Localization of Memory Traces

views updated


The brain consists of a vast number of individual cells called neurons. Individual neurons form highly complex patterns of interconnections with many other neurons. Each of these connections is called a synapse and a collection of interconnected neurons is called a neural network. It is within these networks of neurons and synapses that memories are formed and stored. The term memory trace, also called the engram, broadly refers to the change(s) in the brain that serves to store a memory. To fully understand the nature of a memory trace, at least three different but interrelated properties must be elucidated. First, the precise region within the brain where the memory dependent changes occur must be localized. This entails identifying the specific neural network (or neural circuit) that subserves the formation, storage, and retrieval of the particular memory and then localizing the site(s) of change(s) within that network that mediates storage of the memory. Second, once the site of memory storage has been identified, the biophysical properties of the changes that occurred within the neural network as a result of memory formation must be identified. For instance, these changes might involve strengthening synaptic connections between different neurons, a process that might entail expression of different, memory related genes. Finally, in addition to identifying the site of memory formation and the nature of the changes that occur, the memory specific pattern of neural activity within the network that occurs during recall of a memory must be delineated. This entry will focus primarily on the first step in understanding the nature of memory traces: localizing traces within the brain.

History of Memory Localization

In the early days of behaviorism, the observable, quantifiable study of behaviors, it was thought that each memory was represented as a change in the brain at one particular place. Karl Lashley began the search for the memory trace, stressing the now obvious point that in order to analyze the nature of memory traces, it is necessary to find them. In his classic 1929 monograph, Brain Mechanisms and Intelligence, he concluded that memories, at least memories for complex mazes in rats, did not have any particular locus in the cerebral cortex (equipotentiality); the more cortex removed, the more the impairment in memory (mass action). Walter Hunter was quick to point out that removing more cerebral cortex removed more sensory information (visual, auditory, kinesthetic), in effect reducing the number of available cues (e.g., animals that are blind do not learn mazes well). This issue has never really been resolved, at least for complex maze learning in the rat, although we now know that the hippocampus is important for such memories.

Following Lashley's failure to localize memory traces, some scientists adopted the view that they were distributed either widely throughout the brain or widely within certain brain structures like the cerebral cortex. But as more was learned about the anatomical and functional organization of the brain, it became clear that the brain does not have a diffusely distributed organization; instead, it has a highly structured organization. Donald Hebb, in his important and influential book The Organization of Behavior (1949), proposed a resolution of this dilemma by assuming that the organization of a memory trace can be complex and involve a number of brain areas but that the trace can involve specific connections in particular areas. This remains a common view. Hebb also proposed a possible mechanism of memory trace formation that has come to be known as the Hebb synapse. In brief, he argued that at neurons where traces are formed, there must be active input from a to-be-learned source (e.g., conditioned stimulus in Pavlovian terms) at the same time the neuron is firing action potentials. The Hebb synapse has come to be viewed more generally as a strengthening or weakening of one input (synapses) to a neuron if this input is active concurrent with activation from another input to the neuron.

By the end of the twentieth century, the focus had shifted away from memory traces in complex tasks to more specific and discrete learning and memory tasks, and research emphasized identifying the entire circuitries essential (necessary and sufficient) for particular forms of memory. Only after this has been accomplished can the memory traces be localized and analyzed. Well-established methods of lesions, electrical recording of neuronal activity, and electrical stimulation of brain tissue, together with anatomical tracing of pathways in the brain, have enabled much progress in the identification of essential memory circuits in the brain, at least for simpler forms of learning, although the experimental difficulties are formidable. Relatively new methods such as functional imaging, probes for genetic expression, or localized infusion of highly specific receptor antagonists or agonists have also become widely used tools in the search for essential memory circuits. Once the complete circuit for a particular form of learning has been identified, the next step of localizing the memory trace(s) within that circuit is orders of magnitude more difficult. Indeed, there are no universally agreed upon methods for doing so. This aspect of the search for memory traces has become the conceptual center of the field.

Localization of Different Types of Memories

There are many different types of learning and memory; for instance, learning to ride a bicycle differs from memorizing a list of facts. A distinction is often made between two general categories of memory: declarative (learning "what") and procedural (learning "how"). Many other terms have been suggested for this dichotomy; extreme examples of the two types of memory are one's memory of one's own recent experiences (declarative) and classical or Pavlovian conditioning, where a specific conditioned response like salivation or eye blink is learned to a particular conditioned stimulus (procedural). Although both types of memory formation involve many regions of the brain, the brain structures and systems essential for the two types of memories are quite different. Indeed, there are several different memory circuits and systems in the mammalian brain. Some of these will be noted here; each is treated in a separate article in this volume.

In humans and other mammals, the hippocampus appears to play a key role in recent experiential memory (declarative). Extensive damage to the hippocampus can markedly impair recent memory in humans and monkeys. Evidence suggests that the impairment is more in the establishing of memories—a process that appears to take weeks in monkeys and may take years in humans—than in their retrieval. In rodents, recent "working" memory and spatial memory are impaired by hippocampal lesions. Very recent or short-term memory in monkeys also involves the prefrontal areas of the cerebral cortex. The thalamus, the largest subdivision of the diencephalon, also plays a role in recent memory in humans. However, long-term permanent memories, representing our knowledge and our life experiences, are not stored in the hippocampus, prefrontal cortex, or thalamus—and thus are not impaired by damage to these structures. The cerebral cortex is often suggested as the storage site for these long-term memories, but definitive evidence is lacking.

In contrast, the clearest evidence for a high degree of localization of a memory trace exists for classical conditioning of discrete behavioral responses—for instance, the conditioned eyeblink response. This type of learning, which occurs in humans and other mammals, involves associating a neutral stimulus, such as a brief tone, with another stimulus, for instance an air puff to the eye, that evokes a specific movement such as an eye blink. After presenting the tone paired with the air puff, subjects are conditioned to blink their eye to the tone alone. Using the methods of stimulation, lesions, and recordings described above, the neural circuit that mediates this form of motor learning was found to critically involve the cerebellum and its associated brain-stem structures. Once this essential circuit was identified, the memory trace was localized within the circuit to a particular region of the interpositus nucleus in the cerebellum. In addition to the interpositus, there appear to be additional storage sites in the cerebellar cortex, and these sites certainly are distributed, in the sense that many thousands of neurons are involved. Localization of the memory trace for eyeblink conditioning is a critical first step toward elucidating the mechanisms of plasticity and the network-level properties mediating expression of the stored memory.

Unlike learned motor behaviors, such as eyeblink conditioning, that involve the cerebellum, fear conditioning, as in conditioned changes in heart rate and blood pressure following pairing of a tone or light with a painful electric shock, critically involves the hypothalamus and amygdala but not the cerebellum. As with eyeblink conditioning, much of the circuitry essential for conditioned fear has been identified. For this particular form of learning and memory, the amygdala is critically involved. In particular, the critical region for memory formation for conditioned fear appears to be localized to a region of the amygdala called the basolateral complex. However, it is uncertain whether the amygdala is the site of long-term storage of this type of memory. The amygdala is also critically involved in hormonal modulation of memory storage.

Because the memory traces for conditioned motor responses or conditioned fear are fairly localized, the memory traces for procedural learning tasks in general may also be relatively localized. In contrast, memory traces for declarative memories may be much more widely distributed. On the other hand, the fact that damage to speech areas in the human cerebral cortex appears to abolish memory for language suggests that this complex learning and memory process, perhaps the most complex yet evolved in nature, may have a considerable degree of localization.

A somewhat different approach has been taken in the study of "simplified" neuronal circuits in certain invertebrate preparations where the number of neurons is small, their sizes are large, and their interconnections are well specified. Here, simplified neural circuits containing only a few identified neurons can be isolated and particular training procedures, usually classical conditioning, can result in the circuits' showing changes in activity that can be long-lasting and can closely resemble similar associative learning in mammals. In these preparations it is possible to localize the learning-induced changes in the activities of the neurons and analyze the underlying mechanisms in some detail. These mechanisms can then provide models of putative mechanisms of memory storage in the mammalian brain.

Mechanisms of Memory Formation

As more and more memory traces are localized in the brain, understanding the biophysical properties of the changes that occur as a result of memory formation becomes possible. Theories abound regarding the nature of the mechanisms of memory trace formation. One early notion held that each memory was stored in a particular protein molecule. This view is no longer tenable, but proteins of course play key roles in the structure and functioning of nerve cells. Another early view was that the brain grew new pathways; thus, in Pavlovian conditioning a new pathway would grow to connect the conditioned stimulus region of the brain to the unconditioned stimulus or response region. This does not occur. Instead, evidence is uniformly consistent with the more modest view that there are changes in the actions of the synapses that are the sites of the interconnections and interactions among the neurons of the brain. Changes in synaptic actions can occur in many ways: changes in amount of neurotransmitter release, changes in receptor molecules, and a variety of other biochemical processes, ranging from calcium entry into neurons to second messenger systems (cyclic AMP, cyclic GNP, protein kinases, and so forth) to changes in gene expression.

Although the biophysical properties of synaptic plasticity have been extensively studied, definitive proof that these mechanisms actually mediate memory storage remains elusive. Perhaps the clearest evidence that synaptic plasticity is a mechanism of memory storage is in conditioned fear. Here, the evidence strongly indicates that a strengthening (potentiation) of synaptic transmission in the amygdala is critically involved in memory formation. However, whether this form of plasticity mediates long-term storage of the memory or whether it is simply an intermediate process in the long-term memory storage is unknown. The strongest evidence we have for a biological substrate of long-term memory storage concerns long-lasting structural changes in the synaptic interconnections among neurons. Enriched environments and even particular learning experiences can result in changes in the numbers and distributions of spine synapses on neuron dendrites, and even in changes in the number of dendritic branches in certain types of neurons. Possibly all these processes and many more are involved in memory formation. A great deal of progress has been achieved in the identification of essential memory circuits in the brain. The search for the memory trace has become one of the most active and exciting fields in neuroscience and psychology.



Hebb, D. O. (1949). The organization of behavior. New York: Wiley.

Lashley, K. S. (1929). Brain mechanisms and intelligence. Chicago: University of Chicago Press.

LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience 23, 155-184.

Maren, S. (2001). Neurobiology of Pavlovian fear conditioning. Annual Review of Neuroscience 24, 897-931.

Squire, L. R. (1986). Mechanisms of memory. Science 232, 1,612-1,619.

Squire, L. R., Knowlton, B., and Musen, G. (1993). The structure and organization of memory. Annual Review of Psychology 44, 453-495.

Thompson, R. F. (1990). The neurobiology of learning and memory. In K. L. Kelner and D. E. Koshland Jr., eds., Molecules to models: Advances in neuroscience, 219-234. Washington, DC: American Association for the Advancement of Science.

Thompson, R. F., and Kim, J. J. (1996). Memory systems in the brain and localization of a memory. Proceedings of the National Academy of Sciences of the United States of America 93, 13,438-13,444.

Thompson, R. F., and Krupa, D. J. (1994). Organization of memory traces in the mammalian brain. Annual Review of Neuroscience 17, 519-549.

Zola-Morgan, S., and Squire, L. R. (1993). Neuroanatomy of memory. Annual Review of Neuroscience 16, 547-563.

Richard F.Thompson

Revised byDavid J.Krupa