7. STEPS TOWARD AN ANATOMY OF MEMORY

LARRY R. SQUIRE, PHD

Larry Squire presents an overview of the current scientific understanding of memory and how it relates to the brain’s organization. Functions of memory have been found to be distributed throughout the brain, insofar as memories reactivate the original pathways of the experience remembered. But they are also localized in specific brain structures that control memory, independent of other aspects of awareness. Distinctions are drawn between different types of memory. There is short-term and long-term memory; declarative memory in which we consciously hold a representation of some aspect of experience; and nondeclarative, unconscious memory that manifests, for example, as improved performance in a learned skill.

LARRY SQUIRE: Problems of brain function are so diverse and numerous that there are many thousands of investigators in this country studying brain function. There are many groups studying even a problem as specific as memory from many different points of view. This work is panoramic and proceeding at many different levels of analysis: How do single neurons change? What is the anatomy, and the chemistry, of such changes? How do neurons combine themselves into functional networks? How do these networks change? How is memory organized at levels of behavior and cognition? How is memory stored anywhere in the brain? How many different kinds of memory are there?

One might say that in the neurosciences there are two great problems: there is the problem of the initial organization of connections among nerve cells in the brain, and there is the problem of how these original connections can be altered. Initial connections are concerned in part with the things that Dr. Damasio was talking about. We inherit, when we are born, abilities for many different kinds of functions, for example, to perceive visually, auditorily, somesthetically, and by means of olfaction and taste. Infants do well at localizing sounds and responding to melodies. They can distinguish their own mother’s voice from others. They are able to perform considerable coordinated locomotion. And each normal human baby inherits the remarkable capacity for developing language.

We inherit those abilities that have been shaped by selection during millions of years of evolution, a kind of cumulative record or memory of abilities that have favored survival and reproduction among our ancestors.

In addition to initial neuronal connections that we inherit, most animals also have the ability to change perceptions and behavior as a result of experiences in their own lifetimes. Experience can modify the nervous system, and these modifications enable us subsequently to behave differently and think differently, as a consequence of individual experiences. We call that the ability to learn and to remember.

Although we do not understand the physical basis of memory in much detail, most neuroscientists believe that memory is recorded in the brain as changes in the strengths of connections among neurons. Thus, there are networks of neurons that operate initially in a certain way. As a result of experience such networks may exhibit changes in the strength of some connections. Some connections may get stronger, others may diminish in strength. As a consequence of such changes, the network then functions differently.

MEMORY REENACTS PERCEPTION

One can usefully think of memory as an extension of perception. Ordinarily, perception can be transformed into memory. We use the word encoding to refer to memory deposition, or memory registration, at the time of learning. We know that encoding depends on many different factors such as the amount of attention that is being paid to an event, how important the event is to that individual, and the extent to which that individual can categorize or otherwise organize the event in relation to preexisting knowledge.

The results of the encoding can be held in storage so that later on, even many years or decades later, it is possible to recall that event. In part, we think recall occurs through reactivating some of the same neuronal patterns that were important when the information was being registered in the first place. So one of the fundamentals that we hold about memory is that it is directly linked to perception. Not only do we have in our brains specialized mechanisms for reading, perceiving music, and discriminating speech, but we also have the ability to retain in memory the consequences of each of these different kinds of perceptual neuronal processing. We believe that memory is stored in a spread-out, neuronally distributed way in the brain and that memories are stored physically in the same areas that were initially involved in processing and analyzing the event that is to be remembered.

ISOLATING MEMORY: THE EVIDENCE OF DAMAGE

One of the surprising discoveries of this century is that despite the fact that memory is closely linked to perception and attention and other intellectual faculties, nevertheless, damage to one or two very specific areas of the brain will cause isolated memory problems. In these cases, memory can be impaired without there being any impairment of self-awareness or intellectual functions. A very tiny part of the brain is involved, whereby damage can cause such specific memory problems.

Dr. Squire presented a magnetic resonance imaging (MRI) brain scan showing details of the hippocampus.

Here we have an image of the brain of a sixty-five-year-old gentleman who developed a significant memory impairment. In this area of the hippocampus is a very small abnormality—the only abnormality that we can detect using MRI images.

He has a severe memory impairment and is basically unable to learn from new day-to-day experiences. He essentially forgets the moments of life experience almost as fast as they occur. His memory for something that happened five minutes ago might be about as good as your or my memory for some not very important event that happened a couple of weeks ago. He also has some loss of memories for things that occurred in the past, especially the recent past. His intelligence is normal; he is fully aware and has an intact sense of himself and full insight relating to his impairment. He has an intact memory of his childhood. He has a normal personality and normal skills for getting along in the world.

Until recently, scientists thought about memory as being one thing. We were inclined to presume that this individual had a problem that affected his memory in a global way. Indeed, there is a sense in which his deficit is very pervasive because it affects his ability to learn words, to retain conversations, to remember faces, verbal and nonverbal things, spatial layouts, and so forth. His life is terribly disabled, as you can appreciate, because of this memory deficit. Nevertheless, this individual, and other persons like him, have entirely preserved abilities for some other kinds of learning and memory.

DALAI LAMA: Is that person’s long-term memory of events before the disability struck impaired?

LARRY SQUIRE: That is also affected, especially for recent years prior to onset of the disability.

DALAI LAMA: Might a person recollect everything clearly up to a stroke or other impairment?

LARRY SQUIRE: That will often happen, but in cases of specific impairment of memory, it is quite common to suffer losses of memory for events that occurred prior to the onset of the disability. We call that retrograde memory loss.

DALAI LAMA: Isn’t it possible for some patients to remember things from the distant past without being able to recall recent events?

LARRY SQUIRE: Roughly speaking, that’s what we are talking about here. The main deficit is in the ability to store new information. You are particularly able to retrieve older memories of your distant past.

DALAI LAMA: Something that happened five years ago, before the memory deficit began, might be lost, but prior to that time everything would be perfectly recorded and still recallable?

LARRY SQUIRE: That’s exactly right. How far back in time the impairment goes and how dense the memory loss is will depend upon the severity of the brain defect.

DALAI LAMA: But does it happen, on occasion, that the memory up to the time of disease onset remains completely unimpaired, leaving memory deficit only for events after the onset of the disease?

LARRY SQUIRE: It would be unusual to have previous memory complete up to the second. People who have auto accidents may have memories that are accurate up to one or two minutes before the accident, and may be very bad afterward. This individual, however, has difficulty remembering events that happened as long ago as twenty-five years prior to the time his brain became affected.

ANTONIO DAMASIO: Severity really reflects having different kinds of anatomical parts damaged in terms of the retrograde amnesia. In virtually all patients, there are some kinds of memory loss for some time before the accident took place. And in some, if there is a certain type of damage, the loss can be tremendous and can affect retrograde memory for decades.

We have one patient whom we have studied for many years, an individual whom Larry Squire has studied also, who has a retrograde memory loss for virtually fifty of the sixty years of his life. There is very, very little memory left at the level of unique events, although he knows something about his life generally.

ROBERT LIVINGSTON: I would like to know, in this case, whether the hippocampus was in your opinion quite intact prior to the stroke. I ask because otherwise a person might have been an alcoholic or have degeneration relating to atherosclerosis and be deteriorating as far as memory is concerned well prior to the stroke, but perhaps family and physicians didn’t pay much attention to that until a stroke wiped out new memory storage capacity.

LARRY SQUIRE: We are confining ourselves to cases where impairment came on in a single day. In those patients, if the damage is severe enough, they may lose memory access for perhaps ten or fifteen years prior to the disorder. On the other hand, if the damage is less severe, there may be memory impairment for only a year prior to the disorder. All of these patients that I am talking about here have good, intact memories of their early life, say the first fifteen or twenty years.

DALAI LAMA: Does this show that there are physically different mechanisms for storing and recalling these different periods of memory?

LARRY SQUIRE: Yes. The implication is that the hippocampus is important for storing new memories and for retrieving recent memories. But as memories grow older, something changes in the brain, so that they don’t require this structure for retrieval any more.

DALAI LAMA: Is it possible that the storage facilities in the brain for the memories are unimpaired, but that the retrieval system is interfered with? In such a case, might you have the memories cached, so to speak, but not be able to get at them?

LARRY SQUIRE: That is an important question, and often asked. We think that this represents an impairment in storage. This is in part because we have patients who have a memory impairment for a short time, only for one day, for example. And when such patients recover function in these brain sites, they never recover memory for that lost day. It is as though those events were never recorded.

DIFFERENT TYPES OF MEMORY

DALAI LAMA: In Buddhist psychology, we speak of memory in terms of retention and storage. Could some cases involve the capacity for retention, but not for the storing of memory?

PATRICIA CHURCHLAND: You don’t mean retrieval? Or short-term as compared with long-term memory?

ALAN WALLACE: No. Retention is the initial imprinting process, whereas the subsequent storage is deeper and for a longer term. However, one Tibetan term is used for both processes, imprinting and storage.

LARRY SQUIRE: Western neuroscience makes a distinction between what we call short-term memory and long-term memory. These amnesic patients with memory impairment have intact short-term memory. That is, they can repeat back a short sentence and they can understand the meaning of such a sentence. Thus, they have a capable memory for a short period of time.

PATRICIA CHURCHLAND: How long is short-term memory?

LARRY SQUIRE: It can last from many seconds to a few minutes. If you say, “Four, three, six…four, three, six…four, three, six…” at intervals over a period of fifteen minutes, an amnesic patient can respond by repeating it correctly whenever asked. If you leave and come back in a few minutes to ask what that number was, he will probably be able to respond correctly, “Four, three, six.” But then, if you distract him to attend to a different task and ask him again what was that number, he will have no recollection of it.

So the brain makes a distinction between a short-term and long-term memory, and more recently it has become clear that another and perhaps deeper distinction is very important. This is the distinction we would make between declarative, or conscious, memory and nondeclarative, or unconscious, memory.

Amnesic patients have a problem only in acquiring declarative memories, which form the basis later for conscious recollection of past events. Many memories are like that: our memory for conversation, our ability to recall a specific face. But there are other kinds of nondeclarative memories of which we remain generally unconscious.

Dr. Squire demonstrated two tests for the acquisition of nondeclarative memories. The first consisted of a list of uncommon English words printed as though seen in a mirror.

One can learn through practice the skill for reading these words very quickly. Amnesic patients have a completely normal ability to do that. In general, skills are abilities that depend on memories resulting from experience. But skill acquisition does not involve experiences about which we have very much conscious knowledge. We don’t have much conscious knowledge about reading, or tennis service, for example. One doesn’t know consciously what one does exactly even though one has these skills; one simply does them. And the ability to improve at skills such as reading this kind of text can occur in the absence of the hippocampus, and despite otherwise severe memory impairment.

The second example consisted of a list of words, followed by another list of the same words truncated to the first three letters of each.

Scientists call the second example “priming,” as in priming a pump, a useful metaphor for the process. You present a list of words to the subject, and then you display three-letter beginnings of those words. You say to the subject: “This is not a memory test. This is a puzzle. What I would like you to do is to complete each stem or fragment to form the first word that pops into your mind.” For normal subjects, there is a very high probability that the words they think of will be the words that they previously saw. And a memory-impaired patient will have the exact same ability and tendency to say those words.

The normal subject will say in discussion afterward, “Oh, that’s one of those words you just showed me.” He will have a tendency to complete the word and also to remember that it was a word from the list. The memory-impaired patient will produce the correct word, but will not have any conscious recollection that the word had been presented before.

We now think of memory as being composed of many different components. Clearly, we have conscious memory. This kind of memory depends on the integrity of the hippocampus. Conscious memory is concerned with cognition. It is concerned with making representations concerning the external world, about what occurred, and the relationships among events in the world.

Unconscious, nondeclarative memories are not concerned with representations of the external world. They are concerned with behavior, with priming, with improving our ability to correctly identify objects. In these instances, performance improves as a result of experience. But as we have shown, to maintain these kinds of skills, there is no necessity to retain recollections of the past. We simply adapt to the performance requirements for the skill. We simply change our behavior in the course of acquiring skills.

We consider that nondeclarative memories are relatively primitive with respect to evolution. For example, snails and other simple animals have some of these abilities to change behaviorally, to develop skills, as it were. If you touch a turtle on the head many times, it will keep its head inside the shell. The same is true for many snails. The animal simply changes its behavior a result of experience. But we think this is very different from consciously recollecting. There are unconscious abilities to change behavior, such as the abilities to improve the perception of something seen recently or to develop a new movement, which are distinct from the abilities to consciously recollect different occasions of encounters and experiences. And there is one particular brain system, the hippocampus, which is essential for conscious kinds of memory, but not for developing sensory and motor skills.

One of the things that makes this an exciting time to study memory is that we can also study many of these phenomena in animals, including conscious memory in the monkey.

In one such study, the animal sees an object which he must push aside to disclose a raisin reward. Then one minute later, the animal is presented two objects. The monkey will now find a raisin only if he moves aside the new, unfamiliar object. By its performance, the animal informs us that it has recognized the original object as familiar so as to be able to select the other object. Brain lesions in the hippocampus, exactly similar to the patients’ lesions, interfere with performance of this task.

One of the important goals of the neurosciences is to understand the significance of brain connections, not only of what is connected to what, but what purposes are served by given connections. Even for something as simple as perceiving a drinking glass on a table, we know now that the brain has two distinctive areas that are each important for that ability. One area, in the inferior aspect of the temporal lobe, is important for identifying what the objects are; the other area, in the parietal lobe, is important for identifying where they are in relation to other objects in space. This dissects perceptual processes in very fundamental ways.

Now, if perception is to be transformed into memory, there need to be connections down to the hippocampus, through stations¹ in the medial aspect of the temporal lobe. All of these must be functional. So, later on, in order to reactivate the representation of the drinking glass on the table, there must have been activity through these circuits at the time of the original perception. These same circuits are thereafter critical for the ability to recollect conscious memories.