The Disordered Mind

by Eric R. Kandel

  For a long time Milner thought that H.M.’s memory deficit applied to all areas of knowledge. Then she made a remarkable discovery. She asked H.M. to trace the outline of a star while looking at his hand, his pencil, and the paper in a mirror. Everyone who tries this tracing task makes errors on the first day, drawing outside the line of the star and having to adjust back in, but people with normal memory improve to almost perfect performance by the third day. If H.M.’s memory loss applied to all areas of knowledge, he should show no such improvement. Yet after three days, and despite having no memory of practicing the task or of having seen Milner before, H.M. had learned this motor task as well as anyone else (fig. 5.4).

  Because H.M. could not remember having practiced, scientists speculated that motor learning, unlike every other form of learning, must involve a special form of memory. It must be mediated by other systems in the brain.

  Neuroscientists thought this for a very long time—until Larry Squire at the University of California, San Diego, found that people with damage to the medial region of both temporal lobes (the same areas removed in H.M.) can learn more than motor skills. Their capability for language is normal, and they can perform a whole family of learned perceptual skills as well, such as reading mirror-reversed print. They can also acquire habits and other simple forms of learning. If this range of learning capabilities remained, Squire reasoned, then maybe these people were relying on a different kind of memory system.1

  Figure 5.4. Learning a motor task

  Squire came to realize that there are two major memory systems in the brain. One is explicit, or declarative, memory, which allows us to consciously remember people, places, and objects. This is what we mean when we refer to “memory” in everyday language. It reflects our conscious ability to remember facts and events. Explicit memory relies on the medial region of the temporal lobe, which explains why H.M. could no longer remember new facts or people or the events of his passing days.

  The second type of memory, the memory that Squire identified, is implicit, or non-declarative, memory, which our brain uses for motor and perceptual skills that we do automatically, like driving a car or using correct grammar. When you speak, you are usually not conscious of using correct grammar—you just speak. What makes implicit memory so mysterious—and the reason we rarely pay attention to it—is that it is largely unconscious. Our performance of a task improves as a result of experience, but we are not aware of it, nor do we have the sense of using memory when we perform the task. In fact, studies show that performance on implicit tasks can actually be impaired when we consciously contemplate the action.

  Not surprisingly, implicit memory depends on different brain systems than explicit memory does. Rather than relying on higher, cognitive regions such as the medial region of the temporal lobe, implicit memory depends more on regions of the brain that respond to stimuli, for example the amygdala, the cerebellum, and the basal ganglia, or, in the simplest instances, the reflex pathways themselves.

  A particularly important subclass of implicit memory is evident in memory that is associated with conditioning. Aristotle was the first person to suggest that certain types of learning require the association of ideas. For example, whenever you see a tree covered with lights you think of Christmas. This notion was elaborated and formalized by the British Empiricists John Locke, David Hume, and John Stuart Mill, the forefathers of modern psychology.

  In 1910 the Russian physiologist Ivan Pavlov extended this idea one critical step further. In earlier studies of dogs, he had noticed that the animals began to salivate when he entered the room, even when he was not carrying their food. In other words, the dogs had learned to associate a neutral stimulus (his entering the room) with a positive stimulus (food). Pavlov called the neutral stimulus a conditioned stimulus and the positive stimulus an unconditioned stimulus—and he called this form of associative learning conditioning.

  Based on his observation, Pavlov designed an experiment to see whether a dog would learn to salivate in response to any signal that predicted the arrival of food. He rang a bell and then gave the dog food. At first, ringing the bell produced no response. After several instances of pairing the sound of the bell with food, however, the dog salivated in response to the ringing of the bell, even when no food was forthcoming.

  Pavlov’s work had an extraordinary impact on psychology: it marked a decisive shift toward a behavioral concept of learning. To Pavlov, learning involved not only an association between ideas but also an association between a stimulus and behavior. This made learning amenable to experimental analysis: responses to stimuli could be measured objectively, and the parameters of a response could be specified or even modified.

  Squire’s discovery that memory is not a unitary function—that different kinds of memories are processed in different ways and stored in different regions of the brain—was a major advance in our understanding of memory and in our understanding of the brain, but it inevitably raised a new set of questions. How do neurons store these different types of memories? Are different cells responsible for implicit and explicit memories? If so, do they operate differently?

  MEMORY AND THE STRENGTH OF SYNAPTIC CONNECTIONS

  Early studies assumed that it takes a fairly complex neural circuit to form and store a memory of what we have learned. However, my colleagues and I at Columbia University and Jack Byrne, one of my former students, now at the University of Texas Health Science Center at Houston, encountered a mechanism of associative learning in the invertebrate marine snail Aplysia that does not require a complex neural circuit.2 Aplysia has an important defensive reflex that is mediated by the connections between a small number of sensory neurons and motor neurons. Learning leads to the activation of modulatory neurons, which strengthen the connections between the sensory and motor neurons. My colleagues and I found that this mechanism contributes to implicit learning of classical conditioning in invertebrate animals. It also operates in the amygdala, the brain structure in mammals that is critical for implicit learning of emotion, particularly fear.

  Another person who challenged the idea that a complex neural circuit is required for learning was the Canadian psychologist Donald Hebb. Hebb suggested that associative learning could be produced by the simple interaction of two neurons: if neuron A repeatedly stimulates neuron B to fire an action potential—the electrical impulse that travels down the axon to the synapse—a change will take place in one or both of those cells. That change strengthens the synaptic connection between the two neurons. The strengthened connection creates and stores, for a short time, a memory of the interaction.3 Two researchers working at the University of Göteborg in Sweden, Holger Wigström and Bengt Gustafsson, later provided the first evidence suggesting that Hebb’s mechanism might be at work in the formation of explicit memory in the hippocampus.4

  Both implicit and explicit memory can be stored in the short term, for minutes, and in the long term, for days, weeks, or even longer. Each form of memory storage requires specific changes in the brain. Short-term memory results from strengthening existing synaptic connections, making them function better, whereas long-term memory results from the growth of new synapses. Put another way, long-term memory leads to anatomical changes in the brain, whereas short-term memory does not. When synaptic connections weaken or disappear over time, memory fades or is lost.

  MEMORY AND THE AGING BRAIN

  Thanks to a broad range of medical advances, the average American born today is expected to live for about eighty years, in contrast to only fifty years in 1900. For many elderly Americans, however, this welcome increase in life expectancy is marred by a deterioration in cognitive abilities, particularly memory (fig. 5.5).

  Some weakening of memory, beginning around age forty, is normal. Until recently, however, it was not clear whether this age-related memory loss, also called benign senescent forgetfulness, is simply the early phase of Alzheimer’s disease or a distinct entity in its own right. The answer to that question is not only a matter of
considerable scientific interest, it is also a matter of enormous financial and emotional consequence to our society and its aging population.

  Because implicit and explicit memory are controlled by different systems in the brain, aging affects them differently. Implicit memory is often well preserved in old age, even in the early stages of Alzheimer’s disease. That’s because the disease does not affect the amygdala, the cerebellum, or other areas important for implicit memory until quite late in its course. It also explains why people who are unable to recall the names of loved ones can still ride a bicycle, read a sentence, and play the piano. In contrast, explicit memory—the memory of facts and events—degrades early in people with Alzheimer’s disease.

  Figure 5.5. Prevalence of memory loss in the aging population

  To find out whether Alzheimer’s and age-related memory loss are biologically different, two groups of scientists at Columbia University, one led by Scott Small and one by me, compared three variables: namely, age of onset and progression of each disorder, regions of the brain involved, and molecular defects in each of the identified regions.

  To compare age of onset and progression, my colleagues and I turned to mice.5 Mice do not develop Alzheimer’s disease, but we found that they do show an age-related memory loss that is centered in the hippocampus. This memory loss begins in midlife, as age-related memory loss appears to do in people. So in mice, at least, we could see that age-related memory loss exists as a separate entity, independent of Alzheimer’s disease.

  To find out what areas of the brain are involved in age-related memory loss and what areas are involved in Alzheimer’s disease, Small and his group used brain imaging to study human volunteers ranging in age from thirty-eight to ninety. They found, as others had earlier, that Alzheimer’s disease begins in the entorhinal cortex, but they also found that age-related memory loss involves the dentate gyrus, a structure within the hippocampus.6

  Small’s group and mine then collaborated to determine whether the dentate gyrus contains any molecular defects that the entorhinal cortex does not contain.7 To do this, we examined at autopsy the brains of people between the ages of forty and ninety who did not have Alzheimer’s disease. Using Affymetrix GeneChips, a technology that enabled us to analyze changes in the expression of as many as twenty-three thousand genes, we found nineteen gene transcripts that varied with the age of the volunteer. (Transcripts are the single-strand RNA molecules produced in the initial stage of gene expression.) The first and most dramatic change was in a gene called RbAp48. This gene became increasingly less active in the dentate gyrus of older volunteers, resulting in less RNA transcription and less synthesis of the RbAp48 protein. Moreover, the change occurred only in the dentate gyrus, not in any other area of the hippocampus or in the entorhinal cortex.

  RbAp48 turned out to be an interesting protein. It is part of the CREB complex, a group of proteins that are critical for turning on gene expression required for the conversion of short-term memory to long-term memory.

  Finally, Small and I returned to mice to see whether expression of the RbAp48 protein also drops off in the dentate gyrus of mice as they age. We found that it does—and once again, the decrease occurs only in the dentate gyrus. In addition, we found that knocking out the RbAp48 gene caused young mice to perform as poorly on spatial tasks as old mice. Conversely, ramping up the expression of the RbAp48 gene in old mice eliminated age-related memory loss, causing them to perform like young mice.

  At this point a surprise emerged. Gerard Karsenty, a geneticist at Columbia University, had picked up on the discovery that bone is an endocrine organ and that it releases a hormone called osteocalcin. Karsenty found that osteocalcin acts on many organs of the body and also gets into the brain, where it promotes spatial memory and learning by influencing the production of serotonin, dopamine, GABA, and other neurotransmitters.8

  Karsenty and I joined forces to examine whether osteocalcin also affects age-related memory loss.9 My colleague Stylianos Kosmidis injected osteocalcin into the dentate gyrus of mice and found that it leads to increased PKA, CREB, and RbAp48—the proteins needed for memory formation. Mice that were not given the injections had fewer CREB and RbAp48 proteins. Interestingly, when we gave old mice osteocalcin, their performance on memory tasks such as novel object recognition—which had declined with age—improved. In fact, their memory matched that of young mice. Moreover, osteocalcin even improved the learning capabilities of young mice.10

  These findings—that osteocalcin declines with age and that it can reverse age-related memory loss in mice—may provide another explanation for the beneficial effects of exercise on the aging human brain. We know that aging is associated with a decrease in bone mass and that the resulting decrease of osteocalcin contributes to age-related memory loss in mice, and possibly in us as well. We also know that vigorous exercise builds bone mass. Thus it is likely that osteocalcin released by the bones ameliorates age-related memory loss in people as well as mice.

  Clearly, as these studies illustrate, age-related memory loss is a disorder that is distinct from Alzheimer’s disease—it acts on different processes in a different region of the brain. Moreover, the Roman ideal of a sound mind in a sound body now appears to have a scientific basis.

  This is good news for people with a normally aging brain. They can maintain crucial mental functions into old age, provided they eat healthfully, exercise, and interact with others. Just as we have learned to extend the life of the body, we must also extend the life of the mind. Fortunately, as we have seen, several avenues of research encourage us to think that diseases affecting memory may one day be preventable.

  It is also important to note that many of the aspects of cognitive function that don’t require memory mature quite well. Wisdom and perspective certainly increase with age. Anxiety tends to decrease. The challenge for all of us is to maximize the benefits of aging while doing our best to minimize the downside.

  ALZHEIMER’S DISEASE

  Aging seems to target particular areas in the brain, and, as we have seen, the hippocampus is one of the most vulnerable. Sometimes it is damaged by lack of blood flow or cell death, but it is often damaged by Alzheimer’s disease.

  Alzheimer’s disease is characterized by deficits in recent memory. It results from the loss of synapses, the point of contact where neurons communicate. The brain can regrow synapses in the early stages of the disease, but in the later stages, neurons actually die. Our brain cannot regrow neurons, so this cell death results in permanent damage. Treatment for Alzheimer’s is likely to be most effective early on, before extensive cell death, so neurologists are trying to develop functional brain imaging and other methods of identifying the disease as early as possible.

  Scientists have begun to unravel the cascade of events underlying the symptoms of Alzheimer’s disease. They have also learned a great deal about the molecular biology of the disease. Every detail added to that store of knowledge gives us another potential target for a drug, another possible way of halting the progress of this devastating disorder.

  The discovery of Alzheimer’s disease dates back to 1906, when Alois Alzheimer, a German psychiatrist and colleague of Emil Kraepelin, described the case of a fifty-one-year-old woman, Auguste D., who had become suddenly and irrationally jealous of her husband. Soon thereafter, she developed memory deficits and a progressive loss of cognitive abilities. In time, her memory became so impaired that she could no longer orient herself, even in her own home. She hid objects. She started to believe that people intended to murder her. She was admitted to a psychiatric clinic and died less than five years after the onset of her symptoms.

  Alzheimer performed an autopsy on Auguste D. and found three specific alterations in the cerebral cortex that have since proven to be characteristic of the disease. First, her brain was shrunken and atrophied. Second, the outside of the nerve cells contained deposits of a dense material that formed what we now call amyloid plaques. Third, inside the neurons was an accumulation of t
angled protein fibers that we now call neurofibrillary tangles. Because of the importance of this discovery, Kraepelin named the disorder after Alois Alzheimer.

  Some of what a pathologist sees under a microscope at autopsy we can now see with brain imaging. Figure 5.6 shows the amyloid plaques and neurofibrillary tangles that are hallmarks of Alzheimer’s disease. At first, scientists thought these abnormal protein aggregates were just by-products of the disease, but we now know that they are instrumental in causing it. One of the fascinating things about them is that they form ten to fifteen years before a person’s memory or thinking has begun to change. If these structures could be detected when they first appear, it might be possible to prevent damage to the brain and to stop Alzheimer’s disease in its tracks.

  Plaques initially form in specific, restricted areas of the brain. One such site is the prefrontal cortex. As we learned earlier, this part of the brain is involved in attention, self-control, and problem solving. Tangles start in the hippocampus. Amyloid plaques and tangles in these two areas account for the cognitive decline and memory loss in people with Alzheimer’s. At first, the brain is able to compensate well enough that even a family member can’t tell the difference between someone who has this initial damage and someone who does not. Over time, however, as more and more connections are damaged and neurons begin to die, regions like the hippocampus disintegrate and the brain begins to lose crucial functions such as memory storage. Symptoms related to memory loss then become noticeable.

  Figure 5.6. Enhanced photograph of an amyloid plaque and a neurofibrillary tangle in the brain