brainstem itself—also produce their own outputs, all of which converge on the spinal cord and travel from there to the muscles. Complex motor movements—such as picking up a hot coffee cup without spilling the liquid or burning your hand, or picking up an egg without crushing it, or dancing without stepping on your partner’s toes— require exquisitely choreographed interactions between all of these brain structures and the muscles they control. Let’s consider one of these examples in greater detail: you see a cup of coffee and pick it up ( Figure 2.9 ). The process begins with visual input from your eyes traveling to your visual cortex (V1), which helps you find and identify the cup. Regions in your frontal lobes coordinate the necessary plans for grasping the cup, which your motor cortex (M1) then directs by means of outputs through the brainstem, down sets of fibers in the spinal cord, and out to the muscles of the arm and fingers. As you reach for the cup, your basal ganglia and cerebellum continuously track the movement, making tiny adjustments as necessary. These brain regions enable you to exert just the right amount of pressure on the cup: enough to lift it against gravity but not so much that you yank it off the table and spill the contents. As you pick up the cup, sensory information from touch, heat, and pressure receptors in your fingers travels back up your arms, through sensory fibers in the spinal cord, and to the somatosensory cortex (S1), providing evidence that the cup is firmly in your hand. If the handle of the cup is hotter than expected, it could produce a reflexive withdrawal of the hand. This response is the kind of spinal reflex studied by Charles Sherrington; the short path from the hand to the spinal cord and back is sometimes called a reflex arc (a concept introduced by Descartes and described in Chapter 1 ; compare Figure 1.3 with Figure 2.9 ). FIGURE 2.9 How to pick up a cup of coffee (1) Visual input from V1 helps you locate the coffee cup and its handle. (2) The frontal cortex helps you plan the movement. (3) Outputs from the motor cortex (M1) travel through the brainstem and down sets of fibers in the spinal cord to the muscles in the arm, causing you to reach out your hand. (4) The basal ganglia and cerebellum continuously monitor whether your hand is on track, making tiny adjustments to ensure that your hand reaches the correct target. (5) Sensory signals travel back up the arm and spinal cord, through a second set of fibers, to the somatosensory cortex (S1), confirming that the cup has been grasped. Description The parts are numbered from 1 through 5.  The stimulus starts with the Visual cortex (1, V1). It then reaches the Frontal cortex (2), in the front brain, and continues to Motor Cortex (3, M1). Thereon, it is passed to the Basal ganglia (4) and moves to Somatosensory cortex (5, S1). The stimulus is then passed through the spinal cord. The motor control signals are transmitted from the brain through the spinal cord, to the hands of the man, and the sensory inputs are transmitted from the nerve ends in the fingers to the brain through nerves. All that input and output just to pick up a cup—before you have even taken your first sip! Infants of many vertebrate species, including humans, are born fairly clumsy and spend a large part of their infancy and childhood learning how to walk or fly or swim gracefully, reach accurately, move throat and tongue muscles to produce coherent sounds, and so on. This relatively long period spent learning coordinated motor control reflects both the complexity of the operations and the many brain structures that have to interact with one another and with the outside world to perform those operations. Observing Brains in Action Because learning is a process that can lead to changes in behavior, and brains control behavior through changes in neural activity, it is clear that learning must be associated with new patterns of activity in the brain. However, knowing that your brain is doing something different after years of practicing a skill or after experiencing a traumatic event is a far cry from knowing what it is doing differently or why. Even if structural imaging techniques reveal that experience has led to physical changes in parts of neurons or to increases in the volume of a brain region, understanding how these changes contribute to performance is not straightforward. To gain a clearer understanding of how experiences change brain function, neuroscientists attempt to monitor specific changes in activity that occur within the central nervous system before, during, and after such experiences. Functional Neuroimaging As noted earlier, structural neuroimaging methods (such as MRI) allow researchers to look at the structure of a living human brain, whereas functional neuroimaging allows them to look at the activity , or function, of a living brain. For example, when a brain structure becomes active, it requires more oxygen. Within 4 to 6 seconds, blood flow (with its cargo of oxygen) increases to that region. On the other hand, when a brain structure becomes less active, it requires less oxygen, and blood flow decreases. By tracking local changes in blood flow (or oxygen usage), researchers can infer which brain regions are more or less active. functional neuroimaging Techniques (such as fMRI) for observing the activity or function of a living brain. Rather than focus on where blood use is heavier in the brain, functional neuroimaging studies typically examine how blood distribution or usage in a particular brain region changes depending on what the person is doing or thinking. To see such changes in blood distribution, researchers may first scan the brain while the person is relaxed and not doing anything. Of course, even though the person is not deliberately performing any task, the brain is still active. Next, the researchers scan the brain while the person performs a given task, such as looking at pictures or reading a story. (The pictures or words are projected onto the inside ceiling of the scanner so that the person can see them while lying on his or her back.) During the task, some areas of the brain should become more active than before. Other brain areas might decrease in activity. The images are often color coded, with white, red, or yellow indicating areas where blood flow increased most during the task relative to the baseline. Colors such as blue and green may indicate where blood flow decreased most during the task. Researchers typically do not rely on measurements from a single person to decide which brain regions are most likely to show changes in activity levels during performance of a particular task. Instead, they usually collect data from multiple individuals and then analyze patterns from the whole group. This approach emphasizes differences in brain activity that are prevalent across many participants, but it does not necessarily reveal all the changes in activity that occur within any particular individual. Through these methods, scientists can discover differences in brain activity that are associated with different kinds of memory tasks (for instance, recognizing faces versus recalling what happened at a recent party) and differences associated with successful recall versus memory failures. The most commonly used functional neuroimaging technology is functional magnetic resonance imaging (fMRI). It uses the same MRI technologies employed for structural imaging described in Section 2.1 , but with fMRI the focus is on differences in oxygen levels in the blood. Oxygenated blood produces slightly different signals than deoxygenated blood, so there are fluctuations in the signal received from areas of the brain that undergo changes in activity levels during task performance. functional magnetic resonance imaging (fMRI) A method of functional neuroimaging based on comparing an MRI of the brain during performance of a task with an MRI of the brain at rest. fMRI is a powerful tool for observing the brain in action, but it only indirectly measures neural activity. Also, because functional neuroimaging studies typically focus on task-related changes in activity, they tend to emphasize associations between specific brain regions and particular functions (in the spirit of Gall) rather than look at the full range of brain activity that contributes to learning and memory processes. Finally, current functional neuroimaging techniques are relatively slow: fMRI allows images to be taken every few seconds, but changes in brain activity occur much more rapidly than that. Tracking learning- or memory-related changes in neural activity in real time requires other technologies. Electroencephalography Electroencephalography (EEG) is a method of measuring electrical activity in the brain. (The Greek word enkephalos means “brain,” so “electro- encephalo-graphy” means drawing or graphing the electrical activity of the brain.) Electrodes placed on a person’s scalp measure the combined tiny electrical charges produced by millions of neurons, especially those near the location on the skull where the electrodes are placed. The resulting picture is called an electroencephalogram (also abbreviated as EEG).  To record neural activity from humans as they perform tasks, researchers attach multiple electrodes to a person’s scalp. Traditionally, signals from electrodes are collected from wires attached to the electrodes (as shown here for the infant), but recent wireless technologies make it possible to record neural activity in any setting. How might EEG be used to measure functional brain changes that have occurred in the adult model as a result of her modeling experiences? Description The photo on the left shows multiple electrodes and wires attached to a baby's head. The photo on the right shows electrodes and wires attached to a person’s scalp. electroencephalography (EEG) A method for measuring electrical activity in the brain by means of electrodes placed on the scalp; the resulting image is an electroencephalogram (also EEG ). Just as blood is always flowing through the brain, so is electrical activity, driven by the firing patterns of neurons. The exact pattern of activity changes depending on what the brain is doing. For example, when a tone sounds, sensory receptors in the ear become active, and signals travel to the primary auditory cortex (A1), affecting electrical activity there. But detecting this particular electrical change in an EEG is difficult because lots of other neurons in other brain areas that are not involved in hearing may also be active—those responding to whatever visual stimuli happen to be in front of you, for instance, or those activated as you wiggle your fingers and think about what you want to have for lunch. To detect an electrical change associated with hearing a sound, such as a tone, researchers typically present the same sound hundreds of times and then average the EEGs produced during those repetitions. Activity in other brain areas will come and go, but only the neurons responding to the specific sensory stimulus will be consistently activated each time the sound is repeated—and so only their activity patterns will survive the averaging process. EEGs averaged across many repetitions of the same event are called event- related potentials (ERPs). Just as functional neuroimaging shows how the brain changes while performing a task, ERPs can be used to show different brain states at different stages of learning or memory processing, such as how a person’s brain responds to different sounds as the person gradually learns to make subtle distinctions between them (discussed in Chapters 3 and 6 ). event-related potential (ERP) Electroencephalograms (EEGs) from a single individual averaged over multiple repetitions of an event (such as a repeated stimulus presentation). Compared with fMRI, EEG recording is a simple and inexpensive way to monitor changes in brain activity during learning and memory tasks. In addition, EEG can detect rapid changes in the brain with more temporal precision than fMRI. Yet what EEG gains in temporal precision, it often sacrifices in spatial precision. Whereas fMRI can localize activation to within a few millimeters, EEG signals show activity over a wide swath of the brain. To take advantage of the strengths of both technologies, some memory researchers are combining functional neuroimaging and EEG methods to generate images that show precisely when and where neural activity occurs during memory storage and recall by humans. Recording from Neurons Memory functions are affected not only by which neurons fire but also by how often they fire. Neuroimaging and EEG studies can reveal the contributions of large areas of the brain to learning and memory, but they do not reveal much about which individual neurons are firing or how often. To gather this information, researchers have to record neural activity directly. Neurophysiology is the study of the activity and function of neurons. neurophysiology The study of the activity and function of neurons. One technique scientists use to measure the firing patterns of individual neurons is single-cell recording (where the single cell in this case is a neuron). The microelectrodes that are employed in this method function somewhat like EEG electrodes, but they are much tinier and can penetrate brain tissue without too much damage. A microelectrode can be inserted in brain tissue until its tip is very close to, or sometimes even inside, a target neuron. In some cases, researchers anesthetize an animal and surgically implant one or more microelectrodes in the brain areas they wish to study. Then, when the animal wakes, the researchers can record from the neuron(s) as the animal goes about its daily business. Most animals do not seem to be much bothered by, or even aware of, the wires connected to their heads. Such experiments allow researchers to determine what role a given neuron or network of neurons might play in the animal’s behavior. Alternatively, if the researcher is interested in looking more closely at how individual neurons interact, it is possible to remove pieces (or “slices”) of a brain, keep the neurons alive in a bath of nutrients, and record their activity. single-cell recording Use of an implanted electrode to detect electrical activity (spiking) in a single cell (such as a neuron). Single-cell recordings have provided some of the most dramatic evidence to date of how neural firing relates to behavior. For example, Apostolos Georgopoulos and colleagues recorded neural firing patterns from the motor cortex of a monkey while the monkey moved a joystick in different directions ( Figure 2.10A ; Georgopoulos, Taira, & Lukashin, 1993 ). Some neurons fired most strongly when the monkey pushed the lever in a particular direction. Figure 2.10B shows recordings from one such neuron as the monkey moved the lever toward different compass points. FIGURE 2.10 Recording from single neurons (A) Researchers implanted recording electrodes into the motor cortex of a monkey that was then trained to move a joystick in different directions. (B) These recordings of a neuron spiking (producing vertical lines) when the monkey moved its arm show that the neuron fired most when the monkey moved its arm toward position 1 and least when it moved its arm toward position 5. Thus, this neuron is tuned to fire during movements away from the monkey’s body. (B) Information from Georgopoulos et al., 1993. Description Section A shows a drawing of a monkey with an electrode implanted in the motor cortex. The monkey is working a joystick that has positions 1 through 8 marked radially around it. Section B shows 8 positions of the joystick and the recorded response rate associated with each position. Spikes represent the response rate. Position 1, at the bottom, has the most spikes. The spikes grow fewer as the joystick moves away from Position 1. Each vertical line in the recording shown in Figure 2.10B represents one action potential, sometimes referred to as a spike. When the monkey moved its arm toward the point labeled 6 in Figure 2.10A , the neuron initially produced several spikes, and then it fell silent. When the monkey moved its arm to a slightly different position, point 7, the neuron produced a more sustained burst of activity, and it continued to spike for the duration of the movement. But when the monkey moved its arm directly away from its body, toward point 1, the neuron really went into action, spiking as fast and frequently as it could. By contrast, when the monkey moved its arm in the opposite direction, toward its body (point 5), the neuron was much less active. Thus, this neuron’s firing patterns are correlated with arm movements, and neuroscientists would say it is specialized, or “tuned,” to fire maximally during movements in a particular direction: away from the body. Georgopoulos and colleagues found that other neurons in the motor cortex are tuned to fire during arm movements in other directions. Given what we know about the motor cortex from functional imaging studies, it is reasonable to assume that these neurons may be playing a direct role in issuing the commands that cause the monkey’s arm to move. Because monkeys generally must be trained to move a joystick in different directions in laboratory experiments, such recordings can potentially reveal how the firing patterns of neurons change as monkeys learn to perform such tasks. In fact, such research has led to new technologies that enable both monkeys and humans to learn to control the movements of robotic arms simply by thinking about where they want the arm to move (discussed in Chapter 8 ). Single-cell recordings can also be collected from humans when doctors need to clinically monitor patients with epilepsy before performing corrective neurosurgery. This practice makes it possible to observe neural activity as a person remembers specific episodes (Suthana & Fried, 2012 ). How Remembering Changes Brain Activity At the start of this chapter, we described the case of Orlando Serrell, who gained remarkable memory abilities after sustaining a blow to the head in a baseball game. Orlando is not the only person to possess phenomenal memory capacities. By studying people with an exceptional ability to remember specific events, researchers can gain unique insights into the systems that make memory possible. Some people who excel at remembering compete in annual competitions to see who can memorize the most information quickly and accurately—things like sequences of numbers or playing cards, names and faces, and so on. These memory champions have trained their brains to absorb details and retain them at least long enough to recite or write them down during the recall portion of the contest. Most such memory experts have learned to use specific strategies or mnemonics to do this (discussed in Chapter 7 ). But memory savants, like Orlando, seem to perform amazing feats of memory involuntarily, without needing to practice, and their memories are typically for very different kinds of information than the items competitors excel at memorizing. One group of memory savants that has received attention in the past decade is made up of people with highly superior autobiographical memory (HSAM ), or hyperthymesia (LePort et al., 2012 ). These individuals appear to be able to access memories of events that happened on almost any day of their lives and can recall seemingly trivial details such as the meals they ate each day. Think about what you did yesterday: whom you met, what you ate, where you went, and what you wore. Now imagine being able to remember any day from the past 10 years with the same level of detail. This is what people with HSAM seem to be doing just as easily as you can envision yesterday. For other kinds of memory, though, people with HSAM perform like everyone else. They do not have “photographic memories.” So, what is different about the brains of people with HSAM? Are they experiencing events in ways that make those events more memorable than they would be for the typical person? Or perhaps their experiences are normal but they have an exceptional ability to access and relive memories of those experiences. It is hard to tell based only on their abnormal accuracy or on their descriptions of what remembering is like for them. Recent functional neuroimaging studies provide a few clues, however, about what is going on in their brains. To see what is happening in people with HSAM when they remember past events, researchers gave them prompts like “the first time you drove a car” and then monitored their brain activity using fMRI while the episode was being remembered (Santangelo et al., 2018 ). They then compared the activity recorded during the remembering of autobiographical events with brain activity that occurred in response to prompts like “examples of vegetables.” The researchers discovered that when remembering past events, the savants showed increased activity in several different brain regions, especially the prefrontal cortex (a subregion of the frontal lobe discussed in Chapters 7 and 9 ), the hippocampus, and a region of cortex at the junction of the frontal and parietal lobes ( Figure 2.11 ). Interestingly, the specific patterns of activation varied considerably across individuals with HSAM, contrary to what you (or Gall) might expect if specific regions of cortex were specialized for remembering personal episodes. FIGURE 2.11 Brain regions active during remembering When people with highly superior autobiographical memory (HSAM group) recall specific days from their past, activity increases in multiple cortical regions, including areas in the frontal lobe, parietal lobe, and temporal lobe. Similar regions become active when typical adults recall past episodes (control group), but the changes in brain activity are not as extensive. Activity recorded by fMRI while participants recalled generic facts was subtracted from activity recorded when participants remembered specific days. Therefore, the images do not show all the brain regions that were active while people reminisced, just those that increased in activity in comparison with a different memory task. Also, these differences are averaged across multiple participants to show only those regions that were consistently more active. Description  The first illustration shows the left hemisphere. Overlapping regions cover the major parts of the left hemisphere. The second illustration shows the top view. The overlapping regions and the other highlighted regions are spread throughout. The third illustration shows the right hemisphere. The overlapping regions are at the top center part of the brain. The other highlighted regions are in the anterior part of the brain. Comparing brain activity during recall of personal episodes to brain activity during recall of facts shows that something different is going on when people with HSAM remember things that happened to them. But is this true only for people with HSAM? Would you see the same kinds of differences if you scanned your own brain activity? To learn whether that might be the case, researchers conducted the same fMRI study using a control group consisting of people without super memory. This comparison revealed that people with HSAM were recruiting more cortical regions than the control participants were using when remembering experienced events (Figure 2.11 ). However, this difference showed up only when participants were initially accessing the personal memories. When participants went on to think about the details of the memory, brain activity was comparable for both groups. This could mean that the way people with HSAM access past personal memories is somehow unique. One complication of simply asking people to “think back” to events that may be long past is that it can be difficult to know how accurate a person’s memory for a particular day really is. A memory may feel like a true record of what really happened, but feelings of certainty are no guarantee of accuracy. (Chapter 7 has more to say about false memories that feel more real than they are.) New personal recording technologies like GoPro enable people to document episodes in their lives, and researchers are beginning to incorporate such devices into fMRI studies, to help trigger and control the personal memories of research participants (Chow & Rissman, 2017 ). Segments of scenes recorded from the viewpoints of study participants or of other individuals can be used to prompt recall of specific episodes and to test participants on their ability to tell whether an event was one they personally experienced. Such tasks activate networks of brain regions that are similar to those seen in people attempting to recall episodes such as their first kiss, providing further support for the idea that many different brain regions work together to make remembering the past possible. Interim Summary Reflexes are natural, automatic responses to stimuli. Sherrington and other early neuroscientists believed that all complex learning involves combining simple spinal reflexes. In the brain, sensory signals (produced by stimuli) are initially processed in cortical regions specialized for processing such signals, and they ultimately lead to activity in other cortical regions, such as the motor cortex, that are specialized for coordinating movements (responses). Functional neuroimaging methods (such as fMRI) allow researchers to track brain activity during the performance of memory tasks by measuring increases and decreases in activity in different brain regions. Electroencephalographic recordings make it possible to track the activity of large populations of neurons over time and to monitor how such activity changes as learning progresses. Single-cell recordings allow researchers to directly monitor and record the electrical activity (or “firing”) of single neurons and to study changes in their firing patterns that occur during learning or the recall of memories. Brain-imaging studies of memory savants and others show that the act of remembering experienced events engages a broad network of brain regions. 2.3 Finding and Manipulating Memories We have discussed two ways in which experience can change the brain: structurally, by changing properties of neurons and their connections; and functionally, by changing patterns of activity in neural circuits. These two kinds of change are closely related because changes in neural activity can lead to physical changes in brain tissue (as proposed by Ramón y Cajal and James), and physical changes partly determine future patterns of neural activity. We have also touched on the idea that different brain regions may be specialized for different kinds of learning or memory functions (an idea that Gall got mostly right!). Baby chicks show localized structural changes in their brains when they imprint on their mothers, and undergraduates show localized changes in brain activation when they recall episodes from their personal lives. Such findings might give you the impression that neuroscientists have solved the riddle of how brains store memories. Some do make this claim, but unfortunately, they do not all agree on what the answer is. Debate rages on about how memories are stored by the brain and especially about where they are stored. Looking for Memories Most neuroscientists are on board with the idea that memories are stored in certain brain regions, and most assume that within those brain regions, memories are stored by physical properties of neurons. There are differing opinions, however, about where specific kinds of memories might be stored, how long they might be stored in a particular location, whether they might change location over time, and whether different people might store them in different locations. Similarly, there are several different proposals about what parts of neurons, or sets of neurons, are the parts that store the memories. Two themes that have dominated research on the brain mechanisms that support learning and memory (in other words, on the neural substrates of learning and memory) are that many memories are stored in the cerebral cortex and, more specifically, that they are stored in connections between cortical neurons. Searching for Memories in the Cerebral Cortex Imagine that a Martian scientist comes to Earth and encounters an automobile, a method of transportation unknown on Mars, powered by an energy source also unknown to Martians. If the Martian speaks no Earth languages and can’t simply ask a mechanic for an explanation, how might she learn how the car works? One way would be to look under the hood and examine the many components there. But studying the car’s “structure” would only get her so far; to learn about the car’s function, she would have to take it for a test drive and see how it behaves when it is operating properly. However, simply seeing cars in action cannot reveal what makes them go. One approach the Martian might use to better understand cars would be to investigate what the different parts do. For instance, she could try disconnecting or removing parts one at a time, noting the consequences in each case. If she removes the axle, she will find that the motor works but cannot transfer energy to make the wheels turn. If she removes the radiator, she will find that the car runs but then quickly overheats. In the end, by discovering the function of each of the car parts, the Martian could probably develop a pretty good idea of how the car works. Neuroscientists trying to understand how nervous systems make it possible for organisms to learn and remember face a challenge similar to the Martian’s. No surprise, then, that one of the earliest approaches researchers attempted was akin to the Martian’s: they examined people whose brains were missing pieces to see how such losses could affect performance. Although no scientist would disassemble the brain of a living human the way the Martian might disassemble a car, humans regularly suffer damage to one or more brain areas—through accident, injury, or disease—making it possible to explore the effects that damaged brain regions can have on learning and memory abilities.  Brain injuries can lead to the loss of large portions of brain tissue. In this MRI image, missing cerebral cortex appears as a dark region on the right side of the image. Is this lesion more likely to be in the temporal lobe or the occipital lobe? Neuropsychology is the branch of psychology that deals with the relationship between brain function and behavior, usually by examining the functioning of patients who have specific types of brain damage. These individuals volunteer their time and effort in experiments that test their learning and memory abilities, as well as other kinds of cognitive function— language, attention, intelligence, and so on. The test results can potentially be used to guide a patient’s rehabilitation, but they also serve a research purpose. By recognizing patterns in the impaired and spared abilities of a group of patients who have experienced damage to a similar region of the brain, researchers hope to build a better picture of that brain region’s normal function—just like the Martian trying to understand what a radiator does by watching what happens to a car that does not have one. neuropsychology The branch of psychology that deals with the relationship between brain function and behavior. Animal researchers have conducted parallel studies by removing or deactivating specific brain regions to create animal “models” of humans with brain damage. Because human brain damage is almost always caused by accident, injury, or illness, every patient’s damage—and disability—is slightly different. By contrast, in animal models, researchers can remove or disable specific brain regions with great precision, making it much easier to compare results across individuals. Instances in which the experimental results from human patients and animal models converge give the clearest picture of how the brain works normally and how it functions after damage. Some of the most famous experimental brain lesion studies of learning and memory were conducted by Karl Lashley (1890–1958), an American psychologist who was looking for the location of the engram — the supposed physical change in the brain that forms the basis of a memory (also referred to as a memory trace ). Lashley would train a group of rats to navigate a maze, and then he would systematically remove a different small area (covering, say, 10%) of the cortex in each rat. He reasoned that once he had found the lesion that erased the animal’s memories of how to run through the maze, he would have located the site of the engram (Lashley, 1929 ). engram A physical change in the brain that forms the basis of a memory. Alas, the results were not quite so straightforward. No matter what small part of the cortex Lashley lesioned, the rats kept performing the task. Bigger lesions would cause increasingly large disruptions in performance, but no one cortical area seemed to be more important than any other. Hence, Lashley could not find the engrams for memories formed during maze learning. Finally, in mock despair, he confessed that he might be forced to conclude that learning “simply is not possible” (Lashley, 1929 ). Eventually, Lashley settled on a different explanation. He endorsed the theory of equipotentiality , which states that memories are not stored in one area of the brain; rather, the brain operates as a whole to store memories. Although Lashley is often credited with formulating this theory, it was actually first proposed in the 1800s as an alternative to phrenology (Flourens, 1824 ). In the theory of equipotentiality, memories are spread over many cortical areas; damage to one or two of these areas will not completely destroy the memory, and over time the surviving cortical areas may be able to compensate for what has been lost. theory of equipotentiality The theory that memories are stored globally by the brain as a whole rather than in one particular brain area. Lashley’s work and his endorsement of the theory of equipotentiality were milestones in the neuroscience of memory because researchers could no longer take for granted the compartmentalized cortical regions that Gall had popularized. But Lashley was only partly correct. Gall was on the right track when he proposed that different brain areas have different specialties; the specialization just wasn’t as extreme as he had thought. Lashley, too, was on the right track when he proposed that engrams are not stored in specific areas of the cortex, but we now know that the cortex is not quite as flexible as he came to believe. The truth is somewhere in the middle. Part of the reason Lashley’s experiments did not work out the way he expected was because of his assumption that memories formed during maze learning were stored only in the cerebral cortex. If Lashley had instead made his lesions beneath the cortex, he might have discovered that other brain regions (such as the hippocampus) more strongly affect spatial learning and memory. (The role of the hippocampus in spatial learning is discussed in Chapters 3 and 7.) Useful as brain lesion experiments are, they are limited in terms of what they can reveal. Suppose a researcher lesions part of a rat’s cortex and then finds, as Lashley did, that the rat can still learn how to get around in a maze. Would that prove that the lesioned cortical area is not involved in spatial memory? Not necessarily; the rat may now be learning the maze in a different way. This would be analogous to your being able to find your way around a house with the lights out, even though you use visual input when it is available. Data from lesion studies are strongest when supplemented by data from other techniques showing that a brain region normally participates in a given behavior. Showing that a particular brain region needs to be intact for learning and memory processes to work normally also does not mean that memories are stored in that region. Your laptop’s memory will not work without a power source, but that does not mean it is storing memories in its battery or power cable. To establish that a memory has been stored in a particular part of your laptop, you would need to show that specific kinds of changes in that part always occur when such a memory is stored; that the changes remain stable as long as the memory is stable; that the changes are specific to the content of the memory that was stored; and that when the memory is retrieved, retrieval processes access the changes made in that part. Researchers use similar tests to try to identify engrams in brains. Usually, these tests focus specifically on the role of physical changes that occur along dendrites. TEST YOUR KNOWLEDGE Equipotentiality Versus Phrenology What are the main differences between the explanations of brain function proposed by Franz Joseph Gall and those ultimately proposed by Karl Lashley? What evidence did each use to support his viewpoint? (Answers appear at the end of the chapter.) The Synapse: Where Neurons Connect So far, we have been describing the transmission of signals into and out of the brain as if these signals flow from one place to another in the nervous system like water through a pipe (reminiscent of the way Descartes described the mechanisms of behavior). What really happens is that neurons throughout the nervous system are continually communicating with one another in vast networks that are similar in some ways to social networking systems such as Twitter or Facebook. This communication between neurons makes learning and memory possible. Generally, neurons that communicate with each other are not actually physically connected. Rather, communicating neurons are separated by a narrow gap of about 20 nanometers (1 nanometer is one-billionth of a meter), called a synapse , across which the neurons pass chemicals ( Figure 2.12A ). Most synapses are formed between the axon of the presynaptic (or sending) neuron and a dendrite of the postsynaptic (or receiving) neuron, but synapses can also be formed between an axon and a cell body, between an axon and another axon, and even between dendrites. FIGURE 2.12 Transmission across a synapse (A) This photo (taken through an electron microscope) shows the tiny gaps, or synapses, between neurons. Vesicles filled with neurotransmitters, ready for release into the synapse, are visible as circular packets inside the presynaptic neuron. (B) A signal is transmitted between neurons when (1) the presynaptic neuron releases neurotransmitter into the synapse and (2) the neurotransmitter molecules dock at receptors on the surface of the postsynaptic neuron. This may activate the receiving neuron.  Leftover neurotransmitter in the synapse is either (3) broken down or (4) reabsorbed into the presynaptic neuron. Description Section A shows a micrograph with the following parts labeled: Dendrite, Synapse, Vesicle, and Axon. Section B shows the release of neurotransmitter molecules from a presynaptic neuron. The regions labeled are as follows. Axon of presynaptic neuron lies at the tip of the neuron. The rounded structures within the axon of the presynaptic neuron are labeled vesicles containing neurotransmitters. The steps are numbered from 1 through 4. The vesicles along the bottom periphery of the axon are open, and the neurotransmitter molecules are released from it (1). These molecules move toward the receptors in the area labeled dendrite of the postsynaptic neuron. The receptors accept the neurotransmitter molecules (2). The unaccepted neurotransmitter is broken down (3) or reabsorbed (4). synapse A narrow gap between two neurons, across which chemical messages can be transmitted. presynaptic On the sending side of a synapse. postsynaptic On the receiving side of a synapse. Neurons contain neurotransmitters , chemical substances that can cross a synapse to affect the activity of a postsynaptic neuron. Neurotransmitters are kept conveniently on hand at the end of the presynaptic axon, in packets known as vesicles. To transmit a signal, one or more vesicles of the presynaptic axon release neurotransmitters into the synapse (Figure 2.12B ). Several different chemicals act as neurotransmitters. Major ones include glutamate, gamma-aminobutyric acid (GABA ), acetylcholine, dopamine, norepinephrine, epinephrine, and serotonin. Once neurotransmitters have been released into the synapse, the next step is for the postsynaptic neuron to collect them. Receptors are molecules embedded in the surface of the postsynaptic neuron that are specialized to bind with and respond to particular kinds of neurotransmitters. neurotransmitter One of several classes of molecule released by neurons to carry chemical messages to other neurons. receptor A specialized molecule, located on the surface of a neuron, to which one or more particular neurotransmitters can bind; when a neurotransmitter activates a receptor, effects may be initiated in the neuron. The effect of a particular neurotransmitter depends on what its corresponding postsynaptic receptors do when activated. Some receptors open a channel for the flow of electrically charged particles into or out of the cell, thus changing the charge characteristics in a small area of the neuron. Similar electrical changes may be occurring simultaneously in other locations on the neuron as other receptors on other dendrites become active. When a neuron fires, sending an electrical charge to the end of its axon, it causes the release of neurotransmitters there. Some neurotransmitters—glutamate, for example—are excitatory , activating receptors that tend to increase the likelihood of the postsynaptic neuron firing. Other neurotransmitters—such as GABA—are inhibitory , activating receptors that tend to decrease the likelihood of the postsynaptic neuron firing. A neuron may produce and release only one kind of neurotransmitter or may release more than one kind. In either case, the neuron may be able to respond to signals from many different presynaptic neurons, each releasing a different kind of neurotransmitter. After a neuron fires, there is a brief period, called a refractory period , during which it cannot fire again, no matter how much input it receives. When this refractory period has passed, the neuron is again open for business. If the neuron is still receiving a lot of input from its neighbors, it may fire again and again in rapid succession, interrupted only by the refractory period after each action potential. If the excitatory inputs are less frequent or less strong, or if there is a lot of inhibitory input, some time may pass before the neuron fires again. In the meantime, neurotransmitter molecules have to be cleared out of the synapse so that the synapse can receive future signals. In some cases, this consists of breaking down the neurotransmitter molecules into their constituent parts through a process called inactivation. In other cases, they are brought back into the presynaptic neuron and recycled for future use in a process called reuptake. When cleanup is complete, the synapse and receptors are ready to receive new transmissions. Several areas in the brainstem contain neurons that send axons widely throughout the brain; when these neurons fire, they release neurotransmitters called neuromodulators that can affect activity in entire brain regions rather than just at a single synapse. Neuromodulators alter, or modulate, how neurons transmit and receive signals, although they themselves are not part of the signal. For example, acetylcholine often functions as a neuromodulator, and one of its effects is to temporarily alter the number of receptors that must be active before a postsynaptic neuron can fire. If you think of synaptic transmission as a message, then acetylcholine levels help determine whether the message is heard as a whisper or a shout. Many human diseases that affect learning and memory seem to involve a global decline in neuromodulators. Examples include Alzheimer’s disease, which is associated with a reduction in acetylcholine (Francis, Palmer, Snape, & Wilcock, 1999 ), and Parkinson’s disease, which is characterized by a reduction in dopamine (Evans & Lees,