Neuropsychology Review, Vol. 3, No.

2, Z992

Memory Deficits in Alzheimer's Patients: A

Comprehensive Review
Giovanni A. C a r l e s i m o 1,2 and Marlene Oscar-Berman 3

Despite considerable experimental work on Alzheimer's disease (AD), the un-

derlying cognitive mechanisms as well as the precise localization of
neuropathological changes critical for memory loss remains undefined. A re-
view of the neuropsychological literature on long-telm memory deficits in AD
patients suggests that AD patients display (a) a pervasive deficit of explicit
memory, (b) a partial deficiency of implicit memory for verbal and visuoper-
ceptual material (as measured by repetition priming procedures), and (c) a
substantial sparing of implic# memory for visuomotor skills. The explic# mem-
ory loss is likely a result of encoding as well as consolidation difficulties. A
faulty lexicaLsemantic knowledge structure appears responsible for deficient
repetition priming effects. Since neuropathological changes diffusely affect the
brain of A D patients, establishing a clear relationship between localization of
cerebral lesions and memory deficits is particularly difficult. Nevertheless, data
suggest that extensive involvement of the hippocampal-amygdala complex plays
a major rote in explicit memory loss. Damage to associative cortical areas likely
is involved in repetition priming deficits. The relative integrity of primary motor
and sensory col¢ical areas and of the basal ganglia likely subsume, by contrast,
the normal learning of visuomotor skills.
KEY WORDS: Alzheimer; memory; cognition; amnesia.

lIstituto di Neurologia, II Universit/t di Roma, Italy.

2To whom correspondence should be addressed at Clinica Neurologica, lI UniversitS. di Roma,
Ospedale S. Eugenio, P.le dell'Umanesimo, 10, 00144 Roma, Italy.
3Psychology Service, U.S. Department of Veterans Affairs Medical Center, Boston, and
Department of Neurology and Division of Psychiatry, Boston University School of Medicine,
Boston, Massachusetts.

119
1040-7308/92/0600-0119506.50/0 © 1992Plenum PublishingCorporation
120 Carlesimo and Oscar-Berman

INTRODUCTION

Alzheimer's disease (AD) is the most common cause of dementia,

being responsible for more than 50% of all dementing syndromes
(Katzman, 1986; Tomlinson et al., 1970). AD is characterized neuropsy-
chologically by a progressive deterioration of multiple aspects of cognition,
personality, and emotional behavior (McKhan et al., 1984). The disease
typically begins after age 65, and is characterized by diffuse neuronal de-
generation of cortical as well as subcortical areas of the brain (Terry and
Katzman, 1984). A disturbance of memory is frequently the first symptom
of complaint by patients and relatives, and generally is regarded as a nec-
essary (but not a sufficient) clinical feature for the diagnosis (Katzman and
Karasu, 1975; McKhan et al., 1984).
The primary aim of this paper is to review the neuropsychological
literature on memory deficits in AD patients, in the attempt to provide a
theoretical account of the disorder. For this purpose, a number of meth-
odological and other fundamental problems must be faced. The first one
deals with diffuse cognitive deficits that may contribute to an overall decline
in memory. For example, linguistic, attentional, and visuoperceptual defi-
cits, as well as impairments in abstract reasoning, can prevent the normal
processing of information, thereby interfering with efficient encoding and
retrieval processes. A complete theoretical account of memory disorders
in AD must include the possible influence exerted by these variables on
objective measures of memory.
Another problem concerns the choice of a theoretical model of
memory in which the experimental evidence can be coherently concep-
tualized. Memory is, at present, no longer regarded as a unitary phe-
nomenon, and a number of subtypes of memory have been proposed to
account for functional dissociations observed in normal as well as in am-
nesic subjects. One model with considerable pragmatic and theoretical
utility makes a distinction between explicit and implicit forms of memory
(Squire, 1987). According to many authors (e.g., Richardson-Klavehn and
Bjork, 1988; Roediger, 1990; Tulving and Schacter, 1990), explicit meas-
ures of memory, such as overt recall and recognition, typically require
the conscious recollection of information; by contrast, implicit measures
of memory, such as priming, assess the influence exerted by past expe-
riences on ongoing behavior without requiring conscious recollection of
the information. In the present paper the explicit-implicit dichotomy is
accepted, and experimental evidence is reviewed concerning the perform-
Memory Deficits in Aizheimer's Disease 121

ance of AD patients on tasks designed to measure these aspects of mem-

ory dysfunction.
A final methodological problem in reviewing the neuropsychological
literature on AD deals with the diagnostic inclusionary/exclusionary criteria
adopted in different studies for subject selection. Because definitive bio-
logical markers of the disease are lacking, the diagnosis in vivo of AD is
based on an integrated evaluation of clinical, neuroradiotogical, and
neuropsychological data. The use of less than accurate diagnostic criteria
likely results in biased experimental samples, including subjects affected by
other dementing and nondementing medical illnesses. The present review
will be limited to studies in which experimental samples were selected ac-
cording to standard diagnostic criteria (see, for example, American Psychi-
atric Association, 1980; McKhan et aL, 1984).
This article is organized into three main sections. In the first section,
we review a large amount of research concerned with explicit measures of
memory in AD. This section is in turn divided into four subsections, each
one related to different aspects of the memory function: acquisition, en-
coding, storage, and retrieval of the information. In the second main section
of the paper, we review a more sparse literature on implicit measures of
memory in AD. Here, one subsection deals with studies that investigated
the ability of AD patients to learn visuomotor procedures, while a second
subsection is concerned with studies of repetition priming. The third and
final main section of the present review summarizes the evidence, discusses
the principal hypotheses proposed to account for memory disorders in AD,
and attempts to relate memory deficits to anatomical and neurochemical
substrates of AD.

EXPLICIT MEMORY

Procedures used to assess explicit memory generally consist of a study

phase (during which the stimulus material is presented) followed by an im-
mediate or delayed retrieval phase of the same material (by means of free
recall, cued recall, or recognition). Deficits in explicit memory can occur
as a result of dysfunctions affecting one or more of four successive
phases--namely, acquisition, encoding, storage, and retrieval of information
(see Baddeley, I982, for a review). The literature on explicit memory in
AD will be reviewed according to this traditional conceptual framework.
In each of the four subsections, a brief outline of the principal cognitive
theories will be followed by a detailed review of experimental data on AD
122 Carlesimo and Oscar-Berman

patients. An attempt to synthesize the available experimental evidence in

a coherent hypothesis will conclude each subsection.

Deficits in Acquisition

Early experimental work on memory disturbances in dementia gen-

erally assumed that the underlying deficiency was one of short-term mem-
ory (STM; Miller, 1977). A defect in STM would result in limited
acquisition of information, as well as deficient transfer of information from
STM to long-term memory (LTM; Miller, 1977).
Early research on STM and LTM deficits in AD was conducted by
Miller (1971, 1972, 1973, 1977). Miller's work fit within a theoretical frame-
work called the "modal model" of memory (Murdock, 1967). According to
this model, STM plays a major role in the flow of information between the
external environment and an internal LTM store (Atkinson and Shiffrin,
1968, 1971). All information reaching the body's receptors from the external
environment temporarily is retained in STM store (usually with the aid of
rehearsal mechanisms) prior to being transferred to the LTM store. Simi-
larly, STM is responsible for activating retrieval strategies when the recol-
lection of information from LTM store is requested. The main differences
between STM and LTM stores are capacity (limited vs. virtually unlimited),
durability of trace (seconds vs. minutes to years), and level of processing
(phonological vs. semantic). In the modal model, the distinction between
STM and LTM systems was based largely on evidence from the serial po-
sition effect observable in tasks of free recall of word lists (Glanzer and
Cunitz, 1966). That is, when the number of items correctly recalled in these
tasks is plotted as a function of their positions in the list, a U-shaped curve
is obtained with the earliest and final items of the list best recalled. Ac-
cording to the modal model, the elevation of the early part of the curve
(primacy effect) occurs because the first items of the list already have been
stored in LTM. By contrast, the elevation of the final part of the curve
(recency effect) occurs because the last items of the list are still being re-
tained in STM store. The magnitudes of primacy and recency effects in a
free recall task traditionally have been regarded as reliable measures of the
strength of LTM and STM, respectively. Another frequently used measure
of STM store capacity is the number of digits or words that are correctly
recalled in an immediate memory task (word span), while the Brown-Pe-
terson technique (Brown, 1958; Peterson and Peterson, 1959) is commonly
considered an indicator of the rate of forgetting from STM store.
In his early research, Miller (1971) compared the performance of AD
patients and age-matched control subjects on a task of free recall of word
Memory Deficits in Alzheimer's Disease i23

lists. In accordance with the hypothesis that deficient LTM in AD is a con-

sequence of insufficient acquisition of information by STM, results of the
first experiment showed that both primacy and recency effects in AD pa-
tients were reduced as compared to control subjects. In the second experi-
ment, Miller (1971) tested the hypothesis that an additional difficulty in
storing information in LTM could result from deficient transfer of infor-
mation from STM to LTM stores. Glanzer and Cunitz (1966) had observed
with normal subjects that when the rate of presentation of word lists was
slowed down, the number of words recalled in the early and middle parts
of the list was enhanced, while the recall of last items was not affected.
The authors' interpretation of this phenomenon was that the increased time
period during slowed presentation permitted a larger number of items to
be transferred into LTM store. Miller (1971, exp. 2) administered to ad-
ditional samples of AD and control subjects lists of words at two different
rates of presentation (1.5 sec per word and 3 sec per word). Results, even
in this case, were in accord with expectations. In fact, while normal subjects
showed the expected increase of recall of early and middle items of the
list as a function of the slower rate of words presented, AD patients did
not benefit from the increase in available time. For AD patients, in fact,
the number of words recalled did not differ as a function of the two rates
of presentation. In a subsequent study, Miller (1973) confirmed these re-
sults, demonstrating that AD subjects suffer both from a reduced capacity
of STM sore (deficient word span) and an impairment in storing informa-
tion in LTM (deficient word supraspan learning). In another study, Miller
(1972) evaluated the hypothesis that reduced STM capacity in AD could
result from ineffective phonological encoding mechanisms. This assumption
was supported by the finding that, in a task requiring the immediate recall
of word lists, the phonological similarity among words exerted a less detri-
mental effect on recall by AD patients than by control subjects. On the
basis of this pool of data, Miller (1977) advanced a two-component hy-
pothesis of memory disorders in AD patients: He proposed that both the
acquisition of information by STM and the establishment of new material
in the LTM store are seriously compromised in these patients.
Miller's two-component hypothesis has been successively challenged
on experimental and theoretical grounds. From the experimental point of
view, Miller's results have not consistently been replicated. As for the re-
cency effect, Wilson, Bacon, Fox, and Kaszniak (1983) confirmed Miller's
results, pointing out deficient performance by AD patients on both the re-
cent and the early portion of the serial position curve. Further, in this study
primacy and recency scores were significantly related in the AD but not
in the control sample, thus giving support to the hypothesis that deficient
LTM recall in AD patients can, at least in part, be related to dysfunction
124 Carlesimo and Oscar-Berman

of STM. Other authors, however, failed to confirm a deficient recency ef-

fect in subjects with early signs of AD. Gibson (1981) and Spinnler et aL
(1988), for example, noted comparable recency effects in mildly demented
AD patients and control subjects. Similarly, a deficient verbal span in AD
subjects has been reported by some authors (Kaszniak and Garron, 1979;
Morris, 1984; Spinnler et al., 1988), but not by others (Brandt et aL, 1988;
Martin et al., 1985; Orsini et al., 1988; Weingartner et al., 1983). Finally,
Miller's claim of a deficient phonological encoding in STM by AD patients
has been questioned by Morris (1984). Morris argued that the lack of dif-
ference between immediate recall of phonologically similar and dissimilar
words in AD patients (Miller, 1972) could actually have been biased by a
floor effect. In fact, when a similar experiment was repeated using letters
instead of words, and the floor effect was avoided (by administering to
every subject strings of phonologically similar or dissimilar letters of the
same length of his own span), similar decremental effects produced by the
phonological similarity were observed in AD and age-matched control sub-
jects (Morris, 1984).
Discrepancies in results obtained by various authors likely reflect,
among other things, dissimilar levels of severity of cognitive impairment in
the different AD samples. Methodological differences in testing procedures
could represent a further source of variability. For example, in the four
studies of recency cited above (Gibson, 1981; Miller, 1971; Spinnler et aL,
1988; Wilson, Bacon, Fox, and Kaszniak, 1983), four different methods
were employed to analyze the same effect. Bearing this in mind, it is pos-
sible that differences between experimental samples and testing procedures
could account for the heterogeneous results. However, even though con-
trasting experimental results could be reconciled by homogeneity of sam-
ples and tasks, a deeper difficulty in interpreting Miller's results lies in his
underlying theoretical assumptions. Data from experimental literature on
normal and brain-damaged subjects make the modal model of memory no
longer tenable as originally interpreted by Atkinson and Shiffrin (1968,
1971). There are two impressive pieces of evidence. First, patients with fo-
cal brain damage in the left temporal lobe have displayed severe impair-
ments in verbal STM, but normal learning ability and retrieval from LTM
(Basso et aL, 1982; Shallice and Warrington, 1970). This finding clearly is
at variance with the hypothesis that STM plays a central role in storing
and retrieving information in LTM. Second, the concept of a unitary, mul-
timodal, capacity-limited STM store is contradicted by the fact that (a) left-
temporal patients could demonstrate an impaired verbal STM in the face
of a normal visuospatial STM; and (b) normal subjects whose STM load
was filled with verbal information were capable of near normal levels of
verbal learning, reasoning, and comprehension (Baddeley and Hitch, 1974).
Memory Deficits in Alzheimer's Disease 125

It should be noted also that problems with the modal model have been
paralleled by increasing skepticism about the reliability of traditional meas-
ures of STM. For example, experimental work on normal subjects has
claimed that measures such as verbal span, recency effect, and the Brown-
Peterson procedure all have large LTM components (Baddeley and Scott,
1971; Craik, 1970; Keppel and Underwood, 1962).
The Working Memory (WM) model (Baddeley, 1976, 1986; Baddeley
and Hitch, 1974) represents a current and influential attempt to redefine
the structure and role of STM, and to discover reliable experimental pro-
cedures for assessing the plausibility of underlying theoretical assumptions.
According to this model, WM is subserved not by a unitary store but by
the cooperation of two major systems. The first is a Central Executive Sys-
tem, a limited capacity central processor able to temporarily store and proc-
ess multimodal information. The second major system actually consists of
a number of peripheral Slave Systems, or limited capacity systems, which
temporarily store and rehearse information belonging to a single modality
when the flow of data fills the capacity of the Central Executive. The best
studied Slave System is the Articulatory Loop, which subserves a temporary
store of verbal material. The loop is assumed to be formed by a pas~sive
phonological input store (sensitive to the phonological similarity of items
to be retained) and by an active articulatory rehearsal mechanism (sensitive
to concurrent articulation and to the length of words to be retained) that
continually refreshes the memory trace in the phonological store. The so-
conceived WM plays a main role in maintaining and processing information
during tasks of comprehension, reasoning, and learning. Common experi-
mental procedures in the domain of WM consist of the assessment of
changes in accuracy or latency on a primary task, at the same time that
the Central Executive or the Slave Systems are occupied by a concurrent
secondary task. It has been demonstrated that tasks presumed to exhaust
the processing resources of the Central Executive (e.g., rehearsing a six-
digit load), interfere with sentence comprehension, reasoning, and storage
of information in LTM; on the other hand, tasks that fill the capacity of
the Articulatory Loop (e.g., articulatory suppression) exert only a marginal
influence on these abilities (Baddeley and Hitch, 1974). Reformulating the
WM model in these terms would explain findings in the above-cited reports
that there are patients who suffer from a selective disorder of verbal STM.
According to this model, it is probably the Articulatory Loop that is mal-
functioning in these patients, while their intact Central Executive and visu-
ospatial Slave Systems are capable of normal performances on tasks of
verbal comprehension, reasoning, and learning (VaIlar and Baddetey, 1984)
as well as visuospatial STM.
126 Carlesimo and Oscar-Berman

Studies that have applied the conceptual model of WM to memory

disturbances of AD patients are relatively recent. In a series of consecutive
studies Morris (1984, 1986, 1987a, 1987b) was able to demonstrate that
mildly demented AD subjects show a substantially unimpaired Articulatory
Loop. His AD subjects, in fact, despite a moderate reduction in span, pre-
sented normal phonological similarity effects and word length effects in a
serial free recall task (Morris, 1984), as well as the normal contribution of
articulatory rehearsal to the memory span (Morris, 1987a, 1987b). Two
studies that used a dual-task procedure obtained, instead, results supporting
a deficient Central Executive System in AD patients. Baddeley et al. (1986)
administered a pursuit tracking task to samples of AD, age-matched nor-
mal, and young normal subjects. A comparable basic level of performance
on this task was obtained for all subjects simply by varying the speed of
the moving stimulus. In three different experimental conditions, the sec-
ondary tasks were, respectively, concurrent articulation, rapid responding
to tones, and a standard digit span procedure. While all three groups
showed a nonsignificant decrement in accuracy on the pursuit tracking task
during the concurrent articulation condition, the decrements manifested by
AD patients during the other two conditions were the only significant find-
ings. The authors interpreted the deficient capacity of AD patients in si-
multaneously performing two concurrent tasks as a demonstration of
reduced processing resources by their Central Executive System. Similar
conclusions were reached by Morris (1986), who assessed changes in accu-
racy of AD patients and age-matched controls in performing the Brown-
Peterson task when the complexity of the distractory task was varied. One
of five different tasks filled the short delay between presentation and recall
of three verbal items: no distractory task, simple articulation, digit reversal,
digit addition, and finger tapping. With the exception of the no distractory
task condition, in all t h e o t h e r conditions AD subjects performed the mem-
ory task at a consistently lower level than age-matched controls. Even in
this case, it was argued that a deficient Central Executive System was re-
sponsible for AD patients' inefficiency in simultaneously maintaining the
memory load and performing the secondary task.
As for the relationship between WM and LTM disorders in AD, an
intriguing hypothesis is that the disability in learning and/or retrieving new
information in these patients can be related, at least in part, to a reduced
overall capacity of the Central Executive System in processing incoming
and outgoing information. This assumption, however, is based only on in-
direct evidence (reduced processing resources in AD patients, as well as a
decremental influence on LTM exerted, in normal subjects, by concurrent
tasks that fill the Central Executive to capacity). The experimental dem-
onstration that learning and/or retrieval abilities are more susceptible to
Memory Deficits in AIzheimer's Disease 127

attention-demanding concurrent tasks in AD than in control subjects would

represent significant support for this hypothesis.

Encoding Deficits

A second influential hypothesis regarding memory disorders in AD

posits that demented patients fail to remember because of a deficient en-
coding of incoming information. This hypothesis reflects a general tendency
of memory theorists from the early 1970s to shift the focus of their attention
from structural models of memory (e.g., the modal model) to models em-
phasizing the role of information-processing activities in memory operations
(Cermak, 1972; Craik and Lockhart, 1972). To briefly explain: these theo-
ries claim that normal subjects, when requested to intentionally learn some
kind of material, engage in organizational and elaborative processes in or-
der to enhance the quality of encoding and, as a consequence, the effec-
tiveness of using cues during retrieval. From an experimental point of view,
it is well known that different kinds of manipulations during the study phase
can modify the level of subsequent retrieval, presumably by influencing the
quality of encoding processes (e.g., Craik and Tulving, 1975, and Jacoby,
1982). Some of these manipulations deal with the qualitative features of
to-be-remembered material. It is known, in fact, that different levels of re-
trieval can be obtained, maintaining the same experimental procedure, sim-
ply by modifying specific characteristics of the stimuli (e.g., high- vs.
low-frequency words, concrete vs. abstract words, lists formed by unrelated
vs. related words, etc.; Deese, 1959; Jacoby and Dallas, 1981; Paivio, 1969).
A more direct demonstration of the influence exerted by information-proc-
essing activities on memory operations lies in experiments showing in-
creased levels of retrieval for the same material following manipulations
in study procedures (e.g., slowing down the rate of presentation, leading
subjects to process semantic rather than phonological aspects of the words,
etc.; Glanzer and Cunitz, 1966; Craik and Tulving, 1975). A list of experi-
mental manipulations of the study phase, known to be effective in modi-
fying levels of retrieval in normal subjects, is presented in Table I.
Several attempts have been made to provide a conceptual framework
to account for this vast amount of experimental evidence. Craik and Lock-
hart's "levels of processing" theory (1972) is undoubtedly the most popular
and frequently cited one. It assumes that different aspects of the same in-
formation can be encoded, but deeper levels of processing (i.e., dealing
with the semantic meaning of the stimuli) are more effective than shallow
levels (i.e., concerning the phonological or the physical aspects of the
words) in producing memory traces that can be easily accessed during re-
128 Carlesimo and Oscar-Berman

Table h Studies Concerned with the Effects of Experimental Manipulations of the Study
Phase on Retrieval Levels of AD Patients
Experimental evidence
Experimental Experimental in AD patients
Manipulations in the evidence in
study phase normal subjects Active Nonactive

Manipulations of material
Semantic relatedness Deese, 1959 Cushman et al., 1988 Weingartner et al.,
of words in the list 1981
Delis et al., 1991
Prior knowledge of Huttsch and Backman and
task-relevant material Dixon, 1983 Herlitz, 1990
Approximation to Miller and Nebes et aL, 1984
ortographic, Setfridge, t950 Nebes et al., 1989
semantic, and
sunctactic rules
Pictures vs. words Paivio, 1971 Rissenberg and
Glanzer, 1986
Concrete vs. abstract Paivio,1969 Rissenber and
words Glanzer, 1987a
Recognition of low- Jacoby and Wilson, Bacon,
frequency vs. high- Dallas, 1981 Kramer et aL, 1983
frequency words
Manipulations of presentation procedures
Levels of processing Craik and Martin et aL, 1985b Corkin, 1982b
Tulving, 1975 Herlitz et aL, 1991b Wilson et aL, 1982b
Butters et aL, 1983
Generation vs. Slamecka and Mitchell et al., 1986
reading of words Graf, 1978 Dick et al., 1989b
Self-performed vs. Cohen, 1981 Dick et aL, 1989a
verbally described Karlsson et aL, 1989
tasks
Encoding specificity Thomson and Granholm and
Tulvin, 1970 Butters, 1988
Rate of presentation Glanzer and Miller, 1971
Cunitz, 1966
Number of repetitions Jacoby and Strauss et at., 1985
Dallas, 1981
aConcreteness effect probably biased by a particular difficulty of AD patients in retrieving
abstract words.
bClassical Levels of Processign experimental paradigm (Craik and Tulvin, 1975).

trieval. O t h e r a u t h o r s stress the role o f c o m m u n i c a t i o n b e t w e e n d i f f e r e n t

r e p r e s e n t a t i o n s (codes) of the s a m e stimulus. F o r example, Paivio (1969,
1971) i n t e r p r e t e d the p i c t u r e effect (pictures m o r e effectively r e m e m b e r e d
t h a n words) a n d the c o n c r e t e n e s s effect ( b e t t e r m e m o r y for c o n c r e t e t h a n
a b s t r a c t words) as the d e m o n s t r a t i o n that a d u a l coding (verbal a n d visual)
Memory Deficits in Alzheimer's Disease 129

increases the effectiveness of retrieval, by providing multiple routes (cues)

for accessing relevant information. An alternative point of view has been
advanced, among others, by Jacoby (1982) who, observing that task diffi-
culty during the study phase (e.g., solving a crossword puzzle vs. simply
reading the word) is positively related to the subsequent retrievability of
the item, proposed to replace the concept of "levels of processing" with
the notion of "distinctiveness." Attention-demanding tasks would produce
more distinctive traces than automatic ones, so facilitating their subsequent
retrieval. A final theory (Tulving, 1983) assumes that memory performances
are function not solely of the quality of encoding operations, but rather of
the interaction between encoding and retrieval processes. In other words,
the strength of a memory trace cannot be specified in absolute terms, since
access to relevant information critically depends on the similarity between
retrieval cues and aspects of the information encoded during the study
phase.
A direct way of testing the hypothesis that impaired memory in AD
patients depends, at least in part, on deficient encoding of incoming infor-
mation consists in examining the effects on retrieval of experimental ma-
nipulations during the study phase. The hypothesis would be supported by
the finding that procedural variables known to be active in improving the
memory of normal subjects are ineffective for AD patients. One experi-
mental design that tests this prediction directly consists of comparing per-
formances of age-matched controls and AD patients on two tasks differing
in study conditions (e.g., learning of lists formed by abstract vs. concrete
words or procedures leading subjects to process shallow vs. deep features
of the words). A significant Group x Test interaction would indicate that
AD patients do not benefit from the experimental manipulation as much
as controls and would support the hypothesis that memory deficits of AD
patients may depend on a reduced ability to process stimuli. By contrast,
a nonsignificant interaction would suggest that AD patients, despite a tow
level of performance, profit in the same way as normal subjects by opera-
tions applied to improve encoding (thereby suggesting that encoding proc-
esses are actually effective in AD patients). According to Shimamura
(1990), two main methodological problems, possibly giving rise to spurious
statistical interactions, emerge from this type of experimental design. The
first problem is that absolute levels of performance are different for normal
and AD patients. As a consequence, a larger deficit by AD patients on
one test compared to some other test (giving rise to a significant Group x
Task interaction) could result simply from different difficulty levels of the
two tests (AD patients being especially impaired on the more difficult one).
A related methodological problem lies in the possible presence of floor or
ceiling effects in the performance of AD and control subjects, respectively,
130 Carlesimo and Oscar-Berman

which could mask or emphasize different patterns of performance by the

groups. One way of dealing with these problems would be to decrease (i.e.,
weaken) the performance of normal subjects (e.g., by testing them after a
longer delay, or by presenting longer lists of stimuli) in order to equate
levels of retrieval with those of AD group. This kind of experimental ma-
nipulation, however, has been used only sporadically (e.g., Herlitz et al.,
1991; Nebes et al., 1984, 1989).
The above-mentioned theoretical as well as methodological concerns
must be kept in mind when reviewing studies of the relationship between
acquisition in the study phase and subsequent levels of retrieval. A sum-
mary of studies that have examined effects of study-phase manipulations
on subsequent retrieval by AD patients is presented in Table I. A number
of studies investigated the ability of AD patients to take advantage of the
semantic context of to-be-remembered material. Weingartner et al. (1981)
and Cushman et al. (1988) asked AD and age-matched controls to recall
lists of semantically related and unrelated words; contrasting results were
obtained in the two studies. In the first study (Weingartner et al. 1981)
controls (not the AD subjects) benefited from the semantic relationship
between words (significant Group x List interaction). In the Cushman et
al. study (1988) the recall levels of both groups were superior for seman-
tically related words. While differences in severity of memory loss between
two different AD samples could account for these contrasting results, it
should be noted that in the Weingartner et al. study (1981) a possible floor
effect in the performance of the AD group may have obscured a real effect.
Recently, Delis et al. (1991) administered a list of 16 words belonging to
four categories (California Verbal Learning Test) to AD, Korsakoff, and
Huntington patients. All three groups of patients were poor in the use of
semantic clustering, thus demonstrating a deficient semantic elaboration of
the to-be-remembered material. In another study, Backman and Herlitz
(1990) evaluated, by means of a facial recognition test, the effect of task-
relevant prior knowledge on memory performances of normal elderly and
AD subjects. The results confirmed previous findings of a positive effect
of prior knowledge on memory performance. The normal elderly and the
AD patients alike identified as familiar more dated that contemporary fa-
mous faces. In a later recognition test, however, only the normals (not the
AD patients) performed better for dated than for contemporary faces, sug-
gesting a reduced ability of the AD group in utilizing the semantic knowl-
edge to enhance encoding operations. Using a different experimental
paradigm, Nebes et al. (1984, 1989) reached opposite conclusions. In the
first study, the authors presented to AD and age-matched control subjects
strings of letters and words with different degrees of approximation to the
orthographic and semantic rules of English language. A comparable level
Memory Deficits in Alzheimer's Disease 131

of percent correct recall in two groups was obtained by varying the length
of the strings presented in the study phase. Results indicated that perform-
ances of both groups were similarly affected by the degree to which the
to-be-remembered material conformed to the rules of the English language.
In a second study (Nebes et al., 1989), the authors confirmed and extended
these results, pointing out that AD patients were as capable as normal sub-
jects of profiting from regular semantic and syntactic sentence structure to
improve subsequent recall. The authors interpreted these data as proof that
AD patients maintain linguistic knowledge (semantic and syntactic), allow-
ing them to organize strings of letters (into word chunks) or words (into
multiword chunks) for efficient encoding.
In other research investigations, the general characteristics of to-be-
remembered stimulus materials were kept constant while manipulations
were performed on the mental processes needed to elaborate the stimuli.
Four studies utilized, with some adaptation, a classical experimental para-
digm from the "levels of processing" framework (Craik and Tulving, 1975).
Corkin (1982) manipulated the level of processing requiring subjects to an-
swer different kinds of questions about 30 words in a list. Some questions
led subjects to process words at a sensory level ("Does a man/woman say
the word?"), others at a phonological level ("Does the word rhyme with
. . . ?"), and others at a semantic level ("Is the word a type of . . . ?"). In
an unannounced multiple-choice recognition test, normal subjects displayed
the expected advantage for semantically processed words, compared to sen-
sorially and phonologically processed words; three subgroups of mildly,
moderately, and severely demented AD patients performed at the same
poor levels regardless of the encoding strategy adopted in the study phase.
Similar results were obtained by Wilson et al. (1982). In that study, AD
and control subjects were required to answer questions about rhyme and
semantic category of 48 words, and retention was measured with a yes-no
recognition test. Somewhat contrasting results were obtained by Herlitz et
al. (1991), who measured the recall of five lists of words, each presented
under a different encoding condition. Age-matched control subjects, and
three groups of mildly, moderately, and severely demented AD patients,
were compared. In one list, only the names of objects were presented (ver-
bal condition); in another list, the names were accompanied by the visual
presentation of the objects (object condition); in a third condition, the ver-
bal description of the use of the object was requested (semantic condition);
in a fourth list, subjects were asked to show the use of the objects (seman-
tic-motor condition); and in a last condition, subjects had to pantomime
the use of objects that only were named by the experimenter (motor con-
dition). In this study, possible floor or ceiling effects were avoided by ad-
ministering lists of different lengths to control (15 items) and AD subjects
132 Carlesimo and Oscar-Berman

(9 items). Results confirmed previous findings that control subjects recalled

significantly more items from lists encoded in the semantic and semantic-
motor condition than from any of the other conditions, mildly demented
patients benefited marginally only from the object and semantic encoding
conditions, and severely demented patients seemed to profit only from the
motor condition. Unfortunately, since the Group x Encoding Condition in-
teraction was not reported, it is not possible to determine whether the con-
trols and mildly demented AD patients profited at the same level of the
semantic encoding. Completely contradictory results were reported by
Martin et al. (1985). These authors contrasted free and cued recall of four
lists of words in AD patients and age-matched controls. For each list, the
subjects were asked to make a different type of response, thereby creating
four different encoding conditions: free (no specific type of response re-
quired), rhyme (a word that sounds the same), place (where the object can
be found), and praxic (pantomime a movement associated with the use of
the object). Although recall by AD patients was poorer than the normal
controls, both groups showed the same pattern of results: better recall in
"place" and "praxic" conditions than in the "rhyme"condition (significant
Group and List main effects, and nonsignificant Group x List interaction).
In a different experimental paradigm, Butters et aL (1983) presented
a set of pictures to normal, AD, Korsakoff, Huntington, and right hemi-
sphere damaged subjects. Each picture contained three human or animal
figures against a scenic background. Half of the subjects were given short
stories linking the figures to the background scenes. In the retrieval phase,
only the background scenes were presented, and subjects were asked to
recognize the figures that had appeared originally in those scenes, as well
as their original position. Controls, Huntington, and right hemisphere
groups, despite different absolute levels of performance, showed a positive
effect of the story condition. By contrast, the Korsakoff and AD subjects
not only obtained the lowest absolute performance levels, but also demon-
strated no difference between the story and no-story conditions. Although
the authors claimed that AD patients were unable to benefit from verbal
mediation to improve their encoding operations, clear floor effects in the
AD and Korsakoff samples may have confounded the results.
Two studies tested the ability of AD patients to profit from a "gen-
eration effect" (Slamecka and Graf, 1978), a procedural requirement that
calls upon a subject's own ability to generate object names. Mitchell et al.
(1986) presented subject-verb-object sentences to young and elderly nor-
mal adults, and AD patients. In half of the sentences, the object was lacking
and had to be generated by the subjects. Results of a subsequent free recall
test showed a robust Group × Condition interaction. In fact, while AD and
normal elderly subjects obtained the same low levels of performance for
Memory Deficits in Alzheimer's Disease 133

the read words, the AD patients failed to profit from the generation con-
dition that had greatly improved the recall of the young and elderly normal
subjects. In a very complex experimental design, Dick et aI. (198%) varied
two operations requested during the study phase (reading vs. generation
of words, and semantic vs. phonological level of processing), as well as the
procedures to be used in retrieval (free recall, word completion, and forced-
choice recognition). In order to equate the two groups' retrieval levels,
longer lists of sentences were administered to the controls. For our purpose,
the most relevant outcome of this study (even if confounded by a probable
floor effect in the AD group) was the lack of any generation effect in the
AD subjects, regardless of the retrieval procedure they used. Mitchell et
al. (1986) and Dick et al. (1989a) interpreted their results as the proof that
AD patients were unable to utilize the generation procedure because of a
diminished ability to activate and retrieve information from semantic mem-
ory. An alternative account for their results might be that AD patients do
not benefit from increasing the effort (e.g., generating a word) needed to
create a distinctive memory representation (Jacoby, 1982).
The ability of AD patients to use multiple encoding representations
in order to improve their memory has been tested in several studies. Using
words and pictorial stimuli, Rissenberg and Glanzer (1986) compared levels
of retention in AD and normal control subjects. While normals showed
the expected advantage of pictures over words (Paivio, 1971), AD patients
failed to show any difference in retrieval of the two types of materials.
Contradictory results were obtained in a successive study by the same
authors (Rissenberg and Glanzer, 1987), who measured memory for con-
crete and abstract words in young normals, elderly normals, and AD pa-
tients. All three groups d e m o n s t r a t e d different absolute levels of
performance, but AD patients and young normaIs alike showed the con-
creteness effect (concrete words being better remembered than abstract
words). Elderly normals did not show the concreteness effect, a finding
attributed by the authors to diminished communication between visual and
verbal memory stores as a result of aging. The unexpected concreteness
effect observed in the AD patients was ascribed to their difficulty in ac-
cessing semantic representations for the abstract words (demonstrated in-
dependently in a separate word finding test).
Another way of testing the hypothesis that multiple memory repre-
sentations produce a positive effect on retention is to compare levels of
recall for a self-performed task with recall levels for a task requiring a
verbal description of events. According to Cohen (1981), when a subject
performs a manipulation task in response to a verbal instruction, the subject
also forms verbal, visual, and motor codes at the same time; in contrast,
when a subject simply describes an event, only a verbal code is activated.
134 Carlesimo and Oscar-Berman

The subsequent enhanced recall for self-performed tasks than for verbally
described tasks reflects the multiplicity of codes activated in the first con-
dition. Two studies employed this procedure to assess the encoding abilities
of AD patients. Dick et al. (1989b) argued for an encoding deficit in AD
patients after observing that young and old normal subjects showed higher
recall for self-performed tasks than for verbal descriptions, while levels of
recall of AD patients did not differ in the two conditions (significant Group
x Condition interaction). Data reported in a second study (Karlsson et al.,
1989) are less clear-cut. In fact, although the authors claim that all groups
(normal subjects, as well as mildly, moderately, and severely demented AD
patients) showed better recall for self-performed tasks than for verbal in-
formation, the significant Group x Condition interaction suggested that AD
patients benefited less than normal subjects from the multiple encoding
condition.
One study used the encoding-specificity paradigm (Thomson and
Tulving, 1970) to assess encoding abilities of AD patients. Granholm and
Butters (1988) presented to groups of normal, Huntington, and AD subjects
lists of words that were associated to strong (S), weak (W), or no (0) se-
mantic cues. Five experimental conditions were created by presenting the
same or different cues during the cued recall phase of the study: S-S (pres-
ence of strong cues during presentation and recall), W-W, S-W, W-S, 0-0.
Normal subjects and Huntington patients showed the expected better recall
when the same cues appeared at presentation and recall (S-S and W-W
conditions). However, AD patients improved their performance only when
strong cues were present during recall, regardless whether the same or dif-
ferent cues were present during the study phase. According to the authors,
AD patients failed to encode the semantic relationship between cue and
target words during cue presentation, and simply generated free associa-
tions during retrieval.
Finally, experimental paradigms used in three other studies manipu-
lated characteristics of the to-be-remembered material [low frequency vs.
high frequency words (Wilson, Bacon, Kramer et al., 1983)] or procedures
of presentation [slow vs. fast rates of presentation (Miller, 1971) and repe-
tition frequency of items (Strauss et al., 1985)]. All of these studies showed
an inability of AD patients to profit from the variables that increase levels
of retention in normal subjects (i.e., low-frequency words in recognition
tests, slow rates of presentation, and a high degree of repetition).
In summary, the literature reviewed so far indicates that evidence sup-
porting the view that AD patients suffer from reduced encoding ability is
suggestive but not unequivocal. A significant Group x Condition interaction
(implicating little advantage for AD patients from the use of experimental
conditions that increase retention in normal subjects) has been reported
Memory Deficits in Alzheimer~s Disease 135

in 14 studies of the 19 reviewed, while a nonsignificant interaction (the

same advantage for normal and AD subjects) has been noted in 5 studies.
An adequate interpretation of this pattern of results requires consid-
eration of important methodological and theoretical concerns. From a
methodological point of view, both procedural and sampling questions can
be raised. For example, in at least four positive studies (in which significant
interactions were observed) floor effects in AD patients' performance may
have confounded the results (Butters et al., 1983; Dick et al., 1989a; Strauss
et al., 1985; Weingartner et al., 1981). Furthermore, a low severity of de-
mentia in experimental samples may have given rise to negative results
(nonsignificant interactions). However, since independent measures of re-
sidual cognitive abilities in AD patients are lacking in many studies, the
hypothesis that only mildly demented patients were included in experiments
yielding negative results is merely conjectural.
From a theoretical point of view, the most common interpretation of
the encoding disability in AD patients assumes that these subjects fail to
utilize the semantic knowledge structure needed for efficient processing of
incoming information and for accessing relevant retrieval cues. In terms of
conceptual frameworks noted earlier in the paper, deficient information
processing will result in diminished access to the semantic meaning of
words, in poor communication between different codes, in a lack of dis-
tinctiveness of to-be-remembered items, or finally, in failure to establish
efficient links between target and contextual cues during the study phase.
Different mechanisms can be hypothesized to account for AD pa-
tients' reduced utilization of semantic knowledge to improve encoding op-
erations. The first hypothesis assumes that, because of a low level of
personal initiative, AD patients do not spontaneously engage in encoding
processes. This hypothesis was proposed by Baddeley (1986) to explain the
failure of demented patients to increase the level of retention when more
time was made available for processing the stimuli (Miller, 1971). The hy-
pothesis predicts that memory deficits in AD patients would be particularly
severe under experimental conditions that provide no help in stimulating
encoding operations. A significant improvement in memory performance
would be observed, however, when specific procedures are introduced to
circumvent the lack of initiative and to lead the subjects to engage in elabo-
rative processes. In the experimental literature reviewed above, this pre-
diction was not confirmed. A prevalence of studies showing an inability of
AD patients to benefit from studying conditions designed to enhance en-
coding levels emerges, in fact, either when no direct indication was pro-
vided to utilize the facilitating encoding conditions (e.g., recall of lists
formed by semantically related words) or when specific instructions were
136 Carlesimo and Oscar-Berman

given and the real accomplishment of them was verified (e.g., levels of proc-
essing paradigm, generation of to-be-remembered words, etc.).
Another hypothesis assumes that AD patients fail to utilize the se-
mantic knowledge to enhance their encoding operations because the nor-
mal semantic structure is lost in these patients. In effect, language disorders
are a common symptom of AD (Bayles and Tomoeda, 1983), and poor
comprehension of verbal stimuli could well account for poor verbal reten-
tion. Partial support for this view is provided by studies that have related
memory and linguistic competence in AD patients. Four studies (Martin
et al., 1985; Weingartner et al., 1981, 1983; Wilson et al., 1982) found posi-
tive correlations between verbal memory scores and performance on tests
of linguistic abilities. In one study reviewed earlier (Dick et al., 1989b),
however, the ability of AD patients to profit from the generation effect
did not correlate with scores on linguistic tests. It is not necessary to assume
a complete disruption of the semantic structure to hypothesize a deficient
utilization of the semantic knowledge in encoding operations. Martin et al.
(1985), for example, noting that AD patients frequently chose the errone-
ous semantic alternative in a multiple-choice recognition test, suggested
that these patients are able to access the correct semantic field, but their
encoding operations are not extensive enough to allow them a reliable dis-
crimination between semantically related words.
A somewhat different view has been offered by Nebes (1989). Ac-
cording to his detailed review of the literature on semantic memory in
AD, Nebes suggested that deficient performance on semantic tasks by AD
patients depends, at least in part, on reduced processing resources needed
to access the relevant semantic information rather than to a loss of se-
mantic knowledge. In this vein, Nebes suggested that deficient memory in
AD patients would be observed when the experimental procedures stress
the role of active semantic elaboration. By contrast, when the experimental
paradigm allows use of linguistic knowledge in an automatic way (e.g.,
approximation to English of strings of letters; Nebes et al., 1984, 1989),
AD patients show a performance profile similar to that exhibited by nor-
mal subjects.
It must be emphasized that the foregoing evaluation of the litera-
ture on encoding deficits in AD patients is based on a comparison of
outcomes from many highly diverse studies. Procedural and sampling dif-
ferences among the studies limit the reliability and the generality of the
conclusions that we can draw. Further studies that contrast, in the same
experimental group, predictions deriving from different hypotheses are
needed in order to reach clear conclusions about the nature of encoding
deficits in AD.
Memory Deficits in Alzheimer's Disease 137

Storage Deficits

Since amnesic and, to a lesser degree, AD patients generally display

better STM than LTM (Baddeley and Warrington, 1970; Cermak, 1982;
Miller, 1971; Wilson, Bacon, Fox, and Kaszniak, 1983), it might be sug-
gested that their memory disturbance reflects a deficiency in the consoli-
dation of m e m o r y traces. If so, the amnesic s y n d r o m e would be
characterized by an exaggerated rate of forgetting. Two main approaches
have been employed to evaluate this hypothesis. The first approach consists
of the assessment of memory for remote events (personal or public events
that occurred before the onset of the memory disorder), while the second
approach analyzes the temporal profile of forgetting of information ac-
quired subsequent to the onset of the disease.
Experimental data deriving from amnesic patients show that a remote
memory impairment is a constant feature of the syndrome, but that differ-
ent performance profiles emerge as a function of the etiology. The different
profiles are manifested in terms of the temporal extension of the deficit.
For example, the famous patient H.M. (with bilateral hippocampal resec-
tion; Corkin et aL, 1983), and subjects receiving bilateral electroconvulsive
therapy (ECT; Cohen and Squire, 1981) exhibited deficits involving a rela-
tively brief period immediately preceding the acute onset of amnesia. How-
ever, alcoholic Korsakoff patients (Albert et al., t979) could not remember
information spanning decades, and their remote memory deficit is charac-
terized by a temporal gradient, with the most recent events being remem-
bered most poorly. Performance profiles of Huntington patients (Albert et
aL, 1981) are flat, insofar as their memory disturbance occurred uniformly
across all the decades covered by the remote memory test. Many investi-
gators (e.g., Butters and Miliotis, 1985; Squire and Cohen, 1982) have ar-
gued that the heterogeneity of temporal profiles observed in the various
amnesic groups actually is the result of different mechanisms underlying
the loss of remote memories. In particular, deficient consolidation of re-
cently acquired information would be at the basis of the circumscribed defi-
cits shown by H.M. and ECT patients. By contrast, a multicomponent
hypothesis has been proposed to account for the temporal profile of remote
memory deficit exhibited by alcoholic Korsakoff patients. In these subjects,
deficient access to previously formed memory traces would be responsible
for the reduced ability in recollecting events from all lifetime periods; the
temporal gradient would result from deficient encoding of information dur-
ing the long period of chronic alcoholism responsible for the onset of Kor-
sakoff's disease (Albert et al., 1980). Finally, the performance profile
exhibited by Huntington patients likely is the consequence of reduced ac-
cess to memory traces for events occurring throughout the lifetime.
138 Carlesimo and Oscar-Berman

Studies concerned with the assessment of remote memory in AD sub-

jects have utilized diverse test instruments, dealing with the retrieval of
public or autobiographical events, and using recall or recognition proce-
dures. Probably because of the heterogeneity of neuropsychological instru-
ments used, contrasting results have been obtained regarding the temporal
profile of the remote memory impairment. In particular, Dall'Ora et aL
(1989), using a test of recall of autobiographical events, found a substan-
tially flat profile, with the retrieval of all events during the lifetime being
uniformly affected. Moscovitch (1982), by contrast, explored recognition
memory for famous faces in a small group of AD patients, and he discov-
ered a deficit involving only the two most recent decades. Beatty et aL
(1988) further tested remote memory for famous faces in AD patients by
means of free recall and cued recall procedures, and found a deficit that
extended to all decades and that was accompanied by a temporal gradient.
Kopelman (1989) used a very complex neuropsychological battery (tests of
recall of autobiographical events, and recall and recognition of famous
events and personalities). He obtained a profile characterized by deficient
retrieval across the lifetime, but with a gentle temporal gradient, the earliest
decades being the best remembered. Finally, Wilson et al. (1981) and Sagar
et al. (1988) highlighted a temporal gradient for recall of personal and pub-
lic events, and no temporal gradient (fiat profile) for recognition of famous
events or faces. The discrepancy in results does not permit clear-cut con-
clusions to be drawn. Evidence for absence of a temporal gradient in the
remote memory deficit (Dall'Ora et aL, 1989; recognition tests in Wilson
et al., 1981, and Sagar et aL, 1988) is in favor of the hypothesis that the
access to memories across the life span is uniformly impaired. By contrast,
the finding of a deficit limited to the two decades immediately preceding
onset of the disease (Moscovitch, 1982) is supportive of a deficient con-
solidation hypothesis. Finally, a performance profile characterized by defi-
cient retrieval of events across the life span, plus a temporal gradient
(Kopelman, 1989), could be explained by a combination of two above men-
tioned mechanisms.
Experimental evaluation of the rate of forgetting of information that
had been acquired after the onset of amnesia, was introduced by Huppert
and Piercy (1978). In two consecutive studies, Huppert and Piercy (1978,
1979) presented Korsakoff patients, H. M., and normal controls with slides
of pictures taken from magazines; acquisition was tested after 10 min, while
retention was tested after different intervals by means of a yes-no recog-
nition procedure. In order to equate levels of retention following the 10-
min study phase, the two groups were given different stimulus exposure
times. In this way, Huppert and Piercy (1978, 1979) were able to demon-
strate that rates of forgetting (measured after intervals of 24 hr and 7 days)
Memory Deficits in Alzheimer's Disease 139

were the same for normal and Korsakoff patients, but accelerated in H.
M. Squire (1981) observed abnormal rates of forgetting in ECT patients
and argued for the existence of two distinct forms of amnesia, each char-
acterized by a different rate of forgetting. In one type of amnesia, caused
by diencephalic pathology, forgetting is normal; in the other type of am-
nesia, caused by ECT's effects or by hippocampal lesions, forgetting is ac-
celerated. It should be noted, however, that this conclusion has been
challenged recently by a reanalysis of the forgetting profile in H. M. Freed
et al. (1987) varied H. M.'s stimulus exposure durations in order to equate
his initial level of retention exactly to that of the normal subjects. When
a forced-choice recognition task was used, H. M. showed a completely nor-
mal rate of forgetting; when a yes-no recognition task was utilized, how-
ever, an accelerated forgetting after 24 hr with normal forgetting after 3
and 7 days was observed.
The application of Huppert and Piercy's paradigm to AD patients
consistently has demonstrated normal rates of forgetting in these patients.
Corkin et al. (1984), using a forced-choice recognition procedure and as-
sessing the forgetting rate after 1 and 3 days, found comparable profiles
of forgetting in normal and AD subjects. Analogous results have been ob-
tained by Kopelman (1985), who evaluated the forgetting rate after 1 and
7 days by means of a yes-no recognition procedure. An interesting variation
to the original paradigm was introduced by Hart et al. (1987). These authors
tested acquisition 90 sec after the end of the study phase, and retention
after 10 min, 2 hr, and 18 hr, by means of a yes-no recognition procedure.
Even in this case, the initial acquisition level in normal and AD subjects
was equated by using different stimulus exposure times (2 sec for normal
subjects, and 11 sec for AD patients). The results showed early accelerated
forgetting in AD patients after the first 10 min, while the forgetting profiles
in the subsequent intervals (2 and 18 hr) were parallel in the two groups.
On the basis of these data, Hart et al. (1987) argued that forgetting was
accelerated in AD patients during the first minutes after acquisition of the
information, but normal after longer delays.
Rate of forgetting in the first few minutes after acquisition also can
be assessed by traditional memory tests, by calculating the difference be-
tween immediate and delayed retrieval. In these studies, however, generally
no attempt is made to match initial levels of acquisition in the subject
groups. Therefore, any conclusions that are drawn from the studies must be
viewed with caution. Notwithstanding, they represent a useful test of Hart
et al.'s (1987) suggestion of an accelerated early forgetting in AD patients.
In eight studies (Dells et al., 1991; Eslinger and Damasio, 1986; Helkala et
al., 1989; Moss et al., 1986; Ober et al., 1985; Salmon et al., 1989; Schultz
et al., 1986; Welsh et al., 1991) that reported immediate and delayed (be-
140 Carlesimo and Oscar-Berman

tween 2 and 30 min) recall of verbal material, the rate of forgetting was
between 10 and 25% for the control group, and 50% and 83% in the AD
group. In one study, in which facial recognition was assessed immediately
and after a 20-min delay period, forgetting was 2% in the controls and 29%
in AD patients (Eslinger and Damasio). In contrast, Becker et al. (1987)
argued for a normal rate of forgetting in AD patients, because AD and
control subjects were observed to lose a comparable number of bits of in-
formation passing from an immediate to a delayed (30-min) recall of a short
story and of the Rey-Osterrieth figure. Nevertheless, the lack of Group x
Delay interaction in that study may be tied to the fact that the statistical
analyses were performed on absolute rather than percentage scores. As a
matter of fact, a closer look at the data reveals that, after 30 min, control
subjects forgot 10 and 1.5% of the short story and of the Rey-Osterrieth
figure, respectively, while AD patients forgot 50 and 12%, respectively.
In conclusion, experimental findings reviewed in this section support
the hypothesis of accelerated forgetting of information by AD patients
within the first few minutes following acquisition. Information still present
in the subsequent period (10 min to several days) is retained normally.
Little can be said about retention of information by AD patients after
longer periods (months or years) because of the heterogeneity of results
obtained by studies on remote memory.

Retrieval Deficits

The final hypothesis to be considered here concerns possible retrieval

deficits in AD patients. Using Tulving's (1968) terminology, information
can be stored in memory and be available (i.e., learned) but not accessible
(i.e., retrievable).
The simplest way to conceptualize a retrieval disturbance is to assume
a reduced ability to actively access the stored information. Classical retrieval
models such as "sample-and-recognize" (Shiffrin, 1970) and "generate-and-
recognize" (Dale, 1967) assume an early stage in which information is ac-
tively searched or generated, followed by a recognition phase in which a
decision is made to accept or reject an item (Baddeley, 1976). In regard to
AD, the main prediction deriving from this hypothesis is that these patients
should be particularly advantaged when, in a memory task, the active re-
trieval phase (a) is made easier by providing appropriate cues (phonemic
or semantic cued recall) or (b) is completely circumvented by means of rec-
ognition procedures (forced-choice or yes-no recognition). The most appro-
priate experimental design to test this prediction consists in comparing
normal and AD subjects on tasks of free and cued recall, or recognition.
Memory Deficits in Aizheimer's Disease 141

The expected outcome would be greater improvement in AD patients than

in control groups following the utilization of cued recall or recognition pro-
cedures. To date, published findings generally do not support such a pre-
diction. Of three studies comparing free recall and phonemic cued recall in
normal and AD subjects, only one (Miller, 1975) clearly documented a sig-
nificant interaction in the expected direction; the other two did not (Davis
and Mumford, 1984; Dick et al., 1989a). Davis and Mumford (1984) found
that normal and AD subjects derived similar advantages from phonemic cu-
ing. Dick et al. (1989a) used an unusual method of administering the pho-
nemic cues. They provided explicit memory instructions during the study
phase, and at retrieval they used an implicit memory procedure. Therefore,
it is not possible to derive a clear interpretation of their data.
Among numerous additional studies comparing free and semantic
cued recall in normal and AD subjects, only in one study (using category
cuing; Cushman et al., 1988) was the expected interaction statistically sig-
nificant. No such findings were found by Diesfeldt (1984) using category
cuing, nor by Martin et al. (1985) using rhyme, contextual, and pantomime
cuing. The list is long. Negative results also were reported by Karlsson et
aL (1989) using category cuing and by Herlitz et al. (1991) using category
cuing. Further, Davis and Mumford (1984) and Grober and Buschke (1987)
reported interactions opposite to expectations; by their studies, control sub-
jects benefited more than AD patients from category cuing. Finally, paratlel
advantages of forced-choice or yes-no recognition on free recall procedures
have been documented in normal and AD subjects (Branconnier et al.,
1982; Helkala et al., 1989; Miller, 1975).
An alternative view of retrieval disorders arises from the classical
works of Warrington and Weiskrantz (1968, 1970) on amnesic patients.
These authors observed that when memory-disordered patients (deeply com-
promised on free recall and recognition tasks) were provided with fragments
of to-be-remembered material (e.g., the first letters of words, or degraded
pictures), retrieval dramatically increased to near normal levels. On the basis
of these data, Warrington and Weiskrantz (1968, 1970) argued that the main
source of difficulty for amnesic subjects is interference exerted by irrelevant
memory traces on target items. In other words, memory-disordered patients
would remember too many details and would not be able to discriminate
between relevant and irrelevant information. When the range of possible
choices was restricted by means of appropriate cues, the interference effect
would be circumvented, and amnesic subjects could successfully access the
relevant information. To accept the hypothesis that AD patients fail to re-
member effectively because of increased sensitivity to proactive interference
from irrelevant material, two predictions should be satisfied. According to
the first prediction, AD patients should be particularly advantaged by re-
142 Carlesimo and Oscar-Berman

trieval procedures that opportunely reduce the range of alternative choices.

In particular, AD patients should benefit from different types of cuing with
respect to free recall or recognition procedures. By contrast, increasing num-
ber of distractors in a multiple-choice recognition paradigm should nega-
tively influence their performance by enhancing the range of interfering
alternatives. The second prediction deals directly with the emergence of in-
terfering items when AD subjects are engaged in retrieval tasks. A high
number of false alarms in yes-no recognition tasks, and of intrusions in re-
call procedures, would be expected. Findings from the literature scarcely
support the first group of predictions. As described above, the number of
studies documenting the same recall advantage of phonemic or semantic
cues for AD and normal subjects clearly overrides evidence of a greater
advantage of cuing for AD patients. Two studies, however, reported a sig-
nificant difference between control and AD subjects in recognition tasks,
but no difference when subjects were provided with the first letter (Miller,
1975) or the first three letters (Morris, et al. 1983) of the words to be re-
called. Finally, Miller (1978, exp. 2) noted that the difference between per-
formance of AD and control subjects in a multiple-choice recognition task
grew wider as the number of response alternatives increased from two, to
four, to eight. Agreement does exist among investigators regarding the pro-
pensity of AD subjects to make intrusion errors or false alarms. In fact,
Miller and Lewis (1977), Branconnier et al. (1982), and Helkala et al. (1989)
reported a reduced number of hit rates and an increased number of false
alarms in AD patients compared to normals when tested by a yes-no rec-
ognition procedure, while a proportional (Hart et al., 1986; Miller, 1978,
exp. 1) or an absolute (Cushman et al., 1988; Fuld et al., 1982; Helkala et
al., 1989; Jacobs, Salmon et al., 1990; Jacobs, Troster et al., 1990) increment
of intrusions in free recall tasks has been frequently described in AD patients.
In conclusion, since experimental data only partially confirm the two
predictions regarding (a) the advantage given to AD patients by the avail-
ability of retrieval cuing, and (b) the high number of intrusions and false
alarms in their retrieval performance, the interference hypothesis does not
seem to provide the best account for this pattern of results. Instead, it ap-
pears that the simplest way to interpret these data would be that AD pa-
tients have reduced availability of stored information, as well as a
disinhibited inclination to guess the correct response. Such an explanation
does not directly involve a retrieval deficit.
A final way of conceptualizing defective retrieval mechanisms in AD
is provided by the Encoding Specificity theory. According to Tulving (1972,
1983), retrieval operations can be understood only with reference to en-
coding processes. In particular, the most efficient retrieval cues are not
absolute entities. Rather, they depend strictly on the conditions in which
Memory Deficits in Alzheimer's Disease 143

the information was learned and on the particular attributes of the infor-
mation that were encoded. A retrieval deficit, could, therefore, depend on
a decreased ability to combine available cues with attributes present during
the study phase. As described in the Encoding Deficits subsection of the
present paper, this hypothesis was evaluated in AD patients by Granholm
and Butters (1988). They used a paradigm proposed by Thompson and
Tulving (1970), and were able to demonstrate that while in normal subjects,
the word associated to the target during the study phase was the most ef-
ficient retrieval cue, in AD patients this relationship was lost. Grober and
Buschke (1987) interpreted in a similar vein their results showing a reduced
ability of AD patients to take advantage of category cues that had been
explicitly associated to the targets during the study phase.
While the studies just reviewed seem to support the prediction derived
from the Encoding Specificity hypothesis, the exact nature of the deficits re-
mains unclear. Because of the interactive interrelationship between encoding
and retrieval processes, the above-mentioned results could just as easily be
interpreted in terms of retrieval deficits, or encoding deficits. In other words,
AD patients might be equally unable to match the retrieval cues effectively
with the encoded attributes, or to encode the attributes at all.
In summary, the literature reviewed above can be hardly seen to sup-
port the view that a retrieval deficit is the main cause of memory disorders
in AD patients. Most of the data, documenting no special advantage to
AD patients of procedures to facilitate retrieval operations, are clearly
against any hypothesis suggesting deficient access to normally stored infor-
mation. Other data, taken as supportive of an interference hypothesis (or
a retrieval deficit as predicted by the Encoding Specificity hypothesis), can
perhaps most reasonably be interpreted in terms of difficulties with encod-
ing or in making decisions.

IMPLICIT MEMORY

A great deal of experimental work and theoretical elaboration has

been devoted, in the last decade, to Implicit Memory. As noted in the In-
troduction, the relevant difference between explicit and implicit memory
tasks concerns the contribution of awareness to the act of remembering
(Richardson-Klavehn and Bjork, 1988; Roediger, 1990; Tulving and Schac-
ter, 1990). In particular, while in explicit tasks subjects generally are re-
quested to recall or recognize information previously encountered in a
particular temporospatial context, in implicit memory tasks the influence
of specific training procedures or instructions is assessed without making
reference to the past experience. According to many authors (see, e.g.,
144 Carlesimo and Oscar-Berman

Richardson-Klavehn and Bjork, 1988, and Squire, 1987), further distinc-

tions should be made among different kinds of implicit memory tasks.
Squire (1987) argued, for example, for the existence of three different kinds
of implicit tasks---namely, learning of visuomotor or cognitive skills, repe-
tition priming, and simple classical conditioning.
For neuropsychologists, interest in implicit measures of memory has
been aroused by the discovery of a specific dissociation between implicit
and explicit memory performances in amnesic subjects. These subjects, al-
though severely impaired on classical recall and recognition tasks, generally
perform at the same level as normal subjects on tests of skill learning (Co-
hen and Squire, 1980; Corkin, 1968), repetition priming (Cermak et al.,
1985; Graf et al., 1984), and classical conditioning (Weiskrantz and War-
rington, 1979). More recently, a further neuropsychological dissociation be-
tween different implicit memory tests has been demonstrated. Patients with
Huntington's disease, who perform poorly on explicit memory tests, and
who do not show normal learning of cognitive (Martone et al., 1984) and
visuomotor (Heindel et al., 1988, 1989) skills, nevertheless are capable of
normal repetition priming effects (Heindel et al., 1989, 1990; Salmon et al.,
1988; Shimamura et al., 1987).
With respect to AD, we know of no studies of classical conditioning,
and only few studies have examined the ability to learn visuomotor skills.
Rather, considerable experimental effort has been devoted to assessing the
effects of repetition priming in AD patients. Experimental evidence regard-
ing the learning of visuomotor skills and the repetition priming effect in
AD subjects will be reviewed separately.

Learning of Skills

Results of studies from different laboratories have been thoroughly

consistent in documenting normal visuomotor learning abilities in AD pa-
tients. In the first study, Eslinger and Damasio (1986) administered a pur-
suit-rotor motor learning task to AD patients and age-matched controls.
The subjects were trained to maintain contact between a hand-held stylus
and a small metallic disk rotating on a turntable. Pursuit rotor learning
was assessed by changes in the total time on target, as well as the number
of times the stylus lost and regained contact with the target over trials (in-
cluding a delayed trial 20 min later). Although AD patients were signifi-
cantly worse than control subjects in keeping the stylus on target during
the first trial, their improvement over subsequent trials was comparable to
that of normal subjects such that the performances of the two groups were
undistinguishable, even in the delayed condition. Similar results were ob-
Memory Deficits in Alzheimer's Disease 345

tained by Heindel et al. (1988, 1989) in two successive studies. In the first
study (Heindel et al., 1988), AD patients were compared with amnesics,
Huntington patients, and age-matched normal subjects. In this, as well as
in the subsequent study, each subject's initial level of performance was
equated by varying the rotation speed of the disk. Results from six sub-
sequent blocks of trials indicated that AD, amnesic, and normal subjects
obtained comparable learning curves; only the Huntington patients were
severely impaired. In the second study (Heindel et al., 1989), AD, Huntington,
Parkinson, and normal subjects were tested on the same task. Once again,
the AD and normal control groups showed comparable degrees of improve-
ment over trials, and Huntington patients were impaired. Parkinsonian pa-
tients, who were free of signs of dementia, were able to learn the skill at
a normal rate; however, Parkinson patients with concomitant signs of de-
mentia were slow to learn the motor skill.
A different experimental paradigm was used by Knopman and Nissen
(1987). In this study, AD and control subjects were trained to press one of
four buttons in response to a light that appeared in one of four possible
positions on a computer screen. Visual reaction time was the dependent
variable. Unbeknown to the subjects, the same 10-trial sequence of light
positions was continuously repeated in the first four blocks of trials, while
in the final block the sequence was random. Although reaction times of AD
patients were globally slower than those of the normal controls, both groups
presented the same pattern of learning. Both groups showed a progressive
reduction in response times during the first four blocks, as well as a signifi-
cant increase in reaction times during the final (random) block. On the basis
of these results, Knopman and Nissen (1987) argued that AD patients are
capable of normal implicit acquisition of the stimulus-response sequence.
Gabrieli (1986) examined the ability of AD Patients to learn a mirror
tracing task. In this paradigm, the subjects were asked to trace a five-
pointed star without seeing their hand, the stylus, nor the star except as it
was reflected in a mirror. Over three consecutive days, AD and control
subjects demonstrated the same rates of learning of the visuomotor skill
(using error reduction and reaction time measures).
In another study, Heindel et al. (1991) tested the biasing of weight
judgments in AD and Huntington patients. Control and pathological groups
were first exposed to either a heavy or a light set of weights, and were
later asked to rate the heaviness of a standard set of 10 weights. Results
showed that weight judgments of AD, mildly demented Huntington, and
control patients were similarly biased by previous exposure to light or heavy
weights (they perceived the standard set of weights as heavier or lighter
following the previous exposure to respectively light or heavy weight).
Weight judgments by severely demented Huntington patients, by contrast,
146 Carlesimo and Oscar-Berman

were not significantly biased by previous exposure to light or heavy weights,

thereby suggesting a motor programming deficit in these patients.
Results of a final study were not so clear-cut. Grosse et aL (1991)
gave a finger maze test to AD and age-matched control subjects. Subjects
were asked to trace one particular maze over two consecutive blocks of 10
trials; then they were to trace a different maze for the last block of 10
trials. Unfortunately, in an attempt to equate the initial performance levels
of AD and control groups, control subjects were blindfolded. The utilization
of this very unusual procedure (acquisition of a visual-tactuat skill by AD
patients, learning of a tactual pattern by normal subjects) makes it difficult
to interpret the results. AD patients were generally faster than controls in
performing the mazes. However, since the learning curves were parallel in
two groups, results suggest a normal learning of the skill by AD patients.
In summary, data from many studies are consistent in suggesting that
visuomotor skill learning is preserved in AD. However, the extent to which
the results can be generalized to encompass all procedural learning skills
in AD seems, at the very least, questionable. Experimental paradigms used
in the studies mentioned above all explored the ability of AD patients to
learn a visuomotor skill. By contrast, the ability of AD subjects in acquiring
cognitive skills [e.g., the mirror reading task (Cohen and Squire, 1980), or
the Tower of Hanoi problem (Cohen and Corkin, 1981)] has been under-
investigated. Due to the different characteristics of the tasks, it seems rea-
sonable to predict a dissociation in AD between visuomotor proficiency on
the one hand, and learning cognitive skills on the other. Clearly, further
studies are needed to clarify this issue.

Repetition Priming

Repetition priming can be defined as an increase in accuracy or proc-

essing speed, as a function of previous exposure to a specific stimulus. What
makes repetition priming an "implicit" memory task is that subjects are
not informed of the relationship between stimuli presented during the study
phase and items presented in the testing phase. A different kind of priming
is "semantic priming," which explores the structure of the semantic net-
work. Semantic priming, unlike repetition priming, is generally short lasting
(few seconds). For a review of studies concerned with semantic priming in
AD, see Nebes (1989).
Repetition priming procedures and explicit memory tasks share many
points in common. Both require phases of study and recollection of similar
stimulus materials. Further, the interest in priming procedures was inspired
by the work of Warrington and Weiskrantz (1968, 1970), who were able
Memory Deficits in Aizheimer's Disease 147

to obtain comparable performance levels in normal and amnesic subjects

by making slight changes in the testing procedures of explicit memory. Con-
sequently, a current controversy concerns whether performance on explicit
memory tasks and with repetition priming procedures relies on the same
or different memory systems. To test the hypothesis that performances on
explicit and repetition priming tasks rely on different memory functions,
two dissociations have been invoked. First, it is well known that different
experimental manipulations can have opposite effect on explicit memory
and priming. For example, type of processing and level of attention during
encoding are critical factors for quality of retrieval in the domain of explicit
memory, but these variables exert little or no effect on priming procedures
(Jacoby and Dallas, 1981; Kirsner et al., 1983). In addition, while a change
in presentation modality from study to test phases may attenuate or elimi-
nate priming effects (Kirsner et al., 1983), the same experimental manipu-
lation generally has no effect on explicit retrieval (Graf et aL, 1985).
Further, evidence supporting the hypothesis that different memory systems
underlie performance on explicit memory and perceptual priming tasks
comes from the functional dissociations noted earlier. That is, patients who
are severely impaired on explicit memory tests generally perform at normal
(or near normal) levels on repetition priming tasks [e.g., amnesics (Cermak
et al., 1985; Graf et al., 1984; Shimamura, 1986), and Huntington patients
(Heindel et al., 1989, 1990; Salmon et al., 1988)]. A different position is,
however, held by other authors (Jacoby, 1983; Roediger, t990). Roediger
(1990) has argued, for example, that performance on explicit as well as
implicit tasks actually relies on the same memory for episodes; the differ-
ence between them is whether access to the information is conscious or
unconscious. The functional dissociations noted above are interpreted by
Roediger (1990) as stemming from different processing requests posed by
the experimental situations: Explicit memory tests generally draw on the
encoded meanings of concepts, while implicit memory tests rely heavily on
perceptual characteristics of the stimuli. In fact, when the experimental pro-
cedures are appropriately manipulated, functional dissociations that do not
respect the classical distinction between explicit and implicit memory tests
have been obtained (Roediger et al., 1989).
A number of different procedures have been utilized to evoke the repe-
tition priming effect. According to several authors (Gabrieli, 1991; Richard-
son-Klavehn and Bjork, 1988; Schacter, 1990; Tulving and Schacter, 1990), a
distinction should be made between priming tasks that rely primarily on the
analysis of the perceptual configurations of stimuli and tasks that require di-
rect access to the lexical-semantic properties of the words. Repetition priming
procedures that are based primarily on perceptual analysis of stimuli include
tasks of perceptual identification of words (Jacoby and Dallas, 1981) or fig-
148 Carlesimo and Oscar-Berman

ures (Jacoby and Brooks, 1984). Tasks that rely mainly on access to the lexi-
cal-semantic properties of the stimuli include free association (Shimamura
and Squire, 1984) or the generation of members of a semantic category (Graf
et al., 1985). In some cases, perceptual as well as lexical factors probably
concur to produce the priming effect, e.g., in tasks requiring stem completion
(Graf et al., 1984) or lexical decision (Kirsner et aL, 1983).
Studies concerned with the performance of AD subjects on repetition
priming tasks will now be reviewed according to this tentative classification.
A summary of the studies is presented in Table II.
In two consecutive studies, Keane, Gabrieli, Fennema et al. (1991)
and Keane, Gabrieli, Growden, and Corkin (1991) administered to A D and
age-matched control subjects tasks of perceptual identification of words and
three-letter nonwords. In this paradigm, the stimuli were presented for a
few milliseconds on a computer screen, and the exposure time was pro-
gressively augmented until subjects were able to identify the stimuli cor-
rectly. Priming effects were demonstrated by the shorter exposure times
needed to read stimuli previously encountered, compared to new items. In
both studies, AD patients required longer exposure times than normal sub-

Table II. Studies Concerned with Repetition Priming Effects in AD Patients

Priming effect
Task Author in AD patients
Perceptual identification of Keane, Gabrieli, Fennema
--Words et al., 1991, exp. 2 Normal
--Nonwords Keane et al., 1991 Normal
- - L i n e drawings Heindel et aL, 1990 Reduced
Gabrieli et aL, 1992, exp. 2 Normal a
Stem completion Shimamura et aL, I987 Reduced
Salmon et al., 1988 Reduced
Heindel et al., 1989 Reduced
Keane, Gabrieli, Fennema
et al., 1991, exp. 3 Reduced
Gabrieli et aL, 1992, exp. 1 Reduced
Grosse et aL, 1990 Normal
Partridge et aL, 1990 Normal
Lexical decision Ober and Shenaut, 1988 Normal b
Free association Salmon et aL, 1988, exp. 2 Reduced
Huff et aL, 1988, exp. 3 Reduced
Brandt et al., t988 Reduced

aA reanalysis of the priming effect in proportional rather than in absolute terms actually shows
a reduced priming effect in the AD sample.
bDue to differences in baseline levels, reanalyses of the priming effects in proportional rather
than in absolute terms likely would show reduced priming in the AD group.
Memory Deficits in Alzheimer's Disease 149

jects. AD and control groups nevertheless read the old items more quickly
than new ones, with no interaction between group and stimulus type. The
finding of normal priming in the AD group was strengthened by the ab-
sence of group differences, whether priming was calculated as a percentage
of baseline performance or as an absolute score.
Two other studies (Gabrieli et al., 1992; Heindel et al., 1990) examined
the perceptual identification of' figures. In the first study (Heindel et al.,
1990), subjects were initially previously presented with 30 line drawings.
During a subsequent testing phase, 15 old and 15 new drawings were pre-
sented, each one at five different levels of perceptual fragmentation. Re-
sults indicated that the priming effect (represented by the difference
between levels of fragmentation required for identifying old and new draw-
ings) was significantly reduced in AD compared to normal and Huntington
subjects (who were unimpaired). Contrasting results were obtained by
Gabrieli et al. (1992, exp. 2). AD and normal control subjects were asked
to perform an incomplete-figures task twice, one hour apart, The task was
the Gollin Incomplete Pictures Task (Gollin, 1960), which requires identi-
fication of 20 line drawings, each at five different levels of perceptual frag-
mentation, and the priming effect was measured as a reduction in the
number of errors in the second test session compared to the first. In this
study, although the AD subjects made man 3, more errors than the controls
in both sessions, their absolute levels of priming was comparable to normal
subjects', k should be noted, however, that differences in baseline perform-
ance (errors in the test session) render an absolute priming effect mislead-
ing. Indeed, when the priming effect was recalculated in proportional rather
than in absolute terms, priming by AD patients was significantly lower than
normal subjects (25 and 61%, respectively).
Stem completion, probably because it is an easy task to perform, is
the repetition priming procedure most frequently used in experimental stud-
ies on AD patients. In this test, subjects first are exposed to a list of words.
Subsequently, they are presented with a list of three-letter stems, and are
asked with each to give the first word that come to mind. The priming effect
is represented by the difference between the percentage of stems completed
with previously read words and a baseline measures (stems completed after
no previous exposure). In numerous different studies of this type, AD pa-
tients evidenced deficient priming (Gabrieli et al., 1992; Heindel et al., 1989;
Keane, Gabrieli, Fennema et al., 1991, exp. 3; Salmon et al., 1988, exp. 1;
Shimamura et al., 1987), even when compared to amnesics of different eti-
ologies (Heindel et al., 1989; Salmon et al., 1988, exp. 1; Shimamura et al.,
t987), Huntington patients (Heindel et al., 1989; Salmon et al., 1988, exp.
1; Shimamura et al., 1987), and nondemented Parkinson patients (Heindel
et al., 1989). By contrast, two other studies using stem completion found
150 Carlesimo and Oscar-Berman

comparable levels of priming in AD and control subjects (Grosse et al., 1990;

Partridge et al., 1990). As usual, sampling and procedural differences could
have contributed to the contrasting results. However, in at least two studies
that obtained contrasting results (Grosse et al., 1990; Shimamura et al.,
1987), the ages and severity of dementia were comparable in the AD sam-
ples. As for possible variations in experimental procedures, the most relevant
difference seems represented by the orienting exercise practiced on target
stimuli during the study phase. Thus, it is possible that the orienting exercise
in the two experiments demonstrating normal priming in AD patients
(Grosse et al., 1990; Partridge et al., 1990) favored deeper processing than
the other studies (Gabrieli et al., 1991b; Heindel et al., 1989; Keane Gabrieli,
Fennema et al., 1991, exp. 3; Salmon et al., 1988, exp. 1; Shimamura et al.,
1987). Partridge et al. (1990) required their subjects to provide an appro-
priate meaning for the words of the study list, and Grosse et al. (1990) asked
their subjects to generate the last word in a number of incomplete sentences.
Keane, Gabrieli, Fennema et al., (1991, exp. 3), by contrast, simply asked
the subjects to read the words, and Shimamura et al (1987), Salmon et al.
(1988, exp. 1), and Heindel et al. (1989) had the subjects rate each word
on a 5-point likability scale. In short, it may be that the type of orienting
exercise during the study phase, perhaps by varying the level of semantic
processing or the attentional resources involved, determines the level of
priming in AD subjects. It must be noted, however, that this hypothesis does
not go unchallenged. There is considerable evidence from stem completion
tasks that type of encoding and level of attention do not influence the level
of priming of normal subjects (Richardson-Klavehn and Bjork, 1988). Fur-
ther, Gabrieli et al. (1992) employed incidental phonological ("Does the
word contain the letter A?"), incidental semantic ("Is the word the name
of something you could touch?"), and intentional ("Remember the word!")
encoding conditions with AD and control subjects, and observed no group
differences in the subsequent priming levels regardless of the type of en-
coding.
In a study of semantic priming using a lexical decision task, Ober and
Shenaut (1988) found comparable repetition priming effects in AD and
control subjects (shorter reaction times for previously presented stimuli
than for new words). However, the baseline of the two groups differed con-
siderably; reaction times of the AD patients were about twice as slow as
those of control subjects. It is possible, therefore, that reanalyses of the
data in terms of proportional rather than absolute levels of priming would
demonstrate deficient priming by the AD patients.
Finally, a free association task was employed in three additional stud-
ies (Brandt et aL, 1988; Huff et al., 1988, exp. 3; Salmon et al., 1988, exp.
2). The task, modeled on a procedure described by Shimamura and Squire
Memory Deficits in Alzheimer's Disease 151

(1984), began by exposing subjects to a list of related word pairs. In the

subsequent testing phase, only the first word of each pair was presented,
and subjects were asked to say the first word that came to mind. Priming
was measured by the difference between percentage of previous word pairs
correctly competed, and baseline (percentage of the same word pairs cor-
rectly completed after no previous exposure). Results of these studies
clearly documented a deficient priming effect in AD patients, whether they
were compared with normal control subjects (Brandt et al., 1988; Salmon
et aL, 1988, exp. 2), aphasic patients (Huff et al., 1988, exp. 3), or Huntington
patients (Salmon et al., 1988, exp. 2).
In summary, most studies have documented deficient repetition priming
effects in AD patients. Nevertheless, a close examination of results of specific
experiments has revealed important dissociations among performances on dif-
ferent tasks. In particular, it seems that AD patients are especially compro-
mised when the priming procedure directly involves the processing of
lexical-semantic properties of stimuli. In three studies that used a free asso-
ciation procedure (Brandt et al., 1988; Huff et al., 1988, exp. 3; Salmon et
al., 1988, exp. 2), AD subjects showed deficient priming. By contrast, when
tasks were used that relied on the analysis of the perceptual configurations
of stimuli, priming more frequently was normal (Gabrieli et al., 1992; Keane,
Gabrieli, Fennema et al., 1991; Keane, Gabrieli, Growden, and Corkin, 1991;
but see the above discussion on interpreting the results of Gabrieli et al.,
1992, exp. 2). Finally, studies employing priming procedures that likely re-
quire both perceptual and lexical-semantic analyses of stimuli (such as stem
completion and lexical decision) have yielded mixed results.
Some investigators (Gabrieli, 1991; Gabrieli et al., 1992; Keane,
Gabrieli, Fennema et alo, 1991; Keane, Gabrieli, Growden, and Corkin,
1991) have argued that dissociations observed in AD patients among prim-
ing effects evoked by different experimental procedures actually reflect dif-
ferential impairments in separate memory systems. Perceptual identification
tasks would be mediated by a memory system that processes perceptual
configurations of words and figures. Priming tasks that rely mainly on access
to the lexical meaning of the words would be mediated by a Iexical-semantic
memory system. Since distinct neural circuits probably subsume the opera-
tions of these different memory systems, it is supposed that pathological
changes in AD would affect these memory systems differentially.
As noted above, however, the assumption that distinct memory, sys-
tems might be involved in explicit memory tasks and in different repetition
priming procedures is far from being universally accepted. For contrasting
opinions, see Roediger (1990) and Schacter (1990). In summary, the only
definitive conclusion that can be drawn from studies of priming in AD pa-
tients is that more data are needed from studies employing similar meth-
152 Carlesimo and Oscar-Berman

odologies and homogeneous patient groups. Further, the cognitive under-

pinnings and underlying neural mechanisms of priming effects deserve con-
siderable research effort, hopefully leading to clarification of the
neurobehavioral functions and brain systems involved.

SYNTHESIS OF DATA AND CONCLUSIONS

The pattern of memory impairments in AD is not uniform across all

domains of memory functions. Thus, it is clear that (a) explicit memory is
severely and consistently disrupted, and (b) acquisition of visuomotor skills
remains intact; (c) the integrity of implicit memory (with contrasting results
from repetition priming experiments) is still uncertain. From a neuropsy-
chological point of view, two questions are particularly relevant. First, is it
possible to identify basic cognitive defects subsuming this pattern of im-
pairment? And second, what is the relationship between the pattern of
memory loss and the underlying neuroanatomical and neurochemical
changes? In attempting to answer these questions at this time, an effort
will be made to synthesize neuropsychological findings on AD patients with
results of neuroanatomical and neurochemical studies.

Explicit Memory

Cognitive Defects

In reviewing the neuropsychological literature concerned with explicit

memory deficits in AD, we chose to integrate the available data within a
traditional conceptual framework that recognizes distinct phases of acqui-
sition, encoding, storage, and retrieval of information. The underlying as-
sumption was that a defect at one or more of these functional levels could
produce the explicit memory deficits of AD subjects.
As discussed earlier (see the section on Deficits in Acquisition), STM
plays an important role in acquisition of information. Early structural mod-
els of memory (e.g., the modal model; Atkinson and Shiffrin, 1968, 1971)
viewed STM as a unitary, capacity-limited store in which incoming infor-
mation is temporarily retained before being transferred to LTM. More re-
cent theoretical frameworks, however, have stressed the active role of STM
(conceptualized as a "working memory," e.g., Baddeley, 1986) in tempo-
rarily storing and processing relevant information during tasks of compre-
hension, reasoning, and learning. According to this view, the role of STM
would not be limited merely to rehearsal of incoming material, but would
Memory Deficits in Alzheimer's Disease 153

encompass mobilization and maintenance of attentional resources as well,

to permit adequate encoding to occur. In this vein, encoding would take
place as a result of the combination of STM (allocation and maintenance
of attentional resources) and general semantic knowledge (access to mul-
tiple codes, such as verbal, perceptual, etc., that organize and integrate in-
formation). After encoding operations have been accomplished and the
memory trace has been formed, it is supposed that consolidation processes
(probably more automatic in nature) act in order to make the trace durable
and available for subsequent retrieval.
Experimental evidence reviewed in this paper has suggested that both
encoding and consolidation operations are defective in AD. More often
than not, AD patients have demonstrated less of an improvement than nor-
mal subjects from experimental procedures that enhance retrieval by im-
proving the quality of encoding processes. While the presence of an
encoding deficit in AD is widely accepted, the basic cognitive mechanisms
of this condition are in debate. It might be useful, once again, to consider
the possibilities of reduced processing resources and deficient utilization
of semantic knowledge as conditions contributing to defective encoding. In
effect, both these components have been found impaired in AD patients.
Reduced processing resources have been demonstrated in dual competition
tasks (Baddeley et al., 1986; Morris, 1986), and disruption in active access
to semantic knowledge has been extensively documented (Nebes, 1989).
Determining the relative contributions of attentional and semantic deficits
to the reduced encoding abilities of AD patients is, however, a difficult
task. For example, the role of a texical-semantic impoverishment at the
basis of deficient memory performance has been emphasized by studies
showing significant correlations between scores on verbal and memory tasks
in AD patients (e.g., Weingartner et al., 1981), yet performance on most
verbal tasks utilized in these studies (e.g., word fluency, Token test) relies
on the availability of normal attentional resources. Moreover, when the lexi-
cal structure of the verbal material can be accessed automatically (i.e., with
a minimum demand on attentional resources), AD patients benefit from
semantic knowledge to the same extent as normal subjects (Nebes et at.,
1984, 1989). In conclusion, then, while it is possible that both attentionaI
and semantic deficits concur in reducing the encoding abilities of AD sub-
jects, clear experimental demonstration of their relative contributions have
not yet been produced. Studies contrasting predictions derived directly from
the "attentional" and the "semantic" deficiency hypotheses are needed, in
order to clarify this issue further.
Rate of forgetting traditionally has been investigated by means of Hup-
pert and Piercy's (1978) experimental paradigm. Studies that applied this
procedure have consistently found a comparable rate of forgetting in A D
154 Carlesimo and Oscar-Berman

and control groups, and have thus inferred efficient memory consolidation
in AD subjects (Corkin et al., 1984; Kopelman, 1985). As previously re-
marked, however, Huppert and Piercy's paradigm explores the forgetting
that occurs 10 min to some days after acquisition of the information. What
happens in the very first minutes following the study phase is not explored
by this paradigm. In fact, results from an experiment that appropriately var-
ied Huppert and Piercy's procedure (Hart et al., 1987; as well as outcomes
of several studies comparing immediate and delayed retrieval--e.g., Eslinger
and Damasio, 1986; Helkala et al., 1989) showed that, in the first 10 min
following acquisition, a significantly larger proportion of information was
lost in AD than in control subjects. This pattern of results would suggest
that defective consolidation of the memory trace in the first minutes after
acquisition results in an exalted early loss of information. Normal consoli-
dation in the subsequent period (after about 10 rain), by contrast, would
permit any remaining traces to be stored normally. Although this hypothesis
seems attractive and sufficiently grounded in experimental evidence, two
cautions need to be expressed. The first deals with the experimental evidence
on remote memory in AD subjects. According to the view of defective early
consolidation in the face of a later normal consolidation into LTM, remote
memories (i.e., those acquired before the onset of the disease) should not
be disrupted by consolidation deficits. Rather, the remote memory impair-
ment generally observed in AD subjects should reflect either a uniform loss
or an impaired access to normally consolidated memory traces. If this is
true, however, the remote memory deficit displayed by AD patients should
involve uniformly all lifetime periods and no temporal gradient would be
expected to emerge. This prediction, unfortunately, has not been consistently
confirmed: Flat temporal profiles (in line with the prediction) as well as
steep temporal gradients (contradicting the prediction) have been reported
in different studies. The second cautionary note about consolidation is theo-
retical in nature. The hypothesis that early and late consolidation processes
are, respectively, impaired and preserved in AD subjects is conceivable only
if one admits that the two processes are functionally dissociable. Unfortu-
nately, since a distinction between early and late forgetting in the domain
of LTM has not been investigated experimentally to our knowledge, any
presumption of its existence in AD patients would be untenable. Neverthe-
less, the hypothesis is worthy of further consideration.

Neuroanatomical Changes

The neuropathological changes that are evident in the brains of AD

patients are extensive. Cortical and subcortical structures are characterized
Memory Deficits in Alzheimer's Disease ][55

by diffusely distributed neuronal loss, senile plaques, and neurofibrillary

tangles. Such pervasive damage no doubt is responsible for the plethora of
cognitive deficits exhibited by AD subjects. In order to clarify the localiza-
tion of brain lesions critical for the memory dysfunctions in AD subjects,
it will be helpful first to refer to studies concerned with the neuroanatomy
of patients with clearly defined anterograde amnesia. Despite controversy
about which structures are critical for normal memory processes, it is widely
accepted that brain damage in amnesic subjects involves the mesial tem-
poral areas (hippocampus and amygdala; Butters and Stuss, 1989; Mishkin,
1982; Scoville and Mflner, 1957) and/or portions of the diencephalon
(mammillary nuclei, mediodorsal nucleus of thalamus, and mammillotha-
lamic tract; Brion and Mikol, 1978; Victor et al., 1971; Von Cramon et al.,
1985). Among these areas, the most consistently and extensively damaged
in AD are the mesial temporal structures. In a series of papers, Hyman et
al., (1984, 1986, 1987, 1990) have documented that the localization of
Alzheimer-related anatomical changes in the hippocampus is not aspecific,
but involves, in a selective manner, large pyramidal neurons of the CA1
area, subiculum, parasubiculum and layers II and IV of the enthorinal cor-
tex. As a result, most of the neurons that assure wide connections between
hippocampus and multiple cortical and subcortical areas are damaged,
thereby leaving the hippocampus functionally isolated. The amygdala also
is severely compromised in AD, especially the accessory basal, lateral basal,
and cortical nuclei (Hyman et al., 1990). Diencephalic structures (mammil-
lary complex and thalamic nuclei), by contrast, are less consistently and
severely affected (for a review, see Van Hoesen and Damasio, 1987).
On the basis of these data, it could be concluded that the underlying
neuropathology in AD patients involves some of the same structures im-
plicated in amnesic patients with mesial temporal lesions (e.g., surgical pa-
tients like H.M. and herpes encephalitis patients), but does not affect
structures implicated in diencephalic amnesia (e.g., Korsakoff's syndrome).
It is difficult to establish, however, if this neuroanatomical distinction has
some functional consequence. The relative contribution of mesiat temporal
and diencephalic structures to explicit memory processes is, to date, un-
known, and no consistent differences have been demonstrated between
memory profiles of amnesics with mesiat temporal vs. diencephalic damage.
As previously noted, in fact, the claim that there are two kinds of amnesia
distinguishable on the basis of rate of forgetting (normal in diencephalic
and exalted in mesial temporal amnesias) has been challenged (Freed et
al., 1987). Moreover, the findings in alcoholic Korsakoff patients of peculiar
deficits such as source amnesia (Shimamura and Squire, 1987), impaired
temporal order judgment (Squire, 1982), failure to release from proactive
interference (Squire, 1982), and attentional disorders (Oscar-Berman,
156 Carlesimo and Oscar-Berman

1973), all of which generally are absent in mesial temporal amnesics, is

probably related more to concomitant damage of frontal cortex than to
damage of specific diencephalic structure per se (Oscar-Berman, 1980; Os-
car-Berman et al., 1991; Squire, 1987).
In order to fully characterize basic mechanisms of memory impairment
in AD, another consideration must be kept in mind besides mesial temporal
and diencephalic involvement. As above noted, encoding of information and
consolidation of the memory trace can be considered sequential phases in
the learning process. A simple way to conceptualize the relationship between
brain structures and memory processes would be to make the following as-
sumptions: cortical association regions of the brain play a specific role in
processing structural and semantic properties of incoming stimuli, while
mesial temporal areas, in concert with diencephalic structures, are involved
in consolidation of the memory trace (for a detailed discussion, see Squire,
1987; but see, for a different view, Weiskrantz, 1982). Functional commu-
nication between these two systems is assured by prolific afferent and ef-
ferent projections connecting the hippocampal-amygdala complex with a
wide range of unimodal and multimodal cortical association areas (Amaral,
1987). If one accepts these assumptions, a natural conclusion would be that
amnesic and AD subjects differ with respect to degree of involvement of
brain structures devoted to encode or consolidate the information. The two
groups share a severe damage to brain regions presumed necessary for con-
solidation of the memory trace (mesial temporal or diencephalic structures
in amnesics, mesial temporal areas in AD subjects). However, in anterograde
amnesia (uncomplicated by alcohol-related cortical damage), the anatomic
damage generally is confined to these structures, thereby accounting for the
selectivity of the memory deficit. In AD, in contrast, neuroanatomicat
changes exceed the boundaries of mesial temporal areas, affecting other cor-
tical structures presumed important for encoding. Cortical areas known to
be damaged in AD are extensive, including the frontal, parietal and temporal
lobes; primary motor and sensory areas are relatively spared (Brun and
Englund, 1981; Pearson et aL, 1985; for a review, see Van Hoesen and
Damasio, 1987). Thus, on the basis of neuropathological findings, it seems
likely that different brain systems are responsible for the memory impair-
ment of AD and amnesic subjects. It follows, then, that while both conditions
would be characterized by impaired consolidation of the memory trace, only
AD would present the additional deficiency in encoding.
Although this general hypothesis is attractive for its simplicity, it does
not derive support from much neuropsychological evidence. First of all, the
assumption that anterograde amnesia results from defective consolidation
of memory traces is, at the very least, questionable. As noted earlier, the
remote memory deficit of amnesics generally has been attributed to defec-
Memory Deficits in Alzheimer's Disease 157

tive consolidation mechanisms. It should be noted, however, that the ap-

plication of Huppert and Piercy's (1978) paradigm generally has shown am-
nesic subjects to have a normal rate of forgetting. Second, if damage to
different brain systems underlies the cognitive changes responsible for the
memory impairments of AD and amnesic subjects, one would expect to
find clear and relevant differences in the memory profiles of these two
populations. However, there are only a handful of studies directly compar-
ing the explicit memory performances of AD and amnesic groups, and the
data from these studies suggest that the memory profiles of two groups
are quite similar. Thus, while Strauss et al. (1985) documented that amne-
sics, but not AD subjects, were able to profit from multiple exposures to
stimuli in order to enhance the level of recall, Corkin (1982) and Butters
et at. (1983) found that both AD and amnesics were deficient in utilizing
the semantic processing of stimuli to improve their memory performances.
As for consolidation processes, Kopelman (1985) documented (by means
of Huppert and Piercy's paradigm) a normal rate of forgetting in both AD
and Korsakoff patients. Moss et al. (1986), however, compared immediate
and delayed recall in AD and Korsakoff groups, and found abnormally
rapid early forgetting only by the AD patients. Finally, Kopelman (1989)
and Dall'Ora et al. (1989) described patterns of remote memory impairment
that were similar in amnesic and AD subjects.
In summary, results of research to date give the distinct impression
that the memory profiles of AD and amnesic subjects are similar in many
respects, including the ability to utilize the semantic processing of stimuli
to increase learning. Neuropsychological findings, in other words, fail to
substantiate predictions derived from neuropathological data. A possible
account of this discrepancy is that the neuropathology in different amnesic
syndromes is so diversely heterogeneous that "true" functional differences
between memory profiles of anterograde amnesia and degenerative demen-
tia are obscured (see, for example, the distinction between amnesics af-
fected by widespread or focal cerebral damage in Dall'Ora et al. 1989). It
is also possible that deficits in acquisition and consolidation processes in
AD and amnesic subjects are so severe as to mask possible differences in
encoding processes.

Neurochemical Changes

A final aspect of the Alzheimer-related cerebral changes in need of

discussion concerns an impoverishment of specific brain neurotransmitters.
This topic is constantly updated by new research findings, but some clear
information has emerged. Most consistently observed has been a reduction
158 Carlesimo and Oscar-Berman

of cholinergic activity in neocortical and hippocampal regions (Rossor et

al., 1984). An important decrease of glutamate content has also been de-
scribed in the cortical and hippocampal regions of the brain of these pa-
tients (Maragos et aL, 1987). Diminished levels of the neuropeptide
somatostatin, moreover, have been found in cortical and subcortical areas
of AD brains (Davies et at., 1980). Other neurotransmitters (e.g., no-
repinephrine and serotonin), finally, appear normal or only slightly de-
creased (for comprehensive reviews, see Hardy et aL, 1985; Perry and Perry,
1986; Wallin and Gottfries, 1990). The cholinergic deficit in AD patients
is especially relevant here, because of the well-established role of acetyl-
choline in normal human memory. Pharmacological studies have shown,
for example, that the administration of cholinergic blockers to normal sub-
jects significantly reduces performances on a wide range of memory tasks
(see Kopelman, 1986, for a review).
It is generally accepted that reduced cholinergic activity in cortical
and hippocampal regions in AD is due, at least in part, to neuronal loss
within the basal forebrain, particularly, the nucleus basalis of Meynert
(Whitehouse et aL, 1982). Ascending projections from nuclei of the basal
forebrain provide the most conspicuous source of neocortical and hippo-
campal acetylcholine. Interestingly, neuronal impoverishment of the nu-
cleus basalis of Meynert also has been reported in brains of Korsakoff
patients (Arendt et al., 1983), and focal lesions of the basal forebrain have
been linked to amnestic behaviors (Damasio et al., 1985).
Despite considerable convergent (albeit, indirect) evidence for cholin-
ergic transmitters having a role in memory, the manner in which cholinergic
insufficiency relates to memory impairment in AD is still unknown. It is
not clear, for example, whether the neurochemical change is directly re-
sponsible for the cognitive changes, nor to what extent the cholinergic de-
pletion of the neocortex is due to the loss of forebrain neurons or is
consequence of a regional degeneration of cholinergic neurons (Van Hoe-
sen and Damasio, 1987). To complicate matters further, it should be noted
that the profile of memory impairments produced in normal subjects by
inducing a cholinergic blockade pharmacologically only partially resembles
that exhibited by AD patients. Kopelman and Corn (1988) reported, for
example, that normal subjects who had taken scopolamine performed
poorly on long-term explicit memory tasks, but showed no deficit on STM
tasks, nor on tests of semantic memory, remote memory, and implicit mem-
ory. In all of these tasks (with the exception of visuomotor learning and
some forms of repetition priming), by contrast, AD subjects are deficient.
It would seem reasonable to conclude, therefore, that while the cholinergic
deficit in AD may indeed contribute to memory loss, it is not the only
critical determinant.
Memory Deficits in Alzheimer's Disease t59

Implicit Memory

As repeatedly remarked, implicit memory is only partially defective

in AD. Visuomotor learning generally is spared, as is repetition priming in
many instances. The existence of neuropsychological dissociations among
different kinds of memory tasks probably is the most cogent argument in
favor of a fractionation of memory into a number of subparts, each one
mediated by distinct brain systems (Roediger, 1990; Schacter, 1990; Squire,
1987; Tulving and Schacter, 1990). For example, the finding that pure am-
nesics perform normally on implicit memory tasks (Cermak et al., 1985;
Cohen and Squire, 1980; Graf et aL, 1984) suggests that mesial temporal
and diencephalic structures (damaged in amnesic subjects) support explicit,
but not implicit, memory operations. Further evidence for fractionating
memory systems comes from the finding that Huntington patients perform
normally on repetition priming tasks (Heindel et aL, 1990; Shimamura et
al., 1987) but are impaired on learning certain cognitive tasks (Martone et
al., 1984) and visuomotor procedures (Heindel et aL, 1988); thus, distinct
neural circuits appear to subsume performance on two kinds of implicit
tasks. It is interesting to note that AD patients show a pattern of implicit
memory impairment opposite to that exhibited by Huntington subjects. In
AD, performance on repetition priming tasks typically is defective, while
visuomotor learning is normal. This double dissociation (documented in
the same experimental sample by Heindel et aL, 1989) suggests that the
learning of visuomotor skills relies on a neural system involving basal gan-
glia structures (damaged in Huntington's disease but generally spared in
AD), while repetition priming involves neocortical associative areas (rela-
tively preserved in Huntington disease but heavily damaged in AD).
One of most widely (if not universally) accepted interpretations of
the repetition priming effect postulates that when a subject is exposed to
a given stimulus the previous representation of the same stimulus in se-
mantic memory is converted to a state of "functional activation." The suc-
cessive presentation of the same stimulus (or a fragment of it) results, as
a consequence, in faster processing or higher likelihood of recall. In line
with this view, some authors (Heindel et aL, 1989; Salmon et al., 1988; Shi-
mamura et aL, 1987) have suggested that repetition priming is defective in
AD because of a faulty semantic knowledge structure (which, as just men-
tioned, would otherwise support the priming effect). A somewhat different
view, however, is held by other authors, who have demonstrated that AD
subjects are not uniformly deficient on repetition priming procedures
(Gabrieli, 1991; Gabrieli et aL, 1992; Keane, Gabrieli, Fennema et aL, t991;
Keane, Gabrieli, Growden, and Corkin, 1991). These authors have sug-
gested that performance is reduced when repetition priming tasks depend
160 Carlesimo and Oscar-Berman

on access to the lexical-semantic features of stimuli; however, normal prim-

ing effects can be demonstrated by employing experimental paradigms that
depend upon visuoperceptual processing of stimuli. Keane and colleagues
(Keane, Gabrieli, Fennema et al., 1991; Keane, Gabrieli, Growden, and
Corkin, 1991) have argued that this neuropsychological dissociation occurs
because priming effects in "perceptual" and "lexical-semantic" procedures
actually are supported by distinct memory systems. Specifically, perceptual
priming tasks would be mediated by a memory system for the visual con-
figuration of words and objects; on the basis of metabolic studies on normal
subjects (Petersen et al., 1990), Keane, Gabrieli, Fennema et al., (1991)
have suggested that the neural substrate for this memory system is located
in the occipital cortex. Lexical-semantic priming, by contrast, would be me-
diated by the operations of a semantic memory system whose neural sub-
strate would be located in the temporoparietal associative areas of the left
hemisphere. That neuropathological changes in AD do not involve uni-
formly all cortical areas of the brain would account for the profile of per-
formance of these subjects: T e m p o r o p a r i e t a l association cortex is
consistently and severely damaged in AD (preventing normal activation of
lexical-semantic units), but striate and peristriate cortex of the occipital
lobes is relative spared (accounting for a normal priming effect in percep-
tual tasks; Brun and Englund, 1981). As noted above, however, available
neuropsychological data only partially support the hypothesis that repeti-
tion priming in visuoperceptual tasks is selectively spared in AD. Results
from one study that examined the perceptual identification of figures
showed an absolute decrement in priming of AD patients (Heindel et al.,
1990), while outcomes of a second study are ambiguous (Gabrieli et al.,
1992, exp. 2, described above). Further, normal priming for word identifi-
cations was observed in AD patients in only one laboratory (Keane,
Gabrieli, Fennema, et al., 1991; Keane,Gabrieli, Growden, and Corkin,
1991). In conclusion, additional neuropsychological evidence from AD (as
well as other neuropsychological populations) is needed to confirm this
view.

Concluding Remarks

The neuropsychological investigation of memory disorders in AD is

a field in evolution, in part because memory, deficits are the earliest and
most dramatic manifestations of the disease. Disorientation, combined with
an inability to recall day-to-day events, contribute to an apparent disinte-
gration of personality in AD patients. Consequently, there is hope that re-
search into biological and behavioral underpinnings of AD will contribute
Memory Deficits in Alzheimer's Disease 161

to a level of understanding sufficient to generate effective treatment regi-

mens, and perhaps even disease prevention.
Over the last decade, our knowledge of behavioral mechanisms and
neural systems critical for normal memory has advanced as a result of the
unique cooperation between cognitive psychology and neuropsychological
research on human amnesics. The availability of subjects affected by cir-
cumscribed memory disorders (e.g., pure amnesics), however, is very lim-
ited. AD, by contrast, has been compared to an epidemic condition
affecting about 5% of people over 65 years of age (Katzman, 1986). Data
deriving from neuropsychological research on AD subjects, therefore, can
integrate and enrich experimental work on pure amnesic patients. None-
theless, a great deal of research remains to be done before the complex
puzzle of memory loss in AD will be solved.

ACKNOWLEDGMENTS

This research was supported by N.A.T.O.-C.N.R. Grant 215.23/04 to

G. A. Carlesimo and by research funds from the U.S. Department of Vet-
eran Affairs, and NIAAA (AA07112), to M. Oscar-Berman.

REFERENCES

Albert, M. S., Butters, N., and Brandt, J. (1981). Patterns of remote memory in amnesic and
demented patients. Archives of Neurology 38: 495-500.
Albert, M. S., Butters, N., and Levin, J. (1979). Temporal gradients in the retrograde amnesia
of patients with alcoholic Korsakoff's disease. Archives of Neurology 36: 211.
Albert, M. S., Butters, N., and Levin, J. (1980). Memory for remote events in chronic alco-
holics and alcoholic Korsakoff patients. In Begleiter, H. (ed.), Biological Effects of Alcohol,
Plenum Press, New York, pp. 719-730.
Amaral, D. G. (1987). Memory: Anatomical organization of candidate brain regions. In Plum,
F. (ed.), The Handbook of Physiology, the Nervous System, V, Williams and Wilkins, Bal-
timore, MD, pp. 211-294.
American Psychiatric Association (1980). Diagnostic and Statistical Manual of Mental Disorders
(3rd ed.), Washington, DC.
Arendt, T., Bigl, V., Arendt, A., and Tennstedt, A. (1983). Loss of neurons in the nucleus
basalis of Meynert in Alzheimer's disease, paralysis agitans and Korsakoff's disease. Acta
Neuropathologica 61: 101-108.
Atkinson, R. C., and Shiffrin, R. M. (1968). Human memory: A proposed system and its
control processes. In Spence, K.W., and Spence, J. T. (eds.), The Psychology of Learning
and Motivation: Advances in Research and Theory, Vol. 2, Academic Press, New York,
pp. 89-195.
Atkinson, R. C., and Shiffrin, R. M. (1971). The control of short-term memory. Scientific
American 225: 82-90.
Backman, L., and Herlitz, A. (1990). The relationship between prior knowledge and face rec-
ognition memory in normal aging and Alzheimer's disease. Journal of Gerontology 45:
P94-P100.
162 Carlesimo and Oscar-Berman

Baddeley, A. D. (1976). Psychology of Memory, Basic Books, New York.

Baddeley, A. D. (1982). Amnesia: A minimal model and an interpretation. In Cermak, L. S.
(ed.), Human Memory and Amnesia, Lawrence Erlbaum Associates, Hillsdale, NJ, pp.
305-336.
Baddeley, A. D. (1986). Working Memory, Oxford University Press, London.
Baddeley, A. D., and Hitch, G. (1974). Working memory. In Bower, G. H. (ed.), The Psy-
chology of Learning and Motivation (Vol. 8), Academic Press, New York, pp. 47-90.
Baddeley, A. D., and Scott, D. (1971). Short-term forgetting in the absence of proactive in-
hibition. Quarterly Journal of Experimental Psychology 23: 275-283.
Baddeley, A. D., and Warrington, E. K. (1970). Amnesia and the distinction between long-
and short-term memory. Journal of Verbal Learning and Verbal Behavior 9: 176-189.
Baddeley, A., Logic, R., Bressi, S., Della Sala S., and Spinnler, H. (1986). Dementia and
working memory. The Quarterly Journal of Experimental Psychology 38A; 603-618.
Baso, A., Spinnler, H., Vallar, G., and Zanobio, M. E. (1982). Left hemisphere damage and
selective impairment of auditory verbal short-term memory. A case study. Neuropsycholo-
gia 20: 263-274.
Bayles, K. A., and Tomoeda, C. K. (1983). Confrontation naming in dementia. Brain and
Language 19: 98-114.
Beatty, W. W., Salmon, D. P., Butters, N., Heindel, W. C., and Granholm, E. L. (1988).
Retrograde amnesia in patients with Alzheimer's disease or Huntington's disease. Neu-
robiology of Aging 9: 181-186.
Becker, J. T., Boiler, F., Saxton, J., and McGonigle-Gibson, K. (1987). Normal rates of for-
getting of verbal and non-verbal material in Alzheimer's disease. Cortex 25: 59-72.
Branconnier, R. J., Cole, J. O., Spera, K. F., and Devitt, D. R. (1982). Recall and recognition
as diagnostic indices of malignant memory loss in senile dementia: A bayesian analysis.
Experimental Ageing Research 8: 189-193.
Brandt, J., Folstein, S. E., and Folstein, M. F. (1988). Differential cognitive impairment in
Alzheimer's disease and Huntington's disease. Annals of Neurology 23: 555-56t.
Brandt, J., Spencer, M., McSorley, P., and Folstein, M. F. (1988). Semantic activation and implicit
memory in Alzheimer disease. Alzheimer Disease and Associated Disorders 2: 112-119.
Brion, S., and Mikol, J. (1978). Atteinte du noyau lateral dorsal du thalamus et syndrome de
Korsakoff alcoolique. Journal of Neurological Science 38: 249-261.
Brown, 3. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of
Experimental Psychology 10: 12-21.
Brun, A., and Englund, E. (1981). Regional pattern of degeneration in Alzheimer's disease:
Neuronal loss and histopathological grading. Histopathology 5: 9-564.
Butters, N., and Miliotis, P. (1985). Amnesic disorders. In Heilman K. M., and Valenstein,
E. (eds.), Clinical Neuropsychotogy, Oxford University Press, New York, Oxford, pp. 403-452.
Butters, N., and Stuss, D. T. (1989). Diencephalic amnesia. In Boiler, E. F., and Grafman,
J. (eds.), Handbook of Neuropsychology, Vol. 3, Elsevier Science Publishers, New York,
pp. 107-148.
Butters, N., Albert, M. S., Sax, D. S., Miliotis, P., Nagode, J., and Sterste, A. (1983), The
effect of verbal elaborators on the pictorial memory of brain-damaged patients. Neurop-
sychologia 21: 307-323.
Cermak, L. S. (1972). Human Memory: Research and Theory, Ronald Press, New York.
Cermak, L. S. (1982). The long and short of it in amnesia, In Cermak, L. S. (ed.), Human
Memory and Amnesia, Lawrence Erlbaum Associates, Hillsdale, NJ, pp. 43-60.
Cermak, L. S., Talbot, N., Chandler, K., and Wolbarst, L. R. (1985). The perceptual priming
phenomenon in amnesia. Neuropsychologia 23: 615-622.
Cohen, N. J., and Corkin, S. (198I). The amnesic patient H. M.: Learning and retention of
a cognitive skill. Society for Neuroscience Abstracts 7: 571.
Cohen, N. J., and Squire, L. R. (1980). Preserved learning and retention of pattern-analyzing
skill in amnesia: Dissociation of knowing how and knowing that. Science 210: 207-210.
Cohen, N. J., and Squire, L. R. (1981). Retrograde amnesia and remote memory impairment.
Neuropsychologia 19: 337-356.
Memory Deficits in Alzheimer's Disease 163

Cohen, R. L. (1981). On the generality of some memory laws. Scandinavian Journal of Psy-
chology 22: 267-281.
Corkin, S. (1968). Acquisition of rooter skill after bilateral medical temporal-lobe excision.
NeuropsychoIogia 6: 255-265.
Corkin, S. (1982). Some relationship between global amnesias and the memory impairments
in Alzheimer's disease. In Corkin, S., Davis, K. L., Growdon, J. H., Usdin, E., Wurtman,
R. J. (eds.), Alzheimer's Disease: A Report of Progress. Aging (Vol. 19), Raven Press, New
York, pp. 149-164.
Corkin, S., Cohen, N. J., and Sagar, H. J. (1983). Memory for remote personal and public
events after bilateral medial temporal lobectomy. Society for Neuroscience Abstracts 9: 28.
Corkin, S., Growdon, J., Nissen, M. J., Huff, F. J., Freed, D. M., and Sagar, H. J. (1984).
Recent advances in the neuropsychological study of Alzheimer's disease. In Wurtman,
R. J., Corkin, S., and Growdon, J. (Eds.), Alzheimer's Disease: Advances in Basic Research
and Therapies. Proceedings of the Third Meeting of the International Study Group on the
Treatment of Memory Disorders Associated with Aging, Center for Brain Sciences and Me-
tabolism Trust, Cambridge, MA, pp. 75-94.
Craik, F. I. M. (1970). The fate of primary memory items in free recall. Journal of Verbal
Learning and Verbal Behavior. 9: 143-148.
Craik, F. I. M., and Lockhart, R. S. (1972). Levels of processing: A framework for memory
research. Journal of Verbal Learning and Verbal Behavior 11: 671-684.
Craik, F. I. M., and Tulving, E. (1975). Depth of processing and the retention of words in
episodic memory. Journal of Experimental Psychology: General 104: 268-294.
Cushman, L. A., Corno, P. G., Booth, H., and Caine, E. D. (1988). Cued recall and release
from proactive interference in Alzheimer's disease. Journal of Clinical and Experimental
Neuropsychology 10: 685-692.
Dale, H. C. A. (1967). Familiarity and free recall. Quarterly Journal of Experimental Psychology
19: 103-108.
Dall'Ora, P., Della Sala, S., and Spinnler, H. (1989). AutobiographicaJ memory, its impairment
in amnesic syndromes. Cortex 25: 197-217.
Damasio, A. R., Graff-Radford, N. R., Eslinger, P. J., Damasio, H., and Kassel, N. (1985).
Amnesia following basal forebrain lesions. Archives of Neurology 42: 263-271.
Davies, P., Katzman, R., and Terry, R. D. (1980). Reduced somatostatine-like immunoreac-
tivity in cerebral cortex from cases of Alzheimer disease and Alzheimer senile dementia.
Nature 288: 279-290.
Davis, P. E., and Mumford, S. J. (1984). Cued recall and nature of the memory disorder in
dementia. The British Journal of Psychiatry 144: 383-386.
Deese, J. (1959). Influence of inter item associative strength upon immediate free recall. Psy-
chological Reports 5: 305-312.
Delis, D. C., Massman, P. J., Butters, N., Salmon, D. P, Cermak, L. S., and Kramer, J. H.
(1991). Profiles of demented and amnesic patients on the California Verbal Learning
Test: Implications for the assessment of memory disorders. Psychological Assessment 3:
19-26.
Dick, M. B., Kean, M. L., and Sands, D. (t989a). Memory for internally generated words in
Alzheimer-type dementia: Breakdown in encoding and semantic memory. Brain and Cog-
nition 9: 88-108.
Dick, M. B., Kean, M. L., and Sands, D. (1989b). Memory for action events in Alzheimer-type
dementia: Further evidence of an encoding failure. Brab~ and Cognition 9: 71-87.
Diesfeldt, H. F. A. (1984). The importance of encoding instructions and retrieval cues in the
assessment of memory in senile dementia. Archives of Gerontology and Geriatrics 3: 51-57.
Eslinger, P. J., and Damasio, A. R (1986). Preserved motor learning in Alzheimer's disease:
Implications for anatomy and behavior. The Journal of Neuroscience 6: 3006-3009.
Freed, D. M., Corkin, S., and Cohen, N. J, (1987). Forgetting in H.M.: A second look. Neurop-
syehologia 25: 461-471.
Fuld, P. A., Katzman, R., Davies, P., and Terry, R. D. (1982). Intrusions as a sign of Alzheimer
dementia: Chemical and pathological verification. Annals of Neurology 11: 155-159.
164 Carlesimo and Oscar-Berman

Gabrieli, J. D. E. (1986). Memory subsystems of the human brain: Dissociations among learning
capacities in amnesia. Unpublished doctoral dissertation, Massachusetts Institute of Tech-
nology, Cambridge, MA.
Gabrieli, J. D. E. (1991). Differential effects of aging and age-related neurological diseases
on memory subsystems of the brain. In Boller, F., and Grafman, J., (eds.), The Handbook
of Neuropsychology, Elsevier/North Holland, Amsterdam.
Gabrieli, J. D. E., Keane, M. M., Stanger, B. Z., Kjelgaard, M. M., Growdon, J. H., and
Corkin, S. (1992). Dissociations among structural-perceptual, lexieal-semantic, and event-
fact memory systems in amnesic, Alzheimer's, and normal subjects. Manuscript submitted
for publication.
Gibson, A. J. (1981). A further analysis of memory loss in dementia and depression in the
elderly. British Journal of Clinical Psychology 20: 179-185.
Glanzer, M., and Cunitz, A. R. (1966). Two storage mechanisms in free recall. Journal of
Verbal Learning and Verbal Behavior 5: 351-360.
Gollin, E. S. (1960). Developmental studies of visual recognition of incomplete objects. Per-
ceptual and Motor Skills 11: 289-298.
Graf, P., Shimamura, A. P., and Squire, L. R. (1985). Priming across modalities and priming
across category levels: Extending the domain of preserved function in amnesia. Journal
of Experimental Psychology: Learning, Memory, and Cognition 11: 386-396.
Graf, P., Squire, L. R., and Mandler, G. (1984). The information that amnesic patients do
not forget. Journal of Experimental Psychology: Learning, Memory and Cognition 10(1):
164-178.
Granholm, E., and Butters, N. (1988). Associative encoding and retrieval and Alzheimer's
and Huntington's disease. Brain and Cognition 7: 335-347.
Grober, E., and Buschke, H. (1987). Genuine memory deficits in dementia. Developmental
Neuropsychology 3: 13-36.
Grosse, D. A., Wilson, R. S., and Fox, J. H. (1990). Preserved word-stem-completion priming of
semantically encoded information in Alzheimer's disease. Psychology and Aging 5: 304-306.
Grosse, D. A., Wilson, R. S., and Fox, J. H. (1991). Maze learning in Alzheimer's disease.
Brain and Cognition 15: 1-9.
Hardy, J., Adolfsson, R., Alafuzoff, I., Bucht, G., Marcusson, J., Nyberg, P., Perdahl, E.,
Wester, P., and Winblad, B. (1985). Transmitter deficit in Alzheimer's disease. A review.
Neurochemistry International 7: 454-563.
Hart, R. P., Kwentus, J. A., Taylor, J. R., and Harkins, S. W. (1987). Rate of forgetting in
dementia and depression. Journal of Consultant and Clinical Psychology 55: 101-105.
Hart, S., Smith, C. M., and Swash, M. (1986). Intrusion errors in Alzheimer's disease. British
Journal of Clinical Psychology 25: 149-150.
Heindel, W. C., Butters, N., and Salmon, D. P. (1988). Impaired learning of a motor skill in
patients with Huntington's disease. Behavioral Neuroscience 102: 141-147.
Heindel, W. C., Salmon, D. P., and Butters, N. (1990). Pictorial priming and cued recall in
Alzheimer's and Huntington's disease. Brain and Cognition 13: 282-295.
Heindel, W. C., Salmon, D. P., and Butters, N. (1991). The biasing of weight judgments in
Alzheimer's and Huntington's disease: A priming or programming phenomenon? Journal
of Clinical and Experimental Neuropsychology 13: 189-203.
Heindel, W. C., Salmon, D. P., Shults, C. W., Walicke, P. A., and Butters, N. (1989). Neurop-
sychologicai evidence for multiple implicit memory systems: A comparison of Alzheimer's
Huntington's, and Parkinson's disease patients. The Journal of Neuroscience 9: 582-587.
Helkala, E. L., Laulumaa, V., Soininen, H., and Riekkinen, P. J. (1989). Different error pat-
tern of episodic and semantic memory in Alzheimer's and Parkinson's disease with de-
mentia. Neuropsychologia 27: 1241-1248,
Herlitz, A., Adolfsson, R., Backman, L., and Nilsson, L. (1991). Cue utilization following dif-
ferent forms of encoding in mildly, moderately, and severely demented patients with
Alzheimer's disease. Brain and Cognition 15: 119-130.
Huff, F. J., Mack, L., Mahlmann, J., and Greenberg, S. (1988). A comparison of lexical-se-
mantie impairments in left hemisphere stroke and Alzheimer's disease. Brain and Lan-
guage 34: 262-278.
Memory Deficits in Alzheimer's Disease 165

Hultsch, D. F., and Dixon, R. A. (1983). The role of pre-experimental knowledge in test
processing in adulthood. Experimental Aging Research 9: 17-22.
Huppert, F. A., and Piercy, M. (1978). Dissociation between learning and remembering in
organic amnesia. Nature 275: 317-318.
Huppert, F. A., and Piercy, M. (1979). Normal and abnormal forgetting in amnesia: Effect
of locus of lesion. Cortex i5: 385-390.
Huppert, F. A , and Piercy, M. (1982). In search of the functional locus of amnesic syndromes.
In Cermak, L. S. (ed.), Human Memory and Amnesia, Lawrence Erlbaum Associates,
Hillsdale, NJ, pp. 123-138.
Hyman, B. T , Kromer, L. J., and Van Hoesen, G. W. (1987). Reinnervation of the hippo-
campal perforant pathway zone in Alzheimer's disease. Annals of Neurology 21: 259-267.
Hyman, B. T., Van Hoesen, G. W., and Damasio, A. R. (1990). Memory related neural systems
in Atzheimer's disease: An anatomic study. Neurology 40: 1721-1730.
Hyman, B. T., Van Hoesen, G. W., Damasio, A. R., and Barnes, C. L. (1984). Alzheimer's
disease: Cell specific pathology isolates the hippocampat formation. Science 225: 1168-1170.
Hyman, B. T., Van Hoesen, G. W., Kromer, L. J , and Damasio, A. R. (1986). Perforant pathway
changes and the memory, impairment of Alzheimer's disease. Annals of Neurology 20: 472-481.
Jacobs, D., Salmon, D. P. Troster, A. I., and Butters, N. (1990). Intrusion errors in the figural
memory of patients with Alzheimer's and Huntington's disease. Archives of Clinical
Neuropsychology 5: 49-57.
Jacobs, D., Troster, A. I., Butters, N., Salmon, D. P., and Cermak, L. (1990). Intrusion errors
on the visual reproduction test on the Wechsler Memory Scale and the Wechsler Memory
Scale-Revised: An analysis of demented and amnesic patients. The Clinical Neuropsycholo-
gist 4: 177-191.
Jacoby, L. L. (1982). Knowing and remembering: Some parallels in the behavior of Korsakoff
patients and normals. In Cermak, L. S. (ed.) Human Memory and Amnesia, Lawrence
Erlbaum Associates, Hillsdale, NJ, pp. 97-t22.
Jacoby, L. L. (1983). Remembering the data: Analyzing interactive processes in reading. Jour-
nal of Verbal Learning and Verbal Behavior 22: 485-508.
Jacoby, L. L., and Brooks, L. R., (t984). Nonanalytic cognition: Memory, perception, and
concept learning. In Bower, G. H. (ed.), The Psychology of Learning and Motivation (Vol.
18), Academic Press, New York, pp. 1-47.
Jacoby, L. L, and Dallas, M. (1981). On the relationship between autobiographical memory
and perceptual learning. Journal of Experimental Psychology: General 110: 306-340.
Karlsson, T., Backman, L., Herlitz, A., Nilsson, L. G., Winblad, B., and Osterlind, P. O. (1989).
Memory improvement at different stages of Alzheimer's disease. Neuropsychologia 27:
737-742.
Kaszniak, A. W., and Garron, D. C. (1979). Differential aspects of age and cerebral atrophy
upon span of immediate recall and paired associate learning in older patients with sus-
pected dementia. Cortex 15: 285-295.
Katzman, R. (1986). Alzheimer's disease. New England Journal of Medicine 314: 964-973.
Katzman, R., and Karasu, T. B. (1975). Differential diagnosis of dementia. In Fields, W. S.
(ed.), Neurological and Sensory Disorders in the Elderly, Stratton, pp. 103-134.
Keane, M. M., Gabrieli, J. D. E., Fennerna, A. C., Growdon, J. H , and Corkin, S. (1991).
Evidence for a dissociation between perceptual and conceptual priming in Alzheimer's
disease. Behavioral Neuroscience 105: 326-342.
Keane, M. M., Gabrieli, J. D. E., Growdon, J. H., and Corkin, S. (1991). Patients with Alzheimer's
disease show normal priming in perceptual identification of pseudowords. Society for Neu-
roscience Abstracts 17: 3.
Keppel, G., and Underwood, Bo J. (1962). Proactlve inhibition in short-term retention of single
items. Journal of Verbal Learning and Verbal Behavior. 1: 153-161.
Kirsner, K., Milech, D., and Stander, P. (1983). Common and morality-specific processes in
the mental lexicon. Memory and Cognition 11: 621-630.
Knopman, D. D., and Nissen, M. J. (1987). Implicit learning in patients with probable Alzheimer's
disease. Neurology 37: 784-788.
166 Carlesimo and Oscar-Berman

Kopetman, M. D. (1985). Rates of forgetting in Alzheimer-type dementia and Korsakoff's

syndrome. Neuropsychotogia 23: 623-638.
Kopelman, M. D. (1986). The cholinergic neurotransmitter system in human memory and
dementia: A review. Quarterly Journal of Experimental Psychology 38A: 535-573.
Kopelman, M. D. (1989). Remote and autobiographical memory, temporal context memory
and frontal atrophy in Korsakoff and Alzheimer patients. Neuropsychologia 27: 437-460.
Kopelman, M. D., and Corn, T. H. (1988). Cholinergic 'blockade' as a model for cholinergic
depletion. Brain 111: 1079-1110.
Maragos, W. F , Greenamyre, J. T., Penney, J. B., and Young, A. B. (1987). Glutamate dys-
function in Alzheimer's disease: An hypothesis. Trends in Neuroscience 10: 65-68.
Martin, A., Browers, P., Cox, C., and Fedio, P. (1985). On the nature of the verbal memory
deficit in Alzheimer's disease. Brain and Language 25: 323-341.
Martone, M., Butters, N., Payne, M., Becker, J. T., and Sax, D. S. (1984). Dissociations be-
tween skill learning and verbal recognition in amnesia and dementia. Arch&es of Neurology
41: 965-970.
McKhan, G., Drachman, D., Folstein, M., Katzman, R., Price, D., and Stadlan, E. M. (1984).
Clinical diagnosis of Alzheimer's disease: Report of the NINCDS-ADRDA work group
under the auspices of Department of Health and Human Services Task Force on Alzheimer's
disease. Neurology 34: 939-944.
Miller, E. (1971). On the nature of the memory disorder in presenile dementia. Neuropsy-
chologia 9: 75-81.
Miller, E. (1972). Efficiency of coding and short-term memory deficit in presenile dementia.
Neuropsychologia 10: 221-224.
Miller, E. (1973). Short- and long-term memory in patients with presenile dementia (Alzheimer's
disease). Psychological Medicine 3: 221-224.
Miller, E. (1975). Impaired recall and the memory disturbance in presenile dementia. British
Journal of Social and Clinical Psychology 14: 73-79.
Miller, E. (1977). Abnormal Ageing, Wiley, New York.
Miller, E. (1978). Retrieval from long-term memory in presenile dementia: two tests of an
hypothesis. British Journal of Social and Clin&al Psychology 17: 143-148.
Miller, E., and Lewis, P. (1977). Recognition memory in elderly patients with depression and
dementia: A signal detection analysis. Journal of Abnormal Psychology 86: 84-86,
Miller, G. A., and Selfridge, J. A. (1950). Verbal context and the free recall of meaningful
material. American Journal of Psychology 63: 176-185.
Mishkin, M. (1982). A memory system in the monkey. Philosophical Transactions of the Royal
Soc&ty (London B) 298: 85-95.
Mitchell, D. B., Hunt, R. S., and Schmitt, F. A. (1986). The generation effect and reality
monitoring: Evidence from dementia and normal aging. Journal of Gerontology 4: 79-84.
Morris, R. G. (1984). Dementia and the functioning of the articulatory loop system. Cognitive
Neuropsychology 1: 143-157.
Morris, R. G. (1986). Short-term forgetting in senile dementia of the Alzheimer's type. Cog-
nitive Neuropsychology 3: 77-97.
Morris, R. G. (1987a). Articulatory rehearsal in Alzheimer-type dementia. Brain and Language
30: 351-362.
Morris, R. G. (1987b). The effect of concurrent articulation on memory span in Alzheimer-
type dementia. British Journal of Clinical Psychology 26: 233-234.
Morris, R. G., Wheattey, J., and Britton, P. (1983). Retrieval from long-term memory in senile
dementia: Cued recall revisited. British Journal of Clinical Psychiatry 22: 141-142.
Moscovitch, M. (1982). A neuropsyehological approach to perception and memory in normal
and pathological aging. In Craik, F. I. M., and Trehub, S. (eds.), Aging and Cognitive
Processes, Plenum Press, New York, pp. 55-79.
Moss, M. B., Albert, M. S., Butters, N., and Payne, M. (1986). Differential patterns of memory
loss among patients with Alzheimer's disease, Huntington's disease and alcoholic Kor-
sakoff's syndrome. Archives of Neurology 43: 239-246.
Murdock, B. B. (1967). Auditory and visual stores in short-term memory. Acta Psychologica
27: 316-324.
Memory Deficits in Alzheimer's Disease 167

Nebes, R. D. (1989). Semantic memory in Alzheimer's disease. Psychological Bulletin 106:

377-394.
Nebes, R. D., Brady, C. B., and Huff, F. J. (1989). Automatic and attentional mechanisms
of semantic priming in Alzheimer's disease. Journal of Clinical and Experimental Neurop-
sychology 11: 219-230.
Nebes, R. D., Martin, D. C., and Horn, L. C. (1984). Sparing of semantic memory in Akheimer's
disease. Journal of Abnormal Psychology 93: 312-330.
Ober, B. A., and Shenaut, G. K. (1988). Lexical decision and priming in Alzheimer's disease.
Neuropsychologia 26: 273-286.
Ober, B. A., Koss, E., Friedland, R. P., and Dells, D. C. (1985). Processes of verbal memory
failure in Alzheimer-type dementia. Brain and Cognition 4: 90-103.
Orsini, A., Trojano, L., Chiaccio, L., and Grossi, D. (1888). Immediate memory span in de-
mentia. Perceptual and Motor Skills 67: 267-272.
Oscar-Berman, M. (1973). Hypothesis testing and focusing behavior during concept formation
by amnesic Korsakoff patients. Neuropsychologia 11: 191-198.
Oscar-Berman, M. (1980). Neuropsychological consequences of long-term chronic alcoholism.
American Scientist 68: 410-419.
Oscar-Berman, M., McNamara, P., and Freedman, M. (1991). Delayed response tasks: Par-
allels between experimental ablation studies and findings in patients with frontal lesions.
In Levin, H. S., Eisenberg, H. M, and Benton, A. L. (eds.), Frontal Lobe Function and
Injury, Oxford University Press, New York.
Paivio, A. (1969). Mental imagery in associative learning and memory. Psychological Review
76: 241-263.
Paivio, A. (1971). Imagery and Verbal Processes, Holt, Rinehart & Winston, New York.
Partridge, F. M., Knight, R. G., and Feehan, M. (1990). Direct nd indirect memory perform-
ance in patients with senile dementia. Psychological Medicine 20: 111-118.
Pearson, R. C. A., Esiri, M. M., Hiorns, R. W., Wilcock, G. K., and Powell, T. P. S. (1985).
Anatomical correlates of the distribution of the pathological changes in the neocortex in
Alzheimer disease. Proceedings of the National Academy of Science USA 82: 4531-4534.
Perry, E. K., and Perry R. H. (1986). A review of neuropathological and neurochemical cor-
relates of the Alzheimer's disease. Danish Medical Bulletin 32, (suppl. 1): 27-34.
Petersen, S. E., Fox, P. T. Snyder, A. Z., and Raichle, M. E. (I990). Activation of extrastriate
and frontal cortical areas by visual words and word-like stimuli. Science 249: 1041-i044.
Peterson, L. R., and Peterson, M. J. (1959). Short-term retention of individual verbal items.
Journal of Experimental Psychology 58: 193-198.
Richardson-Klavehn, A., and Bjork, R, A. (1988). Measures of memory. Annual Review of
Psychology 39: 475-543.
Rissenberg, M., and Glanzer, M. (1986). Picture superiority in free recall: The effect of normal
aging and primary degenerative dementia. Journal of Gerontology 41: 64-71.
Rissenberg, M., and Glanzer, M. (1987). Free recall and word finding ability in normal aging
and senile dementia of the Alzheimer's type: the effect of item concreteness. Journal of
Gerontology 42: 318-322.
Rocdiger, H. L. (1990). Implicit memory. Retention without remembering. American Psycholo-
gist 45: 1043-1056.
Roediger, H. L, Weldon, M. S., and Challis, B. H. (i989). Explaining dissociations between
implicit and explicit measures of retention: A processing account. In Roediger, H. L.,
and Craik, F. I. M. (eds.), Varieties of Memory and Consciousness: Essays in Honour of
Endel Tulving, Erlbaum, Hillsdale, NJ, pp. 3-41.
Rossor, M. N., Iversen, L. L., Reynolds, G. P., Mountjoy, C. Q., and Roth, M. (1984). Neuro-
chemical characteristics of early and late onset types of Alzheimer's disease. British .?vtedi-
cal Journal 288: 961-964.
Sagar, H. J., Cohen, N. J., Sullivan, E. V., Corkin, S., and Growdon, J. H. (1988). Remote
memory function in Alzheimer's disease and Parkinson's disease. Brain 111: 525-539.
Salmon, D. P., Granholm, E., McCullough, D., Butters, N., and Grant, I. (1989). Recognition
memory span mildly and moderately demented patients with Alzheimer's disease. Journal
of Clinical and Experimental Neuropsychology 11: 429-443.
168 Carleslmo and Oscar-Berman

Salmon, D. P., Shimamura, A. P., Butters, N., and Smith, S. (1988). Lexical and semantic
priming deficits in patients with Alzheimer's disease. Journal of Clinical and Experimental
Neuropsychology 10: 477-494.
Schacter, D. L. (1990). Perceptual representation systems and implicit memory: Toward a
resolution of the multiple memory systems debate. In Diamond A. (ed.), Developmental
and Neural Bases of Higher Cognitive Function, New York Academy of Sciences, New
York, pp. 543-571.
Schultz, K. A., Schmitt, F. A., Logue, P. E., and Rubin, D. C. (1986). Unit analysis of prose
memory in clinical and elderly populations. Developmental Neuropsychology 2: 77-87.
Scoville, W. B., and Milner, B. (1957). Loss of recent memory after bilateral hippocampal
lesions. Journal of Neurology, Neurosurgery and Psychiatry 20: 1t-2I.
Shallice, T., and Warrington, E. K. (1970). Independent functioning of verbal memory stores:
A neuropsychotogical study. Quarterly Journal of Experimental Psychology 22: 261-273.
Shiffrin, R. M. (1970). Forgetting: Trace erosion or retrieval failure? Science 168: 1601-1603.
Shimamura, A. P. (1986). Priming effects in amnesia: Evidence for a dissociable memory func-
tion. The Quarterly Journal of Experimental Psychology 38A: 619-644.
Shimamura, A. P. (1990). Forms of memory: Issues and directions. In McCaugh, J. L., We-
inberger, N. M., and Lynch, G. (eds.), Brain Organization and Memory. Cells, Systems,
and Circuits, Oxford University Press, New York, Oxford, pp. 159-176.
Shimamura, A. P., and Squire, L. R. (1984). Paired-associate learning and priming effects
in amnesia: A neuropsychological study. Journal of Experimental Psychology: General 113:
556-570.
Shimamura, A. P., and Squire, L. R. (1987). A neuropsychological study of fact memory
and source amnesia. Journal of Experimental Psychology: Learning Memory and Cognition
13: 464-473.
Shimamura, A. P., Salmon, D. P., Squire, L. R., and Butters, N. (1987). Memory dysfunction
and word priming in dementia and amnesia. Behavioral Neuroscience 101: 347-351.
Slamecka, N. J., and Graf, P. (1978). The generation effect: Delineation of a phenomenon.
Journal of Experimental Psychology: Human Learning and Memory 4: 592-604.
Spinnler, H., Della Sala, S., Bandera, R., and Baddeley, A. (1988). Dementia, aging, and the
structure of human memory. Cognitive Neuropsychology 5: 193-211.
Squire, L. R. (1981). Two forms of human amnesia: An analysis of forgetting. Journal of
Neurosc&nce 1: 635-640.
Squire, L. R. (1982). Comparison between forms of amnesia: Some deficits are unique to
Korsakoff's syndrome. Journal of Experimental Psychology: Human Learning and Memory
8: 560-571.
Squire, L. R. (1987). Memory and Brain, Oxford University Press, New York, Oxford.
Squire, L. R., and Cohen, N. J. (1982). Remote memory, retrograde amnesia and the neurop-
sychology of memory. In Cermak, L. S. (ed.), Human Memory and Amnesia, Lawrence
Erlbaum Associates, Hillsdale, N J, pp. 275-303.
Strauss, M. E., Weingartner, H., and Thompson, K. (1985). Remembering words and how
often they occurred in memory-impaired patients. Memory and Cognition 13: 507-510.
Terry, R. D., Katzman, R. (1984). Senile dementia of the Alzheimer type. Annals of Neurology
14: 497-506.
Thomson, D. M., and Tulving, E. (1970). Associative encoding and retrieval: weak and strong
cues. Journal of Experimental Psychology 86: 255-262.
Tomlinson, B. E., Blessed, G., and Roth, M. (1970). Observations on the brain of demented
old people. Journal of Neurological Sciences 11: 205-242.
Tulving, E. (1968). Theoretical issues in free recall. In Dixon, T. R., Horton, D. L. (eds.),
l"erbal Behavior and General Behavior Theory, Prentice-Hall, Englewood Cliffs, NJ.
Tutvin, E. (1983). Elements of Episodic Memory, Oxford University Press, Oxford.
Tulving, E. (1972). Episodic and semantic memory. In Tulving, E., Donaldson, W. (eds.),
Organization of Memory, Academic Press, New York, pp. 381-403.
Tulving, E., and Schacter, D. L. (1990). Priming and human memory systems. Science 247:
301-306.
Memory Deficits in Aizheimer's Disease 169

VaUar, G., and Baddeley, A. D. (1984). Fractionation of working memory: neuropsychological

evidence for a phonological shortqerm store. Journal of Verbal Learning and Verbal Be-
havior 23: 151-161.
Van Hoesen, G. W., and Damasio, A. R. (1987). Neural correlates of cognitive impairment
in Alzheimer's disease. In Plum, F. (ed.), The Handbook of Physiology, The Nervous Sys-
tem, Vol. V, Williams and Wilkins, Baltimore, MD, pp. 871-898.
Victor, M., Adams, R. D., and Collins, G. H. (1971). The Wernicke-KorsakoffSyndrome, Davis,
Philadelphia, PA.
Von Cramon, D. Y., Hebel, N., and Schuri, U. (1985). A contribution to the anatomical basis
of thalamic amnesia. Brain 108: 993-1008.
Wallin, A., Gottfries, C. G. (1990). Biochemical substrates in normal aging and Alzheimer's
disease. Pharmacopsychiatry 23(Suppl. 2): 37-43.
Warrington, E. K., and Weiskrantz, L. (1968). A new method of testing long-term retention
with special reference to amnesic patients. Nature 217: 972-974.
Warrington, E. K., and Weiskrantz, L. (1970). Amnesic syndrome: Consolidation or retrieval?
Nature 228: 628-630.
Weingartner, H., Kaye, W., Smallberg, 8. A., Ebert, M. H., Gillan, J. C., and Sitaram, N.
(1981). Memory failures in progressive idiopathic dementia. Journal of Abnormal Psychol-
ogy 90: 187-196.
Weingartner, H., Grafman, J., Boutelle, W., Kaye, W., and Martin, P. R. (1983). Forms of
memory failure. Science 221: 380-382.
Weiskrantz, L. (1982). Comparative aspects of studies in amnesia. Philosophical Transactions
of the Royal Society (London) B 298: 97-109.
Weiskrantz, L., and Warrington, E. K. (1979). Conditioning in amnesic patients. Neurop~y-
ehologia 17: 187-194.
Welsh, K., Butters, N., Hughes, J., Mohs, R., and Heyman, A. (1991). Detection of abnormal
memory decline in mild cases of Alzheimer's disease using CERAD neuropsychological
measures. Archives of Neurology 48: 278-281.
Whitehouse, P. J., Price, D. L., Struble, R. G., Clark, A. W., Coyle, J. T., and DeLong, M.
R. (1982). Alzheimer's disease and senile dementia: Loss of neurons in the basal fore-
brain. Science 215: 1237-1239.
Wilson, R. S., Kaszniak, A. W., and Fox, J. H. (1981). Remote memory in senile dementia.
Cortex 17: 41-48.
Wilson, R. $., Kaszniak, A. W., Bacon, L. D., and Fox, J. H. (1982). Facial recognition memory
in dementia. Cortex 18: 329-336.
Wilson, R. S., Bacon, L. D., Fox, J. H., and Kaszniak, A. W. (1983). Primary memory and
secondary memory in dementia of the Alzheimer type. Journal of Clinical Neuropsychology
5: 337-344.
Wilson, R. S., Bacon, L. D., Kramer, R. L., Fox, J. H., and Kaszniak, A. W. (1983). Word
frequency effect and recognition memory in dementia of the Alzheimer type. Journal of
Clinical Neuropsyehology 5: 97-104.