Learning to find your way: the biochemical pathways underlying spatial memory is the brain are giving up their secrets
Eric R. Kandel
For all living creatures, knowledge of the surrounding environment and their position within it is key to behavior and critical to survival. At the simplest level spatial “knowledge” may encompass no more than the ability to orient toward or away from a stimulus. In complex organisms, though, the representation of space is a cognitive process, in which inputs from several senses–sight, hearing, the sensations of motion and posture provided by the inner ear and muscle tension–are bound together. Such binding is a function of the brain. How is it accomplished?
The brain represents information about space in many of its areas and in many different ways. For some purposes the brain represents space with egocentric coordinates, that is, from the point of view of the sensing organism. For example, the brain encodes where a light is relative to the fovea of the retina, or where an odor or touch comes from with respect to the body. For other kinds of behavior the brain encodes the organism’s position with respect to the outside world, and the relations of external objects with respect to one another. Such position coordinates, which are centered on the world, are known as allocentric coordinates.
The eighteenth-century German philosopher Immanuel Kant, one of the forefathers of cognitive psychology, argued that the ability to represent space allocentrically is built into the mind. People, in Kant’s view, were born with principles that ordered experience in space and time, and were prepared to interweave sensations automatically within this framework in specific ways, whether the sensations were elicited by objects, melodies, or tactile experiences.
In the early 1970s, John O’Keefe, a cognitive neuroscientist at University College London, applied this Kantian logic about space to explicit memory-memory that is recalled by deliberate, conscious effort. Explicit memory, which concerns such things as facts and events, people and objects, can be contrasted with implicit memory, such as motor or perceptual skills and conditioned responses, which are accessed and performed unconsciously. O’Keefe argued that many forms of explicit memory are associated with spatial coordinates–that is, we typically remember people and events in a spatial context.
This idea is not new. In 55 B.C., Cicero, the great Roman statesman and orator, described a Greek technique for remembering words. The idea was to picture the rooms of a house in sequence, associate words with each room, and then mentally walk through the rooms in the right order. To this day some actors and others who must memorize and recall information rely on the technique.
O’Keefe was the first to realize that rats have a multisensory representation of extrapersonal space localized in a part of the brain known to be involved in explicit memory storage, called the hippocampus. In 1971 O’Keefe probed how individual neurons, or nerve cells, were activated in the hippocampus of laboratory rats, as the animals walked around in an enclosure. Some neurons, he discovered, are activated when that animal moves to one position, whereas others fire when the animal moves elsewhere. He called these neurons “place cells.” On the basis of those findings, it is thought that as an animal explores its surroundings, the brain breaks down the territory into many small, overlapping areas, similar to a mosaic, thereby forming an internal map. The map develops within minutes of the rat’s entrance into a new environment. Under optimal circumstances, it lasts weeks or even months.
I began to think about the spatial map in 1992, wondering how it is formed, how it is maintained, and how attention might direct its formation and maintenance. I was struck by the finding of O’Keefe and others that the spatial map of even a simple locale does not form instantaneously. Instead, it forms over a period of between ten and fifteen minutes after the rat enters the new environment. The delay suggested that forming a spatial map is a learning process, in which practice makes perfect. Thus even though the general capability for forming spatial maps that Kant envisioned may be built into the brain, each particular map of a specific environment is not.
When my colleagues and I began studying spatial maps, nothing was known about the molecular details of their formation. But we did have a research advantage. We had spent many years teasing out some of the biochemical processes whereby neurons alter their responses to stimuli and the connections between neurons are modified as a result of an animal’s experience. In other words, we already understood, in principle, what can make learning possible.
We had gained our understanding through the fortuitous choice of a research subject, the marine snail Aplysia, an organism with a relatively simple neurological organization. Its brain has only about 20,000 neurons, compared with 100 billion or so in the human brain. Moreover, Aplysia neurons are extremely large, some even visible to the naked eye. In that simple animal, we delineated a simple reflex behavior, in which fewer than one hundred nerve cells take part. The reflex could be modified by learning and retained in memory for several weeks. In that way, we were able to pinpoint the cellular and molecular mechanisms that contribute to learning and memory.
One of the rewards of any avenue of scientific investigation is that, as specialized and narrow as it may seem, it can lead to broadly useful insights. Laboratory experiments or field observations that may at first seem to have no practical application can prove helpful or even essential in solving pressing problems. Although it is too soon to say how or when, the accumulating knowledge of the biochemical mechanisms underlying learning and memory may one day help prevent the “normal” memory loss of aging and perhaps even cure Alzheimer’s disease and other dreaded neurological conditions associated with learning disabilities.
Our studies of neurons in Aplysia on a biochemical level made two things clear. First, neurons can adjust their responses to stimuli in the short term, either by becoming more sensitized to an important stimulus (such as one that is harmful) or by becoming habituated to–and therefore ignoring-one that is inconsequential. To make short-term adjustments, the neuron regulates the strength of its connection with other neurons by chemically altering preexisting proteins and increasing or decreasing the efficiency of preexisting synaptic connections.
Second, neurons can adjust their responses over the long term by increasing or decreasing the number of contact points with other neurons. To construct new points of contact, structural proteins are needed; to assemble the proteins, genes that serve as the blueprints for making the proteins have to be turned on, or mobilized, in the nucleus of the neuron.
In the late 1980s, a number of investigators made the first attempts to understand how long-term potentiation, or enhancement, of connections between neurons played a role in spatial memory. At Columbia University, three post-doctoral fellows–Ted Abel, Seth G.N. Grant, and Mark R. Mayford–and I created various fines of genetically modified mice that lacked one or another key protein thought to be involved in long-term potentiation. We then tested the animals’ performance on several well-understood spatial tasks.
For example, we placed a mouse in the center of a large, white, well-lighted circular platform, with forty mouse-size holes drilled into the rim [see illustration at left]. Mice hate being in light, open spaces, but the platform is too high off the floor for a mouse to escape by leaping off its edge or through a hole. The only escape is through one hole that leads through a tunnel to an enclosed chamber. The mice do get a clue about the way out. The platform is mounted in a small room, each of whose walls is decorated with a different, distinctive marking.
When a normal mouse is first exposed to this experimental condition, it races about in a panic, visiting the holes at random in its search for the way out. After repeated trials it adopts a serial strategy-slightly more efficient than random searching, but not by much–starting at one hole and methodically checking out each one in order until it finds the right one. Neither strategy requires the mouse to have an internal map of the environment stored in its brain. Finally, the mouse learns to recognize which marked wall is aligned with the target hole, and then it makes a beeline for that hole. Most mice soon learn to use that spatial strategy.
In one of our special breeds of mice, we had inhibited a gene that encodes a protein–protein kinase A. The protein is important for long-term potentiation because it mobilizes genes that, in turn, code for the structure proteins required for building new synaptic connections. Inhibiting the kinase A gene compromised the strengthening of synapses in a pathway in the hippocampus. The mutant mice never learned to use the markings on the wall as a guide to the escape hole; they kept searching for it with the simpler but inefficient strategies, in trial after trial.
In later work–in collaboration with the neurobiologist Robert U. Muller of the State University of New York Downstate Medical Center in Brooklyn and his student Alexander Rotenberg–Ted Abel, my student Naveen T. Agnihotri, and I discovered that both protein kinase A and the synthesis of structural proteins are needed for a spatial map to become “fixed” over the long term–so that, for instance, a mouse recalls the same map every time it re-enters a particular space.
In his initial formulation, O’Keefe regarded a cognitive map as merely a kind of a navigational tool, comparable to a compass. Such an internal representation of space would enable an animal to move efficiently in the environment by recognizing directions and landmarks, but it would not endow the animal with any long-term memory of the space. In contrast, our experiments showed that the hippocampus may also serve as a memory store for past responses in encountering those landmarks–thus enabling a normal mouse on the platform to appreciate the value of the spatial landmarks in locating the reward–the escape hole. In this sense it endows an animal with an explicit memory of its space.
My colleagues and I were intrigued by the fact that, despite certain similarities, people’s explicit memory of space differs in substantial ways from their implicit memory. For example, hotel guests may remember to proceed to the nearest stairwell if they hear the fire alarm (implicit memory), without remembering that they must pass fifteen rooms before reaching the stairwell (an explicit memory they would possess only if they had consciously counted the doors). Thus, explicit memory requires selective attention for encoding and recall. To examine the relation between neural activity and explicit memory, we decided to study attention.
Selective attention is widely recognized as a powerful factor in perception, action, and memory–in the unity of conscious experience. At any given moment, animals are inundated with a vast number of sensory stimuli–far more than the brain can process. Attention acts as a filter, selecting some objects for further processing. As the American psychologist William James noted in his seminal book, The Principles of Psychology, in 1890:
Millions of items … are present to my senses which never properly enter into my experience. Why? Because they have no interest for me. My experience is what I agree to attend to…. Every one knows what attention is. It is the taking possession by the mind, in clear and vivid form, of one out of what seem several simultaneously possible objects or trains of thought.
Is selective attention required to form and retain a spatial map? Clifford G. Kentros, a postdoctoral fellow, and I exposed mice to experimental conditions that required increasing degrees of spatial attention [see illustration on preceding page]. We implanted a probe in the brain of a mouse that could measure the individual firing of as many as four place cells as we tracked the animal’s position in a test enclosure. The enclosure was circular, with enough visual cues on its walls for the animal to orient itself and, perhaps, to form a spatial map.
One condition was designed simply to establish a baseline level of attention, which we called ambient attention. A mouse was given some time to run around in the enclosure, without any distracting stimuli, and the mouse’s position and the firing of the cells were recorded simultaneously. In another condition, we engineered things so that as the mouse walked around in its enclosure, bright lights and loud sounds, which the mouse hates, came on periodically and at random. The only way the mouse could turn them off was to run to a small “goal region” on the floor of the enclosure and sit there for a moment. The region itself was unmarked, but the mouse could find it by paying attention to the available visual cues. Mice learn this task very well.
Kentros and I determined that even with ambient attention, the mouse forms a spatial map that remains stable for an hour or two. Such a map, however, becomes unstable after three to six hours. When a mouse is forced to pay a lot of attention to a new environment, by having to use visual landmarks to learn a spatial task, the spatial map remains stable for days.
So what is the attentional mechanism in the brain? How does it contribute to the strong encoding of information about space and the ready recall of that information after long intervening periods? Michael E. Goldberg and Robert H. Wurtz, both at the National Eye Institute in Bethesda, Maryland, had already shown that in the visual system, attention enhances the response of neurons to stimuli. One neural pathway that seemed to play a key role in paying attention to a stimulus was mediated by the neurotransmitter dopamine–in other words, the signals transmitted across synapses from one neuron to the next along the pathway were modulated by dopamine. The neurons that make dopamine are clustered in the midbrain; their axons–the long projections from the main body of the nerve cell–send signals to a number of sites in the brain, including the hippocampus.
That suggested an obvious experiment. What would happen to the spatial map of an animal that was paying attention to its surroundings if dopamine was blocked from reaching its hippocampus? My co-workers and I proved just what we had been led to expect: Without dopamine, the spatial map in the mice would not stabilize; the place cells in the mice would not reliably fire as they did when the spatial memory was fixed. Conversely, when we activated dopamine receptors in the hippocampus, the spatial map of an animal became more stable even when the animal was not paying attention.
In The Principles of Psychology, James pointed out that there are at least two forms of attention: involuntary and voluntary. Involuntary attention is supported by automatic neural processes, and it is particularly evident in implicit memory. In classical conditioning, for instance, animals learn to associate two stimuli if and only if the conditioned stimulus is salient or surprising. In the textbook case, a bell rings when food is presented to a dog. After several such training cycles, the sound of the bell alone–the conditioned stimulus–is enough to get the dog to salivate. Involuntary attention is activated by a property of the external world–of the stimulus–and it is captured, according to James, by “strange things, moving things, wild animals, bright things, pretty things, metallic things, words, blows, blood.”
By contrast, voluntary attention, such as paying attention to where staff members sit in a new office environment, is a feature of explicit memory. It arises from the need to process stimuli that are not automatically salient to the nervous system. James argued that voluntary attention is obviously a conscious process in people; therefore, it is likely to be initiated in the cerebral cortex.
The molecular studies my colleagues and I have conducted in Aplysia and mice support James’s contention that both forms of attention exist. In both voluntary and involuntary attention, the short-term memory of a salient stimulus is converted to long-term memory through the activation of genes. In both cases, neurological pathways and chemical transmitters act as modulators, carrying a signal that marks the stimulus for special attention. In response to that signal, genes are turned on and proteins are produced and sent to all the synapses to strengthen the connections between neurons.
For example, Aplysia normally withdraws its delicate gill into its mantle cavity if its siphon is touched. The response weakens, through habituation, if the siphon is touched repeatedly but weakly. But if the weak touch is then paired, through conditioning, with a shock to the animal’s tail, the weak touch alone will stimulate a brisk response. The shock causes the neurotransmitter serotonin to be released along the neural pathway that carries out the effective motor response for withdrawing the gill. The serotonin triggers protein kinase A, which turns on genes that send structural proteins to the synapses most relevant for quickly withdrawing the gill. Similarly, when a mouse learns the spatial task that switches off obnoxious lights and sounds, we presume that dopamine is released along the neural pathways that represent the mouse’s spatial map. The dopamine then triggers the production of protein kinase A, which “fixes” the memory of the map.
But how do stimuli lead to the release of the neurotransmitters along the “right” neural pathways, marking the stimuli for special attention? In implicit memory storage, exemplified by Aplysia’s increased sensitization through conditioning, the attentional signal is called up involuntarily–reflexively–from the bottom up. The sensory neurons, activated by a shock, act directly on the cells that release serotonin. In the mouse’s spatial memory, however, the attentional signal is called up in a fundamentally different way. Dopamine appears to be recruited from the top down. The cerebral cortex activates the ceils in the midbrain that release dopamine, and dopamine modulates activity in the hippocampus. Nevertheless, in learning to attend, both from the top down and from the bottom up, the underlying molecular mechanisms are similar.
In 2004 Kausik Si, a postdoctoral fellow in my laboratory, discovered that Aplysia carries a novel form of a protein known as CPEB. The novel protein, ApCPEB, is present at all the synapses of the sensory neurons of the gill-withdrawal reflex, where it is activated by serotonin and is required for the growth of new synaptic terminals [see illustration on preceding two pages]. Si discovered that one end of ApCPEB has all the characteristics of a prion.
Prions are probably the weirdest proteins known to modern biology. They cause several neurodegenerative diseases, such as mad cow disease in cattle and Creutzfeldt-Jakob disease in people. What is unique about prions is that they can fold into two distinct shapes that function in highly different ways, One shape is dominant, the other recessive. The genes that encode prions give rise to the recessive form. But the recessive form can convert into the dominant form either by chance or as a result of being exposed to the dominant form.
For example, the recessive prions in an animal can take on the dominant form if the animal eats food that contains the dominant form.
Most proteins are subject to constant turnover, degraded and destroyed in a few hours. But dominant prions are self-perpetuating, because they can trigger newly minted recessive prions to switch to the dominant form as well, causing a chain reaction. Thus their influence is tenacious.
Soon after Si discovered the prionlike properties of ApCPEB, we postulated that in Aplysia’s sensory neurons, serotonin might control the conversion of ApCPEB from its inactive, nonpropagating, recessive form to its active, propagating, dominant form. In other words, the modulatory transmitter required for converting short-term to long-term memory acts by creating dominant ApCPEB protein.
And that protein apparently maintains newly grown synaptic connections over long periods, perpetuating memory storage.
If confirmed, the discovery would be the first case in which a physiological signal–serotonin–may be critical in converting one prion form to another. It would be also the first example of a self-propagating prion form that serves a useful physiological function. In all other cases previously studied, the dominant form either causes disease and death by killing nerve cells or, more rarely, is inactive.
After that finding in Aplysia, Martin Theis, a postdoctoral fellow in my laboratory, and I began testing the idea that, in much the same way, mouse dopamine controls the conversion of another prion-like protein known as CPEB-3 in the mouse hippocampus. That raises the intriguing possibility–so far only that–that spatial maps may become fixed when an animal’s attention triggers the release of dopamine in the hippocampus. That dopamine might then initiate a self-perpetuating state maintained by the dominant form of CPEB-3.
If that idea proves correct, it would open up a new biochemical approach to the stabilization of long-term memory. Eventually, then, new drugs might one day exploit those effects to treat Alzheimer’s disease and other disorders of memory.
This article was adapted from In Search of Memory: The Emergence of a New Science of Mind, by Eric R. Kandel, which is being published this month by W. W. Norton & Company, Inc. Copyright [c] 2006 by Eric R. Kandel
COPYRIGHT 2006 Natural History Magazine, Inc.
COPYRIGHT 2006 Gale Group