Overview

One of the most intriguing of the brain’s complex functions is the ability tostore information provided by experience and to retrieve much of it at will.Without this ability, many of the cognitive functions discussed in the preced-ing chapters could not occur. Learning is the name given to the process bywhich new information is acquired by the nervous system and is observablethrough changes in behavior. Memory refers to the encoding, storage, andretrieval of learned information. Equally fascinating (and important) is thenormal ability to forget information. Pathological forgetfulness, or amnesia,has been especially instructive about the neurological underpinnings ofmemory; amnesia is defined as the inability to learn new information or toretrieve information that has already been acquired. The importance ofmemory in daily life has made understanding these several phenomena oneof the major challenges of modern neuroscience, a challenge that has onlybegun to be met. The mechanisms of plasticity that provide plausible cellu-lar and molecular bases for some aspects of information storage have beenconsidered in Chapters 22 through 24. The present chapter summarizes thebroader organization of human memory, surveys the major clinical manifes-tations of memory disorders, and considers the implications of these disor-ders for ultimately understanding human memory in more detailed terms.

Qualitative Categories of Human Memory

Humans have at least two qualitatively different systems of information stor-age, which are generally referred to as declarative memory and nondeclara-tive memory (Figure 30.1; see also Box A). Declarative memory is the storage(and retrieval) of material that is available to consciousness and can beexpressed by language (hence, “declarative”). Examples of declarative mem-ory are the ability to remember a telephone number, a song, or the images ofsome past event. Nondeclarative memory (sometimes referred to as proce-dural memory), on the other hand, is not available to consciousness, at leastnot in any detail. Such memories involve skills and associations that are, byand large, acquired and retrieved at an unconscious level. Rememberinghow to dial the telephone, how to sing a song, how to efficiently inspect ascene, or making the myriad associations that occur continuously are allexamples of memories that fall in this category. It is difficult or impossible tosay how we do these things, and we are not conscious of any particularmemory during their occurrence. In fact, thinking about such activities mayactually inhibit the ability to perform them efficiently (thinking aboutexactly how to stroke a tennis ball or swing a golf club often makes mattersworse).

Chapter 30

733

Memory

Purves30 5/14/04 11:03 AM Page 733

734 Chapter Thir ty

Figure 30.1 The major qualitative cate-gories of human memory. Declarativememory includes those memories thatcan be brought to consciousness andexpressed as remembered events,images, sounds, and so on. Nondeclara-tive, or procedural, memory includesmotor skills, cognitive skills, simpleclassical conditioning, priming effects,and other information that is acquiredand retrieved unconsciously.

While it makes good sense to divide human learning and memory intocategories based upon the accessibility of stored information to consciousawareness, this distinction becomes problematic when considering learningand memory processes in animals. From an evolutionary point of view, it isunlikely that declarative memory arose de novo in humans with the develop-ment of language. Although some researchers continue to argue for differentclassifications in humans and other animals, recent studies suggest that sim-ilar memory processes operate in all mammals and that these memory func-tions are subserved by homologous neural circuitry. In other mammals,declarative memory typically refers to the storage of information whichcould, in principle, be declared through language (e.g., “the cheese is in thebox in the corner”) and that is dependent on the integrity of the medial tem-poral lobe and its associated structures (discussed later in the chapter). Non-declarative memory in other animals, as in humans, can be thought of asreferring to the learning and storage of sensory associations and motor skillsthat are not dependent on the medial temporal portions of the brain.

Temporal Categories of Memory

In addition to the types of memory defined by the nature of what is remem-bered, memory can also be categorized according to the time over which it iseffective. Although the details are still debated by both psychologists andneurobiologists, three temporal classes of memory are generally accepted(Figure 30.2). The first of these is immediate memory. By definition, imme-diate memory is the routine ability to hold ongoing experiences in mind for

Dailyepisodes

Words and their meanings

Motorskills

Primingcues

Puzzle- solving skills

Declarative(available to consciousness)

Nondeclarative(generally not available to consciousness)

Memory

History Associations

Immediate memory (fractions of a second−

seconds)

Working memory(seconds−minutes)

Forgetting

Long-term memory(days−years)Consolidation

Figure 30.2 The major temporal cate-gories of human memory.

Purves30 5/14/04 11:03 AM Page 734

fractions of a second. The capacity of immediate memory is very large andeach sensory modality (visual, verbal, tactile, and so on) appears to have itsown memory register.

Working memory, the second temporal category, is the ability to holdinformation in mind for seconds to minutes once the present moment has

Memor y 735

Box APhylogenetic MemoryA category of information storage notusually considered in standard accountsis memories that arise from the experi-ence of the species over the eons, estab-lished by natural selection acting on thecellular and molecular mechanisms ofneural development. Such stored infor-mation does not depend on postnatalexperience, but on what a given specieshas typically encountered in its environ-ment. These “memories” are no less con-sequential than those acquired by indi-vidual experience and are likely to havemuch underlying biology in commonwith the memories established during anindividual’s lifetime. (After all, phyloge-netic and ontogenetic memories arebased on neuronal connectivity.)

Information about the experience ofthe species, as expressed by endogenousor “instinctive” behavior, can be quitesophisticated, as is apparent in examplescollected by ethologists in a wide rangeof animals, including primates. The mostthoroughly studied instances of suchbehaviors are those occurring in youngbirds. Hatchlings arrive in the worldwith an elaborate set of innate behaviors.First is the complex behavior that allowsthe young bird to emerge from the egg.Having hatched, a variety of additionalbehaviors indicate how much of its earlylife is dependent on inherited informa-tion. Hatchlings of precocial species“know” how to preen, peck, gape theirbeaks, and carry out a variety of othercomplex acts immediately. In somespecies, hatchlings automatically crouchdown in the nest when a hawk passesoverhead but are oblivious to the over-

flight of an innocuous bird. KonradLorenz and Niko Tinbergen used hand-held silhouettes to explore this phenome-non in naïve herring gulls, as illustratedin the figure shown here. “It soonbecame obvious,” wrote Tinbergen, “that… the reaction was mainly one to shape.When the model had a short neck so thatthe head protruded only a little in frontof the line of the wings, it released alarm,independent of the exact shape of thedummy.” Evidently, the memory of whatthe shadow of a predator looks like isbuilt into the nervous system of thisspecies. Examples in primates includethe innate fear that newborn monkeyshave of snakes and looming objects.

Despite the relatively scant attentionpaid to this aspect of memory, it is proba-bly the most important component of thestored information in the brain thatdetermines whether or not an individualsurvives long enough to reproduce.

ReferencesTINBERGEN, N. (1969) Curious Naturalists. Gar-den City, NY: Doubleday.TINBERGEN, N. (1953) The Herring Gull’s World.New York: Harper & Row.LORENZ, K. (1970) Studies in Animal andHuman Behaviour. (Translated by R. Martin.)Cambridge, MA: Harvard University Press.DUKAS, R. (1998) Cognitive Ecology. Chicago:University of Chicago Press.

Direction ofmovement

(A) (B)

(A) Niko Tinbergen at work. (B) Sil-houettes used to study alarm reac-tions in hatchlings. The shapes thatwere similar to the shadow of thebird’s natural predators (red arrows)when moving in the appropriatedirection elicited escape responses(crouching, crying, seeking cover);silhouettes of songbirds and otherinnocuous species (or geometricalforms) elicited no obvious response.(From Tinbergen, 1969.)

Purves30 5/14/04 11:03 AM Page 735

736 Chapter Thir ty

passed. An everyday example of working memory is searching for a lostobject; working memory allows the hunt to proceed efficiently, avoidingplaces already inspected. A conventional way of testing the integrity ofworking memory at the bedside is to present a string of randomly ordereddigits, which the patient is then asked to repeat; surprisingly, the normal“digit span” is only 7–9 numbers.

The third temporal category is long-term memory and entails the reten-tion of information in a more permanent form of storage for days, weeks, oreven a lifetime. There is general agreement that the so-called engram—thephysical embodiment of the long-term memory in neuronal machinery—depends on long-term changes in the efficacy of transmission of the relevantsynaptic connections, and/or the actual growth and reordering of such con-nections. As discussed in Chapter 24, there is good reason to think that boththese varieties of synaptic change occur.

Evidence for a continual transfer of information from working memory tolong-term memory, or consolidation (Figure 30.2), is apparent in the phe-nomenon of priming. Priming is typically demonstrated by presenting sub-jects with a set of items to which they are exposed under false pretenses. Forexample, a list of words can be given with the instruction that the subjectsare to identify some feature that is actually extraneous to the experiment(e.g., whether the words are verbs, adjectives, or nouns). Sometime thereafter(e.g., the next day) the same individuals are given a different test in whichthey are asked to fill in the missing letters of words with whatever letterscome to mind. The test list actually includes fragments of words that werepresented in the first test, mixed among fragments of words that were not.Subjects fill in the letters to make the words that were presented earlier at ahigher rate than expected by chance, even though they have no specificmemory of the words that were seen initially; moreover, they are faster atfilling in letters to make words that were seen earlier than new words. Prim-ing shows that information previously presented is influential, even thoughwe are entirely unaware of its effect on subsequent behavior. The signifi-cance of priming is well known—at least intuitively—to advertisers, teach-ers, spouses, and others who want to influence the way we think and act.

Despite the prevalence of such transfer, the information stored in thisprocess is not particularly reliable. Consider, for instance, the list of words inTable 30.1A. If the list is read to a group of students who are immediatelyasked to identify which of several items were on the original list and whichwere not (Table 30.1B), the result is surprising. Typically, about half the stu-dents report that the word “sweet” was included in the list in Table 30.1A;moreover, they are quite certain about it! The mechanism of such erroneous“recognition” is presumably the strong associations that have previously beenmade between the words on the list in Table 30.1A and the word “sweet,”which bias the students to think that “sweet” was a member of the originalset. Clearly, memories, even those we feel quite confident about, are oftenfalse.

The Importance of Association in Information Storage

The normal human capacity for remembering relatively meaningless infor-mation is surprisingly limited (as noted, a string of about 7–9 numbers orother arbitrary items). This capacity, however, can be increased dramatically.For example, a college student who for some months spent an hour each daypracticing the task of remembering randomly presented numbers was ableto recall a string of up to about 80 digits (Figure 30.3). He did this primarily

TABLE 30.1The Fallibility of Human Memorya

(A)Initial list (B)Subsequentof words test list

candy tastesour pointsugar sweetbitter chocolategood sugartaste nicetoothnicehoneysodachocolateheartcakeeatpie

aAfter hearing the words in list A read aloud, sub-jects were asked to identify which of the items inlist B had also been on list A. See text for theresults.

Purves30 5/14/04 11:03 AM Page 736

by making subsets of the string of numbers he was given signify dates ortimes at track meets (he was a competitive runner)—in essence, givingmeaningless items a meaningful context. This same strategy of association isused by most professional “mnemonists,” who amaze audiences by appar-ently prodigious feats of memory. Similarly, a good chess player can remem-ber the position of many more pieces on a briefly examined board than apoor player, presumably because the positions have much more significancefor individuals who know the intricacies of the game than for neophytes(Figure 30.4). Thus, the capacity of working memory very much depends on

Memor y 737

Practice (5-day blocks)

0

10

20

30

40

50

60

70

80

Dig

it s

pan

5 15 25 35

Figure 30.3 Increasing the digit span by practice (and the development of associa-tional strategies). During many months involving one hour of practice a day for 3–5days a week, this subject increased his digit span from 7 to 79 numbers. Randomdigits were read to him at the rate of one per second. If a sequence was recalled cor-rectly, one digit was added to the next sequence. (After Ericsson et al., 1980.)

Figure 30.4 The retention of brieflypresented information depends on pastexperience, context, and its perceivedimportance. (A) Board position afterwhite’s twenty-first move in game 10 ofthe 1985 World Chess Championshipbetween A. Karpov (white) and G. Kas-parov (black). (B) A random arrange-ment of the same 28 pieces. (C, D) Afterbriefly viewing the board from the realgame, master players reconstruct thepositions of the pieces with muchgreater efficiency than beginning play-ers. With a randomly arranged board,however, beginners perform as well orbetter than accomplished players. (AfterChase and Simon, 1973.)

(C) Real game (D) Randomly arranged

Black Black

(A) (B)

Black

White

Black

White

1 2 3 4 5 6 7Trials

Cor

rect

pie

ces

4

8

12

16

20

24

0

Master

Beginner

1 2 3 4 5 6 7Trials

Cor

rect

pie

ces

4

8

12

16

20

24

0

Master

Beginner

Purves30 5/14/04 11:03 AM Page 737

738 Chapter Thir ty

what the information in question means to the individual and how readily itcan be associated with information that has already been stored.

The ability of humans to remember significant information in the normalcourse of events is, in fact, enormous. Consider Arturo Toscanini, the lateconductor of the NBC Philharmonic Orchestra, who allegedly kept in hishead the complete scores of more than 250 orchestral works, as well as themusic and librettos for some 100 operas. Once, just before a concert in St.Louis, the first bassoonist approached Toscanini in some consternationbecause he had just discovered that one of the keys on his bassoon was bro-ken. After a minute or two of deep concentration, the story goes, Toscaniniturned to the alarmed bassoonist and informed him that there was no needfor concern, since that note did not appear in any of the bassoon parts for theevening’s program.

A parallel example of a prodigious quantitative memory is the mathe-matician Alexander Aitken. After an undistinguished career in elementaryschool, the 13-year-old Aitken was greatly taken with the manipulation ofnumbers. For the next four years he undertook, as a personal challenge, tomaster mental calculation. He began by memorizing the value of π to 1000places, and could soon do calculations in his head with such facility that hebecame a local celebrity. When asked for the squares of three-digit numbers,he was able to give these almost instantly. The square roots for each wereproduced to five significant digits in 2–3 seconds; the squares of four-digitnumbers allegedly took him about 5 seconds. Aitken went on to become aprofessor of mathematics at Edinburgh and was eventually elected a Fellowof the Royal Society for his contributions to numerical mathematics, statis-tics, and matrix algebra. At the age of 30 or so, he began to lose his enthusi-asm for “mental yoga,” as he called his penchant. In part, his waning enthu-siasm stemmed from the realization that the advent of calculators wasmaking his prowess obsolete (it was then 1930). He also discovered that thelast 180 digits of π that he had memorized as a boy were wrong; he hadtaken the values from the published work of another mental calculator, whoerred in an era when there was no way to check the correct value. In fact,Aitken’s feat has long since been superseded. In 1981, an Indian mnemonistmemorized the value of π to 31,811 places, only to have a Japanese mne-monist increase this record to 40,000 places a few years later!

Toscanini’s and Aitken’s mental processes in these feats were not rotelearning, but a result of the fascination that aficionados bring to their specialinterests (Box B). Although few can boast the mnemonic prowess of suchindividuals, the human ability to remember the things that deeply interestus—whether baseball statistics, soap opera plots, or the details of brain struc-ture—is amazing.

Forgetting

Some years ago, a poll showed that 84% of psychologists agreed with thestatement “everything we learn is permanently stored in the mind, althoughsometimes particular details are not accessible.” The 16% who thought oth-erwise should get the higher marks. Common sense indicates that, were itnot for forgetting, our brains would be impossibly burdened with the welterof useless information that is briefly encoded in our immediate memory“buffer.” In fact, the human brain is very good at forgetting. In addition tothe unreliable performance on tests such as the example in Table 30.1, Figure30.5 shows that the memory of the appearance of a penny (an icon seen

Purves30 5/14/04 11:03 AM Page 738

thousands of times since childhood) is uncertain at best, and that peoplegradually forget what they have seen over the years (TV shows, in this case).Clearly we forget things that have no particular importance, and unusedmemories deteriorate over time.

The ability to forget unimportant information may be as critical for nor-mal life as retaining information that is significant. One sort of evidence forthis presumption is rare individuals who have difficulty with the normalerasure of information. Perhaps the best-known case is a subject studied overseveral decades by the Russian psychologist A. R. Luria, who referred to the

Memor y 739

Box BSavant SyndromeA fascinating developmental anomaly ofhuman memory is seen in rare individu-als who until recently were referred to asidiots savants; the current literature tendsto use the less pejorative phrase savantsyndrome. Savants are people who, for avariety of poorly understood reasons(typically brain damage in the perinatalperiod), are severely restricted in mostmental activities but extraordinarilycompetent and mnemonically capaciousin one particular domain. The grosslydisproportionate skill compared to therest of their limited mental life can bestriking. Indeed, these individuals—whose special talent may be in calcula-tion, history, art, language, or music—areusually diagnosed as severely retarded.

Many examples could be cited, but asummary of one such case suffices tomake the point. The individual whosehistory is summarized here was giventhe fictitious name “Christopher” in adetailed study carried out by psycholo-gists Neil Smith and Ianthi-Maria Tsim-pli. Christopher was discovered to beseverely brain damaged at just a fewweeks of age (perhaps as the result ofrubella during his mother’s pregnancy,or anoxia during birth; the record isuncertain in this respect). He had beeninstitutionalized since childhood becausehe was unable to care for himself, couldnot find his way around, had poor hand-

eye coordination, and a variety of otherdeficiencies. Tests on standard IQ scaleswere low, consistent with his generalinability to cope with daily life. Scores onthe Wechsler Scale were, on differentoccasions, 42, 67, and 52.

Despite his severe mental incapacita-tion, Christopher took an intense interestin books from the age of about three, par-ticularly those providing factual infor-mation and lists (e.g., telephone directo-ries and dictionaries). At about six orseven he began to read technical papersthat his sister sometimes brought homefrom work, and he showed a surprisingproficiency in foreign languages. Hisspecial talent in the acquisition and useof language (an area in which savants areoften especially limited) grew rapidly. Asan early teenager, Christopher couldtranslate from—and communicate in—avariety of languages in which his skillswere described as ranging from rudi-mentary to fluent; these included Dan-ish, Dutch, Finnish, French, German,modern Greek, Hindi, Italian, Norwe-gian, Polish, Portuguese, Russian, Span-ish, Swedish, Turkish, and Welsh. Thisextraordinary level of linguistic accom-plishment is all the more remarkablesince he had no formal training in lan-guage even at the elementary schoollevel, and could not play tic-tac-toe orcheckers because he was unable to grasp

the rules needed to make moves in thesegames.

The neurobiological basis for suchextraordinary individuals is not under-stood. It is fair to say, however, thatsavants are unlikely to have an ability intheir areas of expertise that exceeds thecompetency of normally intelligent indi-viduals who focus passionately on a par-ticular subject (several examples aregiven in the text). Presumably, thesavant’s intense interest in a particularcognitive domain is due to one or morebrain regions that continue to work rea-sonably well. Whether because of socialfeedback or self-satisfaction, savantsclearly spend a great deal of their mentaltime and energy practicing the skill theycan exercise more or less normally. Theresult is that the relevant associationsthey make become especially rich, asChristopher’s case demonstrates.

References MILLER, L. K. (1989) Musical Savants: Excep-tional Skill in the Mentally Retarded. Hillsdale,New Jersey: Lawrence Erlbaum Associations.SMITH, N. AND I.-M. TSIMPLI (1995) The Mindof a Savant: Language Learning and Modularity.Oxford, England: Basil Blackwell Ltd.HOWE, M. J. A. (1989) Fragments of Genius: TheStrange Feats of Idiots Savants. Routledge,New York: Chapman and Hall.

Purves30 5/14/04 11:03 AM Page 739

740 Chapter Thir ty

subject simply as “S.” Luria’s description of an early encounter gives someidea why S, then a newspaper reporter, was so interesting:

I gave S a series of words, then numbers, then letters, reading them to himslowly or presenting them in written form. He read or listened attentively andthen repeated the material exactly as it had been presented. I increased thenumber of elements in each series, giving him as many as thirty, fifty, or evenseventy words or numbers, but this too, presented no problem for him. He didnot need to commit any of the material to memory; if I gave him a series ofwords or numbers, which I read slowly and distinctly, he would listen atten-tively, sometimes ask me to stop and enunciate a word more clearly, or, if indoubt whether he had heard a word correctly, would ask me to repeat it.Usually during an experiment he would close his eyes or stare into space, fix-ing his gaze on one point; when the experiment was over, he would ask thatwe pause while he went over the material in his mind to see if he had retainedit. Thereupon, without another moment’s pause, he would reproduce theseries that had been read to him.

A. R. Luria (1987), The Mind of a Mnemonist, pp. 9–10

S’s phenomenal memory, however, did not always serve him well. He haddifficulty ridding his mind of the trivial information that he tended to focuson, sometimes to the point of incapacitation. As Luria put it:

Thus, trying to understand a passage, to grasp the information it contains(which other people accomplish by singling out what is most important)

(B)a

(A)

Time event occurred (years ago)

80

70

60

50

4020 4 6 8 10 12 14

Perc

ent c

orre

ctf

k

b

g

l

c

h

m

d

i

n

e

j

o

Figure 30.5 Forgetting. (A) Different versions of the “heads” side of a penny.Despite innumerable exposures to this familiar design, few people are able to pickout (a) as the authentic version. Clearly, repeated information is not necessarilyretained. (B) The deterioration of long-term memories was evaluated in this exam-ple by a multiple-choice test in which the subjects were asked to recognize thenames of television programs that had been broadcast for only one season duringthe past 15 years. Forgetting of stored information that is no longer used evidentlyoccurs gradually and progressively over the years (chance performance = 25%). (Aafter Rubin and Kontis, 1983; B after Squire, 1989.)

Purves30 5/14/04 11:03 AM Page 740

became a tortuous procedure for S, a struggle against images that kept risingto the surface in his mind. Images, then, proved an obstacle as well as an aidto learning in that they prevented S from concentrating on what was essential.Moreover, since these images tended to jam together, producing still moreimages, he was carried so far adrift that he was forced to go back and rethinkthe entire passage. Consequently, a simple passage—a phrase, for that mat-ter—would turn out to be a Sisyphean task.

Ibid., p. 113Although forgetting is a normal and apparently essential mental process,

it can also be pathological, a condition called amnesia. Some of the causes ofmemory loss are listed in Table 30.2. An inability to establish new memoriesfollowing neurological insult is called anterograde amnesia, whereas diffi-culty retrieving memories established prior to the precipitating neuropathol-ogy is called retrograde amnesia. Anterograde and retrograde amnesia areoften present together, but can be dissociated under various circumstances.Amnesias following bilateral lesions of the temporal lobe and diencephalonhave given particular insight into where and how at least some categories ofmemory are formed and stored, as discussed in the next section.

Brain Systems Underlying Declarative Memory Formation

Three extraordinary clinical cases of amnesia have been especially revealingabout the brain systems responsible for the short-term storage and consoli-dation of declarative information and are now familiar to neurologists andpsychologists as patients H.M., N.A., and R.B. (Box C). Taken together, thesecases provide dramatic evidence of the importance of midline diencephalicand medial temporal lobe structures—the hippocampus, in particular—inestablishing new declarative memories (Figure 30.6). These patients alsodemonstrate that there is a different anatomical substrate for anterogradeand retrograde amnesia, since in each of these individuals, memory forevents prior to the precipitating injury was largely retained.

The devastating deficiency is (or was, in the case of R.B.) the inability toestablish new memories. Retrograde amnesia—the loss of memory forevents preceding an injury or illness—is more typical of the generalized

Memor y 741

TABLE 30.2Causes of Amnesia

Causes Examples Site of damage

Vascular occlusion of Patient R.B. (Box C) Bilateral medial temporal lobe,both posterior cerebral the hippocampus inarteries particular

Midline tumors — Medial thalamus bilaterally(hippocampus and otherrelated structures if tumor islarge enough)

Trauma Patient N.A. (Box C) Bilateral medial temporal lobeSurgery Patient H.M. (Box C) Bilateral medial temporal lobeInfections Herpes simplex Bilateral medial temporal lobe

encephalitisVitamin B1 deficiency Korsakoff’s Medial thalamus and

syndrome mammillary bodiesElectroconvulsive therapy — Uncertain

(ECT) for depression

Purves30 5/14/04 11:03 AM Page 741

742 Chapter Thir ty

Box CClinical Cases That Reveal the Anatomical Substrate for Declarative Memories

The Case of H.M.H.M. had suffered minor seizures sinceage 10 and major seizures since age 16.At the age of 27, he underwent surgeryto correct his increasingly debilitatingepilepsy. A high school graduate, H.M.had been working as a technician in asmall electrical business until shortlybefore the time of his operation. Hisattacks involved generalized convulsionswith tongue biting, incontinence, andloss of consciousness (all typical of grandmal seizures). Despite a variety of med-ications, the seizures remained uncon-trolled and increased in severity. A fewweeks before his surgery, H.M. becameunable to work and had to quit his job.

On September 1, 1953, surgeons per-formed a bilateral medial temporal loberesection in which the amygdala, uncus,hippocampal gyrus, and anterior two-thirds of the hippocampus were re-moved. At the time, it was unclear thatbilateral surgery of this kind wouldcause a profound memory defect. Severeamnesia was evident, however, uponH.M.’s recovery from the operation, andhis life was changed radically.

The first formal psychological examof H.M. was conducted nearly 2 yearsafter the operation, at which time a pro-found memory defect was still obvious.Just before the examination, for instance,H.M. had been talking to the psycholo-gist; yet he had no recollection of thisexperience a few minutes later, denyingthat anyone had spoken to him. He gavethe date as March 1953 and seemedoblivious to the fact that he had under-gone an operation, or that he had be-come incapacitated as a result. None-theless, his score on the Wechsler-Bellevue Intelligence Scale was 112, avalue not significantly different from hispreoperative IQ. Various psychologicaltests failed to reveal any deficiencies in

perception, abstract thinking, or reason-ing; he seemed highly motivated and, inthe context of casual conversation, nor-mal. Importantly, he also performed wellon tests of the ability to learn new skills,such as mirror writing or puzzle solving(that is, his ability to form proceduralmemories was intact). Moreover, hisearly memories were easily recalled,showing that the structures removedduring H.M.’s operation are not a perma-nent repository for such information. Onthe Wechsler Memory Scale (a specifictest of declarative memory), however, heperformed very poorly, and he could notrecall a preceding test-set once he hadturned his attention to another part ofthe exam. These deficits, along with hisobvious inability to recall events in hisdaily life, all indicate a profound loss ofshort-term declarative memory function.

During the subsequent decades, H.M.has been studied extensively, primarilyby Brenda Milner and her colleagues atthe Montreal Neurological Institute. Hismemory deficiency has continuedunabated, and, according to Milner, hehas little idea who she is in spite of theiracquaintance for nearly 50 years. Sadly,he has gradually come to appreciate hispredicament. “Every day is alone,” H.M.reports, “whatever enjoyment I’ve hadand whatever sorrow I’ve had.”

The Case of N.A.N.A. was born in 1938 and grew up withhis mother and stepfather, attendingpublic schools in California. After a yearof junior college, he joined the Air Force.In October of 1959 he was assigned tothe Azores as a radar technician andremained there until December 1960,when a bizarre accident made him a cel-ebrated neurological case.

N.A. was assembling a model air-plane in his barracks room while, unbe-

knownst to him, his roommate was prac-ticing thrusts and parries with a minia-ture fencing foil behind N.A.’s chair.N.A. turned suddenly and was stabbedthrough the right nostril. The foil pene-trated the cribriform plate (the structurethrough which the olfactory nerve entersthe brain) and took an upward courseinto the left forebrain. N.A. lost con-sciousness within a few minutes (pre-sumably because of bleeding in theregion of brain injury) and was taken toa hospital. There he exhibited a right-sided weakness and paralysis of the righteye muscles innervated by the third cra-nial nerve. Exploratory surgery wasundertaken and the dural tear repaired.Gradually he recovered and was senthome to California. After some months,his only general neurological deficitswere some weakness of upward gazeand mild double vision. He retained,however, a severe anterograde amnesiafor declarative memories. MRI studiesfirst carried out in 1986 showed exten-sive damage to the thalamus and themedial temporal lobe, mostly on theright side; the mammillary bodies alsoappeared to be missing bilaterally. Theexact extent of his lesion, however, is notknown, as N.A. remains alive and well.

N.A.’s memory from the time of hisinjury over 40 years ago to the presenthas remained impaired and, like H.M.,he fails badly on formal tests of newlearning ability. His IQ is 124, and heshows no defects in language skills, per-ception, or other measures of intelli-gence. He also learns new proceduralskills quite normally. His amnesia is notas dense as that of H.M. and is more ver-bal than spatial. He can, for example,draw accurate diagrams of material pre-sented to him earlier. Nonetheless, heloses track of his possessions, forgetswhat he has done, and tends to forget

Purves30 5/14/04 11:03 AM Page 742

Memor y 743

who has come to visit him. He has onlyvague impressions of political, social,and sporting events that have occurredsince his injury. Watching television isdifficult because he tends to forget thestoryline during commercials. On theother hand, his memory for events priorto 1960 is extremely good; indeed, hislifestyle tends to reflect the 1950s.

The Case of R.B.At the age of 52, R.B. suffered anischemic episode during cardiac bypasssurgery. Following recovery from anes-thesia, a profound amnesic disorder wasapparent. As in the cases of H.M. andN.A., his IQ was normal (111), and heshowed no evidence of cognitive defectsother than memory impairment. R.B.was tested extensively for the next five

years, and, while his amnesia was not assevere as that of H.M. or N.A., he consis-tently failed the standard tests of theability to establish new declarative mem-ories. When R.B. died in 1983 of conges-tive heart failure, a detailed examinationof his brain was carried out. The onlysignificant finding was bilateral lesionsof the hippocampus—specifically, cellloss in the CA1 region that extended thefull rostral–caudal length of the hippo-campus on both sides. The amygdala,thalamus, and mammillary bodies, aswell as the structures of the basal fore-brain, were normal. R.B.’s case is particu-larly important because it suggests thathippocampal lesions alone can result inprofound anterograde amnesia fordeclarative memory.

ReferencesCORKIN, S. (1984) Lasting consequences ofbilateral medial temporal lobectomy: Clinicalcourse and experimental findings in H.M.Semin. Neurol. 4: 249–259.CORKIN, S., D. G. AMARAL, R. G. GONZÁLEZ, K.A. JOHNSON AND B. T. HYMAN (1997) H. M.’smedial temporal lobe lesion: Findings fromMRI. J. Neurosci. 17: 3964-3979.HILTS, P. J. (1995) Memory’s Ghost: The StrangeTale of Mr. M. and the Nature of Memory. NewYork: Simon and Schuster.MILNER, B., S. CORKIN AND H.-L. TEUBER

(1968) Further analysis of the hippocampalamnesic syndrome: A 14-year follow-upstudy of H.M. Neuropsychologia 6: 215–234.SCOVILLE, W. B. AND B. MILNER (1957) Loss ofrecent memory after bilateral hippocampallesions. J. Neurol. Neurosurg. Psychiat. 20:11–21.SQUIRE, L. R., D. G. AMARAL, S. M. ZOLA-MOR-GAN, M. KRITCHEVSKY AND G. PRESS (1989)Description of brain injury in the amnesicpatient N.A. based on magnetic resonanceimaging. Exp. Neurol. 105: 23–35.TEUBER, H. L., B. MILNER AND H. G. VAUGHN

(1968) Persistent anterograde amnesia afterstab wound of the basal brain. Neuropsy-chologia 6: 267–282.ZOLA-MORGAN, S., L. R. SQUIRE AND D. AMA-RAL (1986) Human amnesia and the medialtemporal region: Enduring memory impair-ment following a bilateral lesion limited tothe CA1 field of the hippocampus. J. Neu-rosci. 6: 2950–2967.

(A)

(B) (C) (D)

Damaged area Anterior Posterior

MRI images of the brain ofpatient H.M. (A) Sagittalview of the right hemi-sphere; the area of the ante-

rior temporal lobectomy is indicated by thewhite dotted line. The intact posterior hip-pocampus is the banana-shaped object indi-cated by the white arrow. (B–D) Coronal sec-tions at approximately the levels indicatedby the red lines in (A). Image (B) is the mostrostral and is at the level of the amygdala.The amygdala and the associated cortex areentirely missing. Image (C) is at the level ofthe rostral hippocampus; again, this struc-ture and the associated cortex have beenremoved. Image (D) is at the caudal level ofthe hippocampus; the posterior hippocam-pus appears intact, although somewhatshrunken. Outlines below give a clearer indi-cation of the parts of H.M.’s brain that havebeen ablated (black shading). (From Corkinet al., 1997.)

Purves30 5/14/04 11:03 AM Page 743

Chapter Thirty

lesions associated with head trauma and neurodegenerative disorders, suchas Alzheimer’s disease (Box D). Although a degree of retrograde amnesiacan occur with the more focal lesions that cause anterograde amnesia, thelong-term storage of memories is presumably distributed throughout thebrain (see the next section). Thus, the hippocampus and related diencephalicstructures indicated in Figure 30.6 form and consolidate declarative memo-ries that are ultimately stored elsewhere.

Other causes of amnesia have also provided some insight into the parts ofthe brain relevant to various aspects of memory (see Table 30.2). Korsakoff’ssyndrome, for example, occurs in chronic alcoholics as a result of thiamine(vitamin B1) deficiency. In such cases, loss of brain tissue occurs bilaterally inthe mammillary bodies and the medial thalamus, for reasons that are notwell understood.

Studies of animals with lesions of the medial temporal lobe have largelycorroborated these findings with human patients. For example, one test ofthe presumed equivalent of declarative memory formation in animalsinvolves placing rats into a pool filled with opaque water, thus concealing asubmerged platform; note that the pool is surrounded by prominent visuallandmarks (Figure 30.7). Normal rats at first search randomly until they find

Fornix

Hippocampus

Hippocampus

Optic nerve

Mammillary body

Lateralventricle

Temporal lobe

Fornix

Thalamus

AmygdalaRhinalcortex

Mammillarybody

Prefrontalcortex

Basalforebrain

(A) Brain areas associated with declarative memory disorders (B) Ventral view of hippocampus and related structures with part of temporal lobes removed

Tail of the caudate nucleus

Inferior horn of the lateral ventricle

Hippocampus

Hippocampalfissure

Optic tract

Inferior portionof the temporallobe

(C) Hippocampus in coronal section

Corpus callosum

Figure 30.6 Brain areas that, when damaged,tend to give rise to declarative memory disor-ders. By inference, declarative memory is basedon the physiological activity of these structures.(A) Studies of amnesic patients have shownthat the formation of declarative memoriesdepends on the integrity of the hippocampusand its subcortical connections to the mammil-lary bodies and dorsal thalamus. (B) Diagramshowing the location of the hippocampus in acutaway view in the horizontal plane. (C) Thehippocampus as it would appear in a histologi-cal section in the coronal plane, at approxi-mately the level indicated by the line in (B).

Purves30 5/14/04 11:03 AM Page 744

744

the submerged platform. After repeated testing, however, they learn to swimdirectly to the platform no matter where they are initially placed in the pool.Rats with lesions to the hippocampus and nearby structures cannot learn tofind the platform, suggesting that remembering the location of the platform

Memor y 745

Figure 30.7 Spatial learning and mem-ory in rodents depends on the hip-pocampus. (A) Rats are placed in a cir-cular tank about the size and shape of achild’s wading pool filled with opaque(milky) water. The surrounding envi-ronment contains visual cues such aswindows, doors, a clock, and so on. Asmall platform is located just below thesurface. As rats search for this restingplace, the pattern of their swimming(indicated by the traces in C) is moni-tored by a video camera. (B) After a fewtrials, normal rats rapidly reduce thetime required to find the platform,whereas rats with hippocampal lesionsdo not. Sample swim paths of normalrats (C) and hippocampal lesioned rats(D) on the first and tenth trials. Ratswith hippocampal lesions are unable toremember where the platform is located(B after Eichenbaum, 2000; C,D afterSchenk and Morris, 1985).

(C) Control rat

( )

(D) Rat with hippocampus lesioned

First trial After 10 trials

First trial After 10 trials

Hidden platform

0

10

20

30

40

50

60

Trials0 2–6 7–12 13–18

Hippocampus-lesioned

Control

Videocamera

Mea

n la

tenc

y

(B)(A)

Purves30 5/14/04 11:03 AM Page 745

746 Chapter Thir ty

relative to the configuration of visual landmarks depends on the sameneural structures critical to declarative memory formation in humans. Like-wise, destruction of the hippocampus and parahippocampal gyrus in mon-keys severely impairs their ability to perform delayed-response tasks (seeFigure 25.13). These studies suggest that primates and other mammalsdepend on medial temporal structures such as the hippocampus and para-hippocampal gyrus to encode and consolidate memories of events andobjects in time and space, just as humans use these same brain regions forthe initial encoding and consolidation of declarative memories.

Brain Systems Underlying Long-Term Storage of Declarative Memory

Revealing though they have been, clinical studies of amnesic patients haveprovided relatively little insight into the long-term storage of declarativeinformation in the brain (other than to indicate quite clearly that such infor-mation is not stored in the midline diencephalic and medial temporal lobestructures that are affected in anterograde amnesia). Nonetheless, a gooddeal of evidence implies that the cerebral cortex is the major long-termrepository for many aspects of declarative memory.

One line of evidence comes from observations of patients undergoingelectroconvulsive therapy (ECT). Individuals with severe depression areoften treated by the passage of enough electrical current through the brain tocause the equivalent of a full-blown seizure (this procedure is done underanesthesia, in well-controlled circumstances). This remarkably useful treat-ment was discovered because depression in epileptics was perceived to remitafter a spontaneous seizure (see Box C in Chapter 24). However, ECT oftencauses both anterograde and retrograde amnesia. Patients typically do notremember the treatment itself or the events of the preceding days, and eventheir recall of events of the previous 1–3 years can be affected. Animal stud-ies (rats tested for maze learning, for example) have confirmed the amnesicconsequences of ECT. The memory loss usually clears over a period of weeksto months. However, to mitigate this side effect (which may be the result ofexcitotoxicity; see Box B in Chapter 6), ECT is often delivered to only onehemisphere at a time. The nature of amnesia following ECT supports theconclusion that long-term declarative memories are widely stored in thecerebral cortex, since this is the part of the brain predominantly affected bythis therapy.

A second line of evidence comes from patients with damage to associa-tion cortex outside the medial temporal lobe. Since different cortical regionshave different cognitive functions (see Chapters 25 and 26), it is not surpris-ing that these sites store information that reflects the cognitive function ofthat part of the brain. For example, the lexicon that links speech sounds andtheir symbolic significance is located in the association cortex of the superiortemporal lobe, and damage to this area typically results in an inability to linkwords and meanings (Wernicke’s aphasia; see Chapter 26). Presumably, thewidespread connections of the hippocampus to the language areas serve toconsolidate declarative information in these and other language-related cor-tical sites (Figure 30.8). By the same token, the inability of patients with tem-poral lobe lesions to recognize objects and/or faces suggests that such mem-ories are stored there (see Chapter 25).

A third sort of evidence supporting the hypothesis that declarative mem-ories are stored in cortical areas specialized for processing particular types of

Purves30 5/14/04 11:03 AM Page 746

information comes from neuroimaging of human subjects recalling vividmemories. In one such study, subjects first examined words paired witheither pictures or sounds. Their brains were then scanned while they wereasked to recall whether each test word was associated with either a pictureor a sound. Functional images based on these scans showed that the corticalareas activated when subjects viewed pictures or heard sounds were reacti-vated when these percepts were vividly recalled. In fact, this sort of reactiva-tion can be quite specific. Thus, different classes of visual images—such asfaces, houses, or chairs—tend to reactivate the same small regions of thevisual association cortex that were activated when the objects were actuallyperceived (Figure 30.9).

These neuroimaging studies reinforce the conclusion that declarativememories are stored widely in specialized areas of the cerebral cortex.Retrieving such memories appears to involve the medial temporal lobe, aswell as regions of the frontal cortex. Frontal cortical areas located on the dor-solateral and anterolateral aspect of the brain, in particular, are activatedwhen normal subjects attempt to retrieve declarative information from long-term memory. Moreover, patients with damage to these areas often fail toaccurately recall the details of a memory and sometimes resort to confabula-tion to fill in the missing information. Finally, whereas the ability of patientssuch as H.M., N.A., and R.B. to remember facts and events from the periodof their lives preceding their lesions clearly demonstrates that the medialtemporal lobe is not necessary for retrieving declarative information held inlong-term memory, other studies have suggested that these structures maybe important for recalling declarative memories during the early stages ofconsolidation and storage in the cerebral cortex.

Memor y 747

Medial view Lateral view

Hippocampus

Widespread projections from associationneocortex converge on the hippocampal region.The output of the hippocampus is ultimatelydirected back to these same neocortical areas.

Figure 30.8 Connections between the hippocampus and possible declarative mem-ory storage sites. The rhesus monkey brain is shown because these connections aremuch better documented in non-human primates than in humans. Projections fromnumerous cortical areas converge on the hippocampus and the related structuresknown to be involved in human memory; most of these sites also send projectionsto the same cortical areas. Medial and lateral views are shown, the latter rotated180° for clarity. (After Van Hoesen, 1982.)

Purves30 5/14/04 11:03 AM Page 747

748 Chapter Thir ty

Brain Systems Underlying Nondeclarative Learning and Memory

H.M., N.A., and R.B. had no difficulty establishing or recalling nondeclarativememories, indicating that this information is laid down by using an anatom-ical substrate different from that used in declarative memory formation. Non-declarative memory apparently involves the basal ganglia, prefrontal cortex,amygdala, sensory association cortex, and cerebellum, but not the medialtemporal lobe or midline diencephalon. In support of this interpretation, per-ceptual priming (the influence of previously studied information on subse-quent performance, unavailable to conscious recall) depends critically on theintegrity of sensory association cortex. For example, lesions of the visual asso-ciation cortex produce profound impairments in visual priming but leavedeclarative memory formation intact. Likewise, simple sensory-motor condi-tioning, such as learning to blink following a tone that predicts a puff of airdirected at the eye, relies on the normal activation of neural circuits in thecerebellum. Ischemic damage to the cerebellum following infarcts of thesuperior cerebellar artery or the posterior inferior cerebellar artery cause pro-found deficits in classical eyeblink conditioning without interfering with the

Figure 30.9 Reactivation of visual cor-tex during vivid remembering of visualview images. (A) Subjects wereinstructed to view either images ofobjects (houses, faces, and chairs) (left)or imagine the objects in the absence ofthe stimulus (right). (B) (Left) Bilateralregions of ventral temporal cortex arespecifically activated during perceptionof houses (yellow), faces (red), andchairs (blue). (Right) When subjectsrecall these objects, the same regionspreferentially activated during the per-ception of each object class are reacti-vated. (After Ishai et al., 2000).

ChairsFacesHouses

ImageryPerception

(B)

(A)

Purves30 5/14/04 11:04 AM Page 748

ability to lay down new declarative memories. Evidence from such double-dissociations endorses the idea that independent brain systems govern theformation and storage of declarative and nondeclarative memories.

A brain system that appears to be especially important for complex motorlearning involves the signaling loops that connect the basal ganglia and pre-frontal cortex (see Chapter 17). Damage to either structure profoundly inter-feres with the ability to learn new motor skills. Thus, patients with Hunting-ton’s disease, which causes atrophy of the caudate and putamen (see Figure17.9B), perform poorly on motor skill learning tests such as manually track-ing a spot of light, tracing curves using a mirror, or reproducing sequencesof finger movements. Because the loss of dopaminergic neurons in the sub-stantia nigra interferes with normal signaling in the basal ganglia (see Figure17.9A), patients with Parkinson’s disease show similar deficits in motor skilllearning, as do patients with prefrontal lesions caused by tumors or strokes.Neuroimaging studies have largely corroborated these findings, revealingactivation of the basal ganglia and prefrontal cortex in normal subjects per-forming these same skill-learning tests. Activation of the basal ganglia andprefrontal cortex has also been observed in animals carrying out rudimen-tary motor learning and sequencing tasks.

The dissociation of memory systems supporting declarative and nonde-clarative memory suggests the scheme for long-term information storage dia-grammed in Figure 30.10. The generality of the diagram only emphasizes therudimentary state of present thinking about exactly how and where long-term memories are stored. A reasonable guess is that each complex memoryis instantiated in an extensive network of neurons whose activity depends onsynaptic weightings that have been molded and modified by experience.

Memory and Aging

Although it is all too obvious that our outward appearance changes withage, we tend to imagine that the brain is much more resistant to the ravagesof time. Unfortunately, the evidence suggests that this optimistic view is notjustified. From early adulthood onward, the average weight of the normalhuman brain, as determined at autopsy, steadily decreases (Figure 30.11). Inelderly individuals, this effect can also be observed with noninvasive imag-ing as a slight but nonetheless significant shrinkage of the brain. Counts ofsynapses in the cerebral cortex generally decrease in old age (although thenumber of neurons probably does not change very much), suggesting that itis mainly the connections between neurons (i.e., neuropil) that are lost as

Memor y 749

Acquisition and storageof declarative information

Long-term storage (a variety of cortical sites: Wernicke’s

area for the meanings of words, temporal cortex for the memories

of objects and faces, etc.)

Short-term memory storage (hippocampus and related structures)

Acquisition and storageof nondeclarative information

Long-term storage (cerebellum, basal ganglia, premotor

cortex, and other sites related to motor behavior)

Short-term memory storage (sites unknown but presumably

widespread)

Figure 30.10 Summary diagram of theacquisition and storage of declarativeversus nondeclarative information.

Purves30 5/14/04 11:04 AM Page 749

750 Chapter Thir ty

Box DAlzheimer’s DiseaseDementia is a syndrome characterizedby failure of recent memory and otherintellectual functions that is usuallyinsidious in onset but steadily pro-gresses. Alzheimer’s disease (AD) is themost common dementia, accounting for60–80% of cases in the elderly. It afflicts5–10% of the population over the age of65, and as much as 45% of the popula-tion over 85. The earliest sign is typicallyan impairment of recent memory func-tion and attention, followed by failure oflanguage skills, visual–spatial orienta-tion, abstract thinking, and judgment.Inevitably, alterations of personalityaccompany these defects.

The tentative diagnosis of Alz-heimer’s disease is based on these char-acteristic clinical features, and can onlybe confirmed by the distinctive cellularpathology evident on postmortem exam-ination of the brain. The histopathologyconsists of three principal features (illus-trated in the figure): (1) collections ofintraneuronal cytoskeletal filamentscalled neurofibrillary tangles; (2) extracel-lular deposits of an abnormal protein ina matrix called amyloid in so-called senileplaques; and (3) a diffuse loss of neurons.These changes are most apparent in neo-cortex, limbic structures (hippocampus,amygdala, and their associated cortices),and selected brainstem nuclei (especiallythe basal forebrain nuclei).

Although the vast majority of ADcases arise sporadically, the disorder isinherited in an autosomal dominant pat-tern in a small fraction (less than 1%) ofpatients. Identification of the mutantgene in a few families with an early-onset autosomal dominant form of thedisease has provided considerableinsight into the kinds of processes thatgo awry in Alzheimer’s.

Investigators suspected that themutant gene responsible for familial AD

might reside on chromosome 21, primar-ily because similar clinical and neu-ropathologic features often occur in indi-viduals with Down’s syndrome (asyndrome typically caused by an extracopy of chromosome 21), but with amuch earlier onset (at about age 30 inmost cases). The prominence of amyloiddeposits in AD further suggested that amutation of a gene encoding amyloidprecursor protein is somehow involved.The gene for amyloid precursor protein(APP) was cloned by D. Goldgaber andcolleagues, and found to reside on chro-mosome 21. This discovery eventuallyled to the identification of mutations ofthe APP gene in almost 20 families with

the early-onset autosomal dominantform of AD. It should be noted, however,that only a few of the early-onset fami-lies, and none of the late-onset families,exhibited these particular mutations. Themutant genes underlying two additionalautosomal dominant forms of AD havebeen subsequently identified (presenilin 1and presenilin 2). Thus, mutation of anyone of several genes appears to be suffi-cient to cause a heritable form of AD.

The most common form of Alzhei-mer’s occurs late in life, and althoughthe relatives of affected individuals areat a greater risk, the disease is clearlynot inherited in any simple sense. Thecentral role of APP in the families with

(A) (B)Neurofibrillary tangle

Amyloid plaque

(A) Histological section of the cerebral cortex from a patient with Alzheimer’s disease, show-ing characteristic amyloid plaques and neurofibrillary tangles. (B) Distribution of pathologicchanges (including plaques, tangles, neuronal loss, and gray matter shrinkage) in Alzheimer’sdisease. Dot density indicates severity of pathology. (A from Roses, 1995, courtesy of Gary W.Van Hoesen; B after Blumenfeld, 2002, based on Brun and Englund, 1981.)

Purves30 5/14/04 11:04 AM Page 750

Memor y 751

the early-onset form of the diseasenonetheless suggested that APP mightbe linked to the chain of events culmi-nating in the “spontaneous” forms ofAlzheimer’s disease. In particular, bio-chemists Warren Strittmatter and GuySalvesen theorized that pathologic depo-sition of proteins complexed with aderivative of APP might be responsible.To test this idea, they immobilized arecombinant form of the APP derivativeon nitrocellulose paper and searched forproteins in the cerebrospinal fluid ofpatients with Alzheimer’s disease thatbound with high affinity. One of the pro-teins they detected was apolipoprotein E(ApoE), a molecule that normally chap-erones cholesterol through the blood-stream.

This discovery was especially pro-vocative in light of a discovery made byMargaret Pericak-Vance, Allen Roses, andtheir colleagues at Duke University Med-ical Center, who found that affectedmembers of some families with the late-onset form of the inherited disease ex-hibited an association with genetic mark-ers on chromosome 19. This finding wasof particular interest because a geneencoding an isoform of apolipoprotein E(the e4 allele) is located in the sameregion of chromosome 19 implicated bythe family studies. As a result, they beganto explore the association of the differentalleles of apolipoprotein E with affectedmembers in families with a late-onset butinherited form of Alzheimer’s disease.

There are three major alleles ofapolipoprotein E, e2, e3, and e4. The fre-quency of allele e3 in the general popula-tion is 0.78, and the frequency of allele e4is 0.14. The frequency of the ε4 allele inlate-onset familial AD patients, however,is 0.52—almost 4 times higher than thegeneral population. Thus, the inheritanceof the e4 allele is a risk factor for late-onset AD. In fact, people homozygousfor e4 are about 8 times more likely todevelop AD compared to individuals

homozygous for ε3. Among individualsin late-onset Alzheimer’s families withno copies of e4, only 20% develop AD byage 75 compared to 90% of individualswith two copies of e4. An increased asso-ciation of the e4 allele has also beenshown in the sporadic form of AD, anespecially important discovery becausethis category constitutes by far the mostcommon form of the disease.

It is not known whether the ε4 alleleof ApoE itself is responsible for theincreased risk, or whether it is linked toanother gene on chromosome 19 that isthe real culprit. The fact that ApoE bindsavidly to amyloid plaques in AD brainsfavors the idea that the e4 allele of ApoEitself is the problem. However, in con-trast to the mutations of APP or presenilin1 and presenilin 2 that cause familialforms of AD, inheriting the e4 form ofApoE is not sufficient to cause AD;rather, inheriting this gene simplyincreases the risk of developing AD.Moreover, some of the individuals withearly-onset forms of familial AD do nothave the e4 allele. Thus, a variety ofrelated molecular anomalies appear tounderlie AD.

A possible common denominator ofAD at the cellular level is the “amyloidcascade” hypothesis. A prominent con-stituent of the amyloid plaques is anabnormal cleavage product of APP calledamyloid-β peptide (or β-A4). The cascadehypothesis proposes that accumulationof β-A4 is critical to the pathogenesis ofAD. Others, however, argue that extra-cellular deposition of β-A4 may not be akey event in the pathogenesis of ADbecause the density of the β-A4 plaquescorrelates only poorly with severity ofthe dementia (the degree of dementiabeing much better correlated with thedensity of neurofibrillary tangles). More-over, a transgenic mouse model of ADbased on a presenilin 1 mutation exhibitsneurodegeneration without amyloidplaque formation.

Clearly, AD has a complex pathologyand probably reflects a variety of relatedmolecular and cellular abnormalities. Itis unlikely that this important problemwill be understood without a great dealmore research, much hyperbole in thelay press notwithstanding.

ReferencesCHUI, D. H., H. AND 12 OTHERS (1999) Trans-genic mice with Alzheimer presenilin 1 muta-tions show accelerated neurodegenerationwithout amyloid plaque formation. NatureMed. 5: 560–564.CITRON, M., T. OLTERSDORF, C. HAASS, L.MCCONLOGUE, A. Y. HUNG, P. SEUBERT, C.VIGO-PELFREY, I. LIEBERBURG AND D. J. SELKOE

(1992) Mutation of the β-amyloid precursorprotein in familial Alzheimer’s diseaseincreases β-protein production. Nature 360:672–674.CORDER, E. H., A. M. SAUNDERS, W. J. STRITT-MATTER, D. E. SCHMECHEL, P. C. GASKELL, G.W. SMALL, A. D. ROSES, J. L. HAINES AND M. A.PERICAK-VANCE (1993) Gene dose of apolipo-protein E type 4 allele and the risk of Alz-heimer’s disease in late-onset families. Sci-ence 261: 921–923.GOLDGABER, D., M. I. LERMAN, O. W.MCBRIDE, U. SAFFIOTTI AND D. C. GAJDUSEK

(1987) Characterization and chromosomallocalization of a cDNA encoding brain amy-loid of Alzheimer’s disease. Science 235:877–880.LI, T. AND 17 OTHERS. (2000) Photoactivated g-secretase inhibitors directed to the active sitecovalently label presenilin 1. Nature 405:689–694.MURRELL, J., M. FARLOW, B. GHETTI AND M. D.BENSON (1991) A mutation in the amyloidprecursor protein associated with hereditaryAlzheimer’s disease. Science 254: 97–99.ROGAEV, E. I. AND 20 OTHERS. (1995) FamilialAlzheimer’s disease in kindreds with mis-sense mutations in a gene on chromosome 1related to the Alzheimer’s disease type 3gene. Nature 376: 775–778.SHERRINGTON, R. AND 33 OTHERS. (1995)Cloning of a gene bearing missense muta-tions in early-onset familial Alzheimer’s dis-ease. Nature 375: 754–760.

Purves30 5/14/04 11:04 AM Page 751

752 Chapter Thir ty

Figure 30.11 Brain size as a functionof age. The human brain reaches itsmaximum size (measured by weight inthis case) in early adult life anddecreases progressively thereafter. Thisdecrease evidently represents the grad-ual loss of neural circuitry in the agingbrain, which presumably underlies theprogressively diminished memory func-tion in older individuals. (After Deka-ban and Sadowsky, 1978.)

humans grow old (consistent with the idea that the networks of connectionsthat represent memories—i.e., the engrams—gradually deteriorate).

These several observations accord with the difficulty older people have inmaking associations (e.g., remembering names or the details of recent experi-ences) and with declining scores on tests of memory as a function of age. Thenormal loss of some memory function with age means that there is a largegray zone between individuals undergoing normal aging and patients suffer-ing from age-related dementias such as Alzheimer’s disease (see Box D).

Just as regular exercise slows the deterioration of the neuromuscular sys-tem with age, age-related neurodegeneration and associated cognitivedecline may be slowed in elderly individuals who make a special effort tocontinue using the full range of human memory abilities (i.e, both declara-tive and nondeclarative memory tasks). Although cognitive decline with ageis ultimately inevitable, neuroimaging studies suggest that high-performingolder adults may to some degree offset declines in processing efficacythrough compensatory activation of cortical tissue that is less fully used dur-ing remembering in poorly performing older adults (Figure 30.12).

Age (years)

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Bra

in w

eigh

t (kg

)

FemalesMales

0 1 3 5 10 20 30 40 50 60 70 80 90

Old–Low Old–HighYoung

R L R L R L

Figure 30.12 Compensatory activa-tion of memory areas in high-function-ing older adults. During remembering,activity in prefrontal cortex wasrestricted to the right prefrontal cortex(following radiological conventions, thebrain images are left-right reversed) inboth young participants and elderlysubjects with poor recall. In contrast,elderly subjects with relatively goodmemory showed activation in bothright and left prefrontal cortex. (AfterCabeza et al., 2002).

Purves30 5/14/04 11:04 AM Page 752

SummaryHuman memory entails a number of biological strategies and anatomicalsubstrates. Primary among these are a system for memories that can beexpressed by means of language and can be made available to the consciousmind (declarative memory), and a separate system that concerns skills andassociations that are essentially prelinguistic, operating at a largely uncon-scious level (nondeclarative or procedural memory). Based on evidence fromamnesic patients and knowledge about normal patterns of neural connec-tions in the human brain, the hippocampus and associated midline dien-cephalic and medial temporal lobe structures are critically important in lay-ing down new declarative memories, although not in storing them (a processthat occurs primarily in the association cortices). In contrast, nondeclarativememories for motor and other unconscious skills depends on the integrity ofthe premotor cortex, basal ganglia, and cerebellum, and is not affected bylesions that impair the declarative memory system. The common denomina-tor of these categories of stored information is generally thought to be alter-ations in the strength and number of the synaptic connections in the cerebralcortices that mediate associations between stimuli and the behavioral re-sponses to them.

Memor y 753

Additional ReadingReviewsBUCKNER, R. L. (2000) Neuroimaging of mem-ory. In The New Cognitive Neurosciences, M.Gazzaniga (ed.). Cambridge, MA: MIT Press,pp. 817–840.BUCKNER, R. L. (2002) The cognitive neuro-science of remembering. Nat. Rev. Neurosci.2: 624–634.CABEZA, R. (2001) Functional neuroimaging ofcognitive aging. In Handbook of Functional Neu-roimaging of Cognition, R. Cabeza and A. King-stone (eds.). Cambridge, MA: MIT Press, pp.331–377.ERICKSON, C. A., B. JAGADEESH AND R. DESI-MONE (2000) Learning and memory in theinferior temporal cortex of the macaque. InThe New Cognitive Neurosciences, M. Gazza-niga (ed.). Cambridge, MA: MIT Press, pp.743–752.MISHKIN, M. AND T. APPENZELLER (1987) Theanatomy of memory. Sci. Am. 256(6): 80–89.PETRI, H. AND M. MISHKIN (1994) Behaviorism,cognitivism, and the neuropsychology ofmemory. Am. Sci. 82: 30–37.SCHACTER, D. L. AND R. L. BUCKNER (1998)Priming and the brain. Neuron 20: 185–195.SQUIRE, L. R. AND B. J. KNOWLTON (2000) Themedial temporal lobe, the hippocampus, andthe memory systems of the brain. In The NewCognitive Neurosciences, M. Gazzaniga (ed.).Cambridge, MA: MIT Press, pp. 765–779.SQUIRE, L. R. (1992) Memory and hippocam-pus: A synthesis from findings with rats, mon-keys, and humans. Psych. Rev. 99: 195–230.THOMPSON, R. F. (1986) The neurobiology oflearning and memory. Science 223: 941–947.

ZACKS, R. T., L. HASHER AND K. Z. H. LI (1999)Human memory. In The Handbook of Aging andCognition. F. I. M. Craik and T. A. Salthouse(eds.). Mahwah, New Jersey: Lawrence Erl-baum Associates, pp. 293–357.ZOLA-MORGAN, S. M. AND L. R. SQUIRE (1993)Neuroanatomy of memory. Annu. Rev. Neu-rosci. 16: 547–563.

Important Original PapersCABEZA, R., N. D. ANDERSON, J. K. LOCANTOREAND A. R. MCINTOSH (2002) Aging gracefully:Compensatory brain activity in high-perform-ing older adults. NeuroImage 17: 1394–1402.GOBET, F. AND H. A. SIMON (1998) Expert chessmemory: Revisiting the chunking hypothesis.Memory 6: 225–255.ISHAI, A., L. G. UNGERLEIDER AND J. V. HAXBY(2000) Distributed neural systems for the gen-eration of visual images. Neuron 28: 979–990.SCOVILLE, W. B. AND B. MILNER (1957) Loss ofrecent memory after bilateral hippocampallesions. J. Neurol. Neurosurg. Psychiat. 20: 11–21. SQUIRE, L. R. (1989) On the course of forget-ting in very long-term memory. J. Exp. Psy-chol. 15: 241–245.ZOLA-MORGAN, S. M. AND L. R. SQUIRE (1990)The primate hippocampal formation: Evi-dence for a time-limited role in memory stor-age. Science 250: 288–290.

BooksBADDELEY, A. (1982) Your Memory: A User’sGuide. New York: Macmillan.CRAIK, F. I. M. AND T. A. SALTHOUSE (1999) TheHandbook of Aging and Cognition. Mahwah,New Jersey: Lawrence Erlbaum Associates.DUKAS, R. (1998) Cognitive Ecology: The Evolu-tionary Ecology of Information Processing and

Decision Making. Chicago: University of Chi-cago Press.GAZZANIGA, M. S. (2000) The New CognitiveNeurosciences, 2nd Ed. Cambridge MA: MITPress.GAZZANIGA, M. S., R. B. IVRY AND G. R. MAN-GUN (1998) Cognitive Neuroscience: The Biologyof the Mind. New York: W. W. Norton & Com-pany.LURIA, A. R. (1987) The Mind of a Mnemonist.Translated by Lynn Solotaroff. Cambridge,MA: Harvard University Press.NEISSER, U. (1982) Memory Observed: Remem-bering in Natural Contexts. San Francisco: W.H. Freeman. PENFIELD, W. AND L. ROBERTS (1959) Speech andBrain Mechanisms. Princeton, NJ: PrincetonUniversity Press. SAPER, C. B. AND F. PLUM (1985) Handbook ofClinical Neurology, Vol. 1(45): Clinical Neuro-psychology, P. J. Vinken, G. S. Bruyn and H. L.Klawans (eds.). New York: Elsevier, pp. 107–128.SCHACTER, D. L. (1997) Searching for Memory:The Brain, the Mind, and the Past. New York:Basic Books.SCHACTER, D. L. (2001) The Seven Sins of Mem-ory: How the Mind Forgets and Remembers.Houghton Mifflin Co.SMITH, S. B. (1983) The Great Mental Calculators:The Psychology, Methods, and Lives of Calculat-ing Prodigies, Past and Present. New York:Columbia University Press.SQUIRE, L. R. (1987) Memory and Brain. NewYork: Oxford University Press, pp. 202–223.ZECHMEISTER, E. B. AND S. E. NYBERG (1982)Human Memory: An Introduction to Researchand Theory. Monterey, CA: Brooks/Cole Pub-lishing.

Purves30 5/14/04 11:04 AM Page 753