H.M. is, arguably, the most famous patient in the history of psychology and neuroscience. He was studied intensively for more than fifty years by hundreds of scientists (including, briefly, the author of this review); he died in 2008, and his brain is still being analyzed. Permanent Present Tense, by Suzanne Corkin, is the story of how these investigations led to a fundamental revolution in our understanding of the human brain and, particularly, of the organization and varieties of memory. Her accessible book places his story in the context of past and present research on memory and describes many of the questions initiated by research on H.M. It is a scientifically exciting and personally moving portrait of a man whose life and brain ended up being devoted to the science of memory.
By the time he was 24, in 1950, H.M. (a k a Henry) had developed severe epilepsy, perhaps from a bicycle accident years earlier, and was referred to the neurosurgeon William Beecher Scoville, who had performed many frontal lobotomies on patients diagnosed as “psychotic.” Scoville had been unsatisfied with the results of frontal lobotomies and was trying a new surgery, bilateral medial temporal lobotomy, in another attempt to treat psychosis. Two of his psychotic patients happened by chance to suffer from epilepsy, and temporal lobotomies had the unexpected effect of reducing their seizures. By 1953, Henry’s epileptic episodes had become more frequent and incapacitating, and Scoville decided to perform a medial temporal lobotomy on him to treat it. He removed, from both sides of H.M.’s brain, a large portion of the medial temporal lobes, including portions of the hippocampus, most of the amygdala, and parts of the adjacent cerebral cortex such as the perirhinal and parahippocampal cortex. After the surgery, H.M.’s seizures diminished significantly, yet the patient could not remember anything that had happened since the procedure, nor could he recognize anybody new or recall a conversation that had transpired a few minutes earlier. He left the hospital with devastating amnesia and could no longer form new memories.
The pioneering neurosurgeon Wilder Penfield, founder of the Montreal Neurological Institute, had been successfully performing unilateral temporal lobe surgery for the treatment of temporal lobe epilepsy. Two of his many patients had developed memory problems after the surgery; it turned out that the unoperated-upon parts of the temporal lobes were abnormal, giving them bilateral temporal lobe damage, as was later the case for H.M. These two patients had their memory examined by Brenda Milner, a McGill University graduate student collaborating with Penfield. When Milner presented their research at the 1954 meeting of the American Neurological Association, Scoville realized the similarity to H.M.’s case and called Penfield, who then sent Milner to examine him. Thus began the modern science of memory.
H.M. turned out to be an ideal subject for the study of memory. Milner found him to be highly intelligent and willing to sit for hours of testing. His perceptual and cognitive abilities, other than memory, were intact. He was able to focus his attention on the task immediately at hand as well as anyone. He was usually very good-tempered and eager to cooperate. He never got bored or restless, perhaps because everything was continually new to him.
Although memory has been a central topic since the founding of experimental psychology in the middle of the nineteenth century, and the effect of brain injury on memory had been studied before Milner’s work on H.M., little was known about the role of different brain areas in memory. The hippocampus was thought to be involved in smell because of its connections with the olfactory system. One widespread view in the early twentieth century, based on the research of Karl Lashley, was that complex memories were distributed throughout the cerebral cortex and therefore could not be localized.
In her initial studies with Henry, published in 1957, Milner established that the hippocampus or adjacent structures were crucial for the long-term storage of information in memory. Milner had discovered that Henry’s digit span (the ability to repeat a gradually increasing series of numbers) and immediate memory were essentially normal, but after about thirty seconds, unless he rehearsed it continually, all record of conscious experience had vanished—nothing new was stored in long-term memory. Furthermore, Henry seemed to have memories intact from several years before the surgery. Milner’s carefully documented experiments revealed several characteristics of the memory mechanism: a specific part of the brain, namely the hippocampus or adjacent tissue, is necessary for the storage of facts and experience; immediate memory is a different process from long-term memory; long-term memory is not permanently stored in the medial temporal region; and severe amnesia can exist in the presence of normal perceptual, motor and language functions.
Milner made another extraordinary discovery when she tested H.M. on a motor skill task called mirror drawing. In this task, the subject looks in a mirror, where he sees both his hand holding a pencil and a piece of paper with two concentric outlines of a five-pointed star. The task is for the subject to draw a line between the outlines on paper while only looking in the mirror. Because the star and the hand can be seen only in the mirror, the image is reversed: for example, to draw a leftward-moving line requires moving the hand to the right. Normal subjects require several trials before they can draw a line between the concentric outlines. H.M. could learn and remember this task as well as normal people. He improved his performance in the first day of trials, and by the third had virtually perfect performance. Yet he had no memory of ever having seen the star or the task before. Milner concluded that “motor skill” learning was an exception to H.M.’s amnesia.
Milner then discovered that H.M. could perform a particular test of perceptual learning that researchers call the Gollin incomplete picture test. In it, subjects are shown drawings of objects in a sequence, from incomplete to complete, and asked to identify the image. The object is too inchoate in the first drawing to be identified, and most people need to see several of the drawings before they can identify it. An hour after seeing the drawings, Henry recognized the objects in far fewer trials than originally, demonstrating perceptual learning and memory. Yet he had no conscious memory of having done the task previously. Milner also showed that H.M. could remember how to do mirror tracing and Gollin figures in spite of his total inability to remember new facts or events.
* * *
Over the next decades, Milner’s paradigm-breaking observations were expanded and fine-tuned by Suzanne Corkin. In 1962, she began to study H.M. as Milner’s graduate student and continued with her research on him for the next forty-six years. He was often tested at her neuropsychological clinic at MIT, from which she has recently retired as professor of behavioral neurosciences. Corkin and her students made many of the important subsequent discoveries about H.M. She was his main liaison with the other scientists studying him, and eventually, as his family died off and he was confined to a nursing home, she became, effectively, his guardian. No one knew him better or longer, personally or scientifically, and no other person could have written Permanent Present Tense. One of the most exciting parts of the book is Corkin’s description of how, after H.M. died, his brain was studied with magnetic resonance imaging (MRI) for many hours and then sent away for detailed anatomical studies. After his death, Corkin revealed to the world, through The New York Times, that H.M. was Henry Gustav Molaison.
Milner’s initial findings raised a host of questions, some of which have been at least partially resolved in the last half-century and others of which remain central to memory research. The advances made since Milner’s original observations have been due to the further study of H.M. as well as of other amnesic patients. After the publication of Scoville and Milner’s 1957 article “Loss of Recent Memory After Bilateral Hippocampal Lesions,” the procedure Scoville performed on H.M. was no longer used on humans. (The ethics of Scoville’s treatment of H.M. are treated less gingerly by Philip J. Hilts in Memory’s Ghost.) The bilateral damage to the hippocampus found in other amnesic patients had a variety of causes, such as carbon monoxide poisoning and viral encephalitis, and these patients have been carefully studied by a number of investigators, such as Larry Squire and Stewart Zola in San Diego. In general, these patients had amnesic syndromes very similar to Henry’s, but they varied in severity from more to usually less severe than his. Other advances since Milner’s early work have been due to theoretical and experimental strides in the study of memory and the introduction of brain imaging. This latter technique has helped to identify the sites of damage in H.M. and other amnesic patients and enabled researchers to directly study brain activity before, during and after memory encoding and retrieval. The use of animal models has also greatly elucidated the critical areas for human memory, because memory mechanisms in monkeys and even rats appear to be similar to those in humans.
One basic question was the role of the hippocampus, as opposed to that of other medial temporal structures destroyed in H.M. Imaging and postmortem examination of the brains of patients with a variety of medial temporal damage made it clear that injury to the hippocampus alone will produce amnesia, but that the memory losses are considerably worse if the surrounding perirhinal and parahippocampal cortex was affected. Research on monkeys by Mortimer Mishkin at the National Institutes of Health and Zola, Squire and their colleagues also found that the severity of memory deficit depended on the amount of damage to both the hippocampus and surrounding cerebral cortex.
Milner had shown that H.M. could remember how to do mirror tracing and Gollin figures despite his total inability to remember new facts or events. What else could H.M. learn and remember? From the work of Corkin, Squire, their colleagues and others, it turned out that there were a number of types of memory that were intact in H.M. and other medial temporal amnesics. What was impaired was their “declarative memory,” consisting of “semantic” memory for general facts and “episodic memory” for specific experiences. What remained intact or relatively intact were types of “non- declarative” memory, each of which involves a brain system other than the medial temporal lobe. (There are many synonyms for these designations, because scientists would no sooner use someone else’s terminology than their toothbrush.) In general, declarative memory is often conscious, involved in “knowing,” reminiscing and planning, and it is what we commonly mean by “memory” and its loss as “amnesia.” The different nondeclarative memory systems tend to be unconscious, involved in “doing” and closer to the perceptual, motor and emotional systems of the brain.
One type of nondeclarative memory is called repetition priming. In the verbal version of the test for priming, the subject is shown a list of words. Later, the subject is shown a list of the first three letters of these words interposed with the first three letters of comparable words not included on the original list. He or she is asked to say the first word that comes to mind when each combination is seen. Memory is measured by the subject’s ability to produce the whole word after having seen it previously as a fragment. H.M. and other medial temporal patients have performed this task almost as well as normal subjects do. By contrast, when H.M. had been asked to memorize the original list, he would remember or recognize few or none of the items. Repetition priming can also be done with images or nonverbal material with the same results. The Gollin partial picture test that H.M. could do is a type of repetition priming, which is believed to depend on sensory areas of the cerebral cortex.
“Classical” conditioning is another type of nondeclarative memory, of which Ivan Pavlov’s experiment is the best-known example. Putting meat in a dog’s mouth produces salivation; after the meat and the sound of a bell are paired, the bell alone will produce salivation. Classical conditioning is intact in amnesics; it depends on the cerebellum, a large structure in the rear of the brain. Emotional memory is a related type of nondeclarative memory. It requires parts of the amygdala, a subcortical structure adjacent to the hippocampus.
One difference between H.M. and most other medial temporal amnesics is that his damage included part of the amygdala. This may be the reason that H.M., apparently unlike other amnesic patients without amygdala damage, was rather insensitive to pain, hunger and thirst and had no interest in sex. Perhaps the amygdala damage was also related to his very placid personality, described in detail by Corkin. The amygdala, she writes, “is critical for processing emotion, motivation, sexuality, and pain responses, particularly feelings of aggression and fear. Was this sweet, tractable man pacified by his operation?”
Departing from the scientific literature, Corkin speculates as well about the similarities between Buddhist meditation and Henry’s state of being:
Buddhism and other philosophies teach us that much of our suffering comes from our own thinking, particularly when we dwell in the past and in the future…. Meditation is a method for training the mind to have a new relationship with time, knowing only the present…. Dedicated mediators spend years practicing being attentive to the present—something Henry could not help but do.
When we consider how much of the anxiety and pain of daily life stems from attending to our long-term memories and worrying about and planning for the future, we can appreciate why Henry lived much of his life with relatively little stress…. [A] part of us all can understand how liberating it might be to always experience life as it is right now, in the simplicity of a world bounded by thirty seconds.
* * *
When testing H.M.’s autobiographical memory with more sophisticated prompts like pictures of TV luminaries, Corkin and her colleagues showed that Henry’s retrograde amnesia stretched back longer than had been realized, up to eleven years before the surgery. His semantic memory (facts, context-free memory) seemed much more intact than his episodic memory (personal experiences, content-rich memory). This discrepancy raises the question of whether these two types of declarative memory might have different critical sites in the brain.
In support of a distinction between the neural bases of semantic and episodic memory, Mortimer Mishkin and Faraneh Vargha-Khadem of University College, London, studied teens who had anoxia when very young that resulted in severe hippocampal damage, but with no damage to the surrounding cortical area. The teens had apparently normal semantic memory—they took regular classes in secondary school—but suffered severe loss of episodic memory, failing to recall what they wore the previous day or had eaten for lunch that day. Squire and Zola have opposing evidence that the degree of impairment on semantic and episodic memory is similar for varying lesions, implying that the two deficits cannot be dissociated by the site or size of damage. Resolution of this controversy will require new evidence, perhaps from more revealing behavioral tests or the development of higher-resolution brain imaging that can more accurately assess the site of damage.
Another current issue in memory research is the distinction between familiarity (I know or feel I have seen that) and recollection (I distinctly remember seeing that in a specific context). Corkin and her colleagues showed that H.M. had some familiarity with the previous pictures he had seen, and suggested that this might be because some of his perirhinal cortex was intact. This idea has been supported by studies of the differential activation of medial temporal structures and results from other patients with partial damage to different medial temporal structures.
Perhaps the central area of investigation in the study of declarative memory is determining the nature of the processes underlying the temporally graded retrograde amnesia manifested by H.M. and other amnesics. Why are relatively recent memories lost and older memories retained? Where are the older memories stored? The generally accepted model of the processes is known as “consolidation theory” and dates from 1900. It is based on a large body of experiments on rats, monkeys and humans involving many investigators and different techniques, from studying brain damage to imaging and molecular markers. In the contemporary model, new information comes into the brain first through a hierarchical series of areas linked to each sensory system; then it enters the hippocampus through the perirhinal and parahippocampal cortex and is “consolidated,” or organized, into declarative memories.
Consolidation—whatever it is—takes time. If the hippocampus is injured before the consolidation has finished, the incipient declarative memories will be lost. The length of consolidation time varies with species and the type of information. In rats and monkeys, consolidation of specific memories of some particular learning experience can take days or weeks. In humans, the consolidation time for some declarative memories can take years, as indicated by Henry’s loss of declarative memories for the eleven years before his surgery.
The process of consolidation involves “transferring” memories from the hippocampus to other regions of the cerebral cortex, where they are stored as long-term memories. There is a multiplicity of competing theories as to what “consolidation” is and the nature of the neural mechanisms underlying the metaphor “transfer” of declarative memory from the hippocampus to the cortex. In the hippocampus of adult mammals, including humans, new neurons form and make functional connections, a process known as “adult neurogenesis.” These neurons are believed to play a role in learning and memory. In rats and monkeys, most of them have a transient existence of days or weeks. As this is in the range of consolidation time for these species, the adult-generated neurons may play some role in temporary storage in the hippocampus before memories are “transferred” to the cortex.
One piece of the puzzle yet to be found is the link between the hippocampus’s role in declarative memory and its role in spatial memory. In the 1970s, John O’Keefe, then at University College, London, found single cells in the hippocampus of the rat that were active (“fired”) only when the rat was in a specific location in space. Thus, each time it ran a fixed route, say, through a maze, these “place cells” would fire in the same sequence. Furthermore, as shown more recently by Matt Wilson and his colleagues at MIT, when a rat that has learned the maze is dreaming (that is, in rapid eye movement sleep), the place cells fire in the same sequence as they did when the rat was running the maze. Although there is less evidence for place cells in humans, there are several lines of evidence for the role of the hippocampus in their spatial memory. The best-known concerns London taxi drivers, whose memorization of routes through the city’s labyrinthine streets qualifies them as having acquired “the Knowledge” and also seems to enlarge the size of their hippocampus. How this function of the hippocampus as a “cognitive map” relates to its role in declarative memory remains to be seen.