Professional Documents
Culture Documents
Developmental
Neuropsychology
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/hdvn20
To cite this article: Ellen Grober & Herman Buschke (1987) Genuine memory
deficits in dementia, Developmental Neuropsychology, 3:1, 13-36, DOI:
10.1080/87565648709540361
Taylor & Francis makes every effort to ensure the accuracy of all
the information (the “Content”) contained in the publications on our
platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy,
completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of
the authors, and are not the views of or endorsed by Taylor & Francis.
The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and
Francis shall not be liable for any losses, actions, claims, proceedings,
demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in
relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any form
to anyone is expressly forbidden. Terms & Conditions of access and use
can be found at http://www.tandfonline.com/page/terms-and-conditions
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
DEVELOPMENTAL NEUROPSYCHOLOGY, 5(1), 13-36
Copyright © 1987, Lawrence Erlbaum Associates, Inc.
EMPIRICAL CONTRIBUTIONS
Requests for reprints should be sent to Ellen Grober, Saul R. Korey Department of Neurol-
ogy, Albert Einstein College of Medicine, Van Etten Hospital, Room 319, 1300 Morris Park Av-
enue, New York, NY 10461.
14 GROBER AND BUSCHKE
tegies, or to other factors that limit learning and memory in the elderly
(Botwinick, 1984; Buschke, 1984; Perlmutter & Mitchell, 1982; Poon, 1985;
Rabinowitz, Craik, & Ackerman, 1982). Apparent memory deficits are
secondary memory deficits because they are due to disruptions in processes
involved in other cognitive activities that are necessary but not sufficient
for learning and memory. Apparent memory deficits can be ameliorated by
the proper control of processing, which induces patients to process the infor-
mation in a semantically appropriate way and provides them with effective
cues for retrieval (Arenberg, 1980; Craik, 1983, 1984; Craik & Byrd, 1982;
Craik & Rabinowitz, 1984;Eysenck, 1974; Hultsch, 1971 ;Macht& Buschke,
1984; Perlmutter, 1978, 1979; Perlmutter & Mitchell, 1982; Poon, 1985;
Rabinowitz & Ackerman, 1982; Smith, 1977,1980; Till & Walsh, 1980; Treat
& Reese, 1976).
Genuine memory deficits, unlike apparent memory deficits, persist even
after the intended processing has been completed. They are primary memory
deficits because they are due to disruptions in specific memory processes such
as encoding and retrieval rather than on other cognitive processes involved in
the initial perception of the items. Genuine memory deficits in the elderly are
usually associated with dementia.
The use of the term apparent memory deficit is not meant to deny that in
real life many elderly people have a memory deficit; rather, it emphasizes
that their memory deficit is secondary to impairment of other cognitive pro-
cesses. The term genuine memory deficit is meant to capture the idea of a
memory deficit that persists despite a person's having carried out effective
encoding and retrieval activities. Because the terms primary memory and sec-
ondary memory already refer in the psychological literature to a distinction
different from the one intended here, the terms genuine memory deficit and
apparent memory deficit have been adopted.
To show that the memory deficits of some elderly persons are due to
genuine memory deficits, it is necessary to show that their best performance
on a memory test is impaired even when they have carried out the same spe-
cific effective processing for encoding and retrieval that elderly persons with
preserved memory do (Buschke, 1984). This means that cognitive processing
must be controlled in a way that induces the elderly to process the to-be-
remembered items in the same way; it also shows that the specified opera-
GENUINE MEMORY DEFICITS IN DEMENTIA 15
tions have been carried out. In addition to control of processing, cued recall
will be needed to obtain maximum recall, because free recall alone will not re-
sult in retrieval of all items available in memory (Tulving & Pearlstone,
1966). The effectiveness of cued recall, in turn, depends on semantic proces-
sing appropriate for encoding specificity (e.g., Tulving, 1979) and on the
coordination of encoding and retrieval (e.g., Schacter & Tulving, 1982).
Research in cognitive psychology by Tulving and his colleagues among
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
others has clarified the conditions needed for the accurate assessment
of memory (e.g., Schacter & Tulving, 1982; Thomson & Tulving, 1970;
Tulving, 1974; Tulving & Osier, 1968; Tulving & Pearlstone, 1966; Tulving &
Thomson, 1973). Based on their findings, a procedure has been developed
that coordinates encoding and retrieval for effective cued recall (Buschke,
1984, in press). It involves a search procedure in which the items to be learned
are placed in front of the participant, who is asked to search for and identify
each item (e.g., "elephant") when its semantic cue is given (e.g., "animal").
After all items have been successfully identified, free recall is attempted, fol-
lowed by cued recall of any of the items not retrieved by free recall. The
search procedure requires that each item be identified on the basis of its cue,
thus circumventing the failure of some elderly persons to spontaneously
process material deeply or to engage in elaborative verbal or visual processing
(Eysenck & Eysenck, 1979; Smith & Fullerton, 1981; Waugh & Barr, 1982).
The correct identification of each item shows that the intended processing
was carried out and that the item was appreciated well enough for an ade-
quate representation of it to be formed. Such effective encoding, however, is
not sufficient for maximizing recall. Free recall must be supplemented by
cued recall to reveal learning not shown by free recall (Tulving & Pearlstone,
1966). The use of effective cued recall is especially important in the case of the
elderly who retrieve even less information than young normal individuals un-
der standard free recall instructions (Hultsch, 1971; Schonfield & Robertson,
1966; Smith, 1977). Controlled learning coordinates the processing of items
during both encoding and retrieval in a way that maximizes the effectiveness
of cued recall for items that cannot be free-recalled.
To evaluate the efficacy of controlled learning in distinguishing between
genuine and apparent memory deficits in the aged, demented and non-
demented elderly persons performed a controlled learning task that consisted
of searching for and identifying 16 items and of then completing three trials
of free and cued recall. This was followed immediately by a recognition test.
Free recall, cued recall, and recognition measures were computed and
analyzed to evaluate four specific predictions: (a) that free recall grossly
underestimates the amount of information stored by demented patients as
well as by nondemented elderly, (b) that nondemented elderly store and relia-
bly retrieve all the information presented for learning, (c) that demented pa-
16 GROBER AND BUSCHKE
tients store less information then nondemented elderly and retrieve what they
do store unreliably, and (d) that "total recall," the sum of free and cued recall,
is a better predictor of dementia than either free recall alone or recognition.
METHOD
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
Participants
TABLE 1
Individual Participant Data for Normal Elderly
oo TABLE 2
Individual Participant Data for Normal Elderly
ent on the Blessed test, /(48) = 12.72, p < .05, ontheDRS, t(45) = 9.18,/J
< .05,andonVIQ,/(46) = 6.66,p < .05. Premorbid VIQ, estimated by the
ability to read irregular words (Nelson & O'Connell, 1978), was well above
average for both groups (122 vs. 115 for nondemented and demented, respec-
tively), but was significantly higher for nondemented participants, f(43) =
2.72,p < .05.
Thirteen of the patients in the demented group were diagnosed as having
Alzheimer's disease (AD) according to the criteria established by a National
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
Stimuli
Two items were selected from each of 16 categories in the Battig and
Montague (1969) norms. Items of comparable difficulty were chosen so that
they were not so common that a correct response by guessing was likely and
not so rare that participants failed to identify them when given cues. One in-
stance from each category was randomly assigned to one of the two 16-item
lists. The mean generation frequency of the items comprising each list was
similar (14.5 vs 11.1), f(30) = .93, ns. Sixteen items were used because in a
previous study (Buschke, 1984), four mildly impaired AD patients succeeded
in recalling all the items from a 12-item list when cued recall was added to free
recall. The materials used in the current search procedure consisted of the
name of each item printed boldly above a picture taken from Snodgrass and
Vanderwart (1980). Labeled pictures were used because some demented par-
ticipants were unable to name the pictures otherwise. Four items were pre-
sented on an 8 Vi- X 11-in. card, one picture in each quadrant. The items on a
card were semantically unrelated. The stimuli for the recognition test
consisted of equal numbers of targets and related and unrelated foils. Be-
cause the same categories were used in the two lists, targets from one list com-
prised the related foils in the other, and vice versa. Unrelated foils were the
same for both lists and consisted of items in the same frequency range as the
targets but drawn from 16 different categories.
20 GROBER AND BUSCHKE
Procedure
card and point to and name each item (e.g., grapes) when its category cue
(e.g., fruit) was given verbally. After all 4 items were identified by naming
each correctly, the card was removed, and immediate verbal cued recall of
just those 4 items was tested, in the order of the search, by reading each cue to
the participant. If the participant failed to recall an item in response to its cue
at this point, the card was represented, the search was performed again, and
the participant was given another opportunity at immediate cued recall. If
the participant was still unable to recall the item, the experimenter verbally
represented the cue-item pair, and a final trial of immediate cued recall was
attempted.
The use of immediate cued recall during the study phase was introduced
because during pilot testing, some demented patients who showed little or no
cued recall after the search alone did show some cued recall when immediate
cued recall was added. Immediate cued recall serves several functions: It pro-
vides an initial successful trial of cued recall that, in turn, may act like an ori-
enting task effectively guiding retrieval (Rabinowitz & Craik, 1986); it pro-
vides retrieval practice (Craik & Rabinowitz, 1984; Landauer&Bjork, 1978);
it adds semantic information to the memory trace (Rabinowitz & Craik,
1986; Whitten, 1978); it confirms correct initial encoding; and it demon-
strates that the participant understands the task. Once immediate cued recall
for a group of 4 items was complete, the next set of items was presented for
study.
The study phase was followed by three trials each of recall preceded by 20
sec of counting backward in order to obtain recall from secondary memory.
Each recall trial consisted of two parts. First, an extended period of time up
to 2 min was provided for participants to free-recall as many items as possi-
ble. Next, cues were provided for those items not retrieved by free recall.
When a participant failed to retrieve the item with the cue, the examiner re-
minded the participant of the missed item, which the participant then
repeated.
Immediately following the last recall trial, a yes-no recognition test was
administered. A word was presented in bold print under computer control.
Participants were instructed (a) to decide on each trial whether the word was
in the memory list and (b) to indicate their response verbally to the examiner.
GENUINE MEMORY DEFICITS IN DEMENTIA 21
RESULTS
The data obtained from free recall, total recall (sum of free and cued recall),
and recognition were analyzed with two objectives in mind: to evaluate the
superiority of total recall over free recall as a measure of learning and mem-
ory and to test the utility of total recall in distinguishing between apparent
memory deficits in normal elderly and genuine memory deficits in dementia.
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
retrieved 15.2 items, and demented patients consistently retrieved 4.5 items
per trial. Because free recall revealed only 25% of the items that patients
managed to encode and retrieve, the use of free recall alone would have
provided a misleading and inaccurate description of their memory capacity.
Although the addition of cued recall provides an estimate of a participant's
best performance, cued recall obtained from demented patients in the ab-
sence of controlled processing and coordination of encoding and retrieval
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
will not necessarily result in improvement over free recall (Caine, Ebert, &
Weingartner, 1977).
Not only does free recall provide a less accurate estimate of learning and
memory than effective cued recall, it may also be less useful for distinguish-
ing between elderly persons with and without dementia. This is illustrated
in Panels A through C of Figure 2. Panel A presents the scores of a non-
demented participant (Participant 23 in Table 1) selected due to his unusually
good free recall. Panel B shows the scores of a nondemented participant
(Participant 13 in Table 1) whose free recall is lower and more representative
of the group as a whole. Despite very different levels of free recall, both par-
ticipants showed perfect total recall. Panel C shows the performance of a de-
mented patient (Participant 45 in Table 2) whose free recall is nearly identical
with that of the participant shown in Panel B but whose total recall is signifi-
cantly impaired. These comparisons suggest that free recall may not be the
most appropriate indicator of dementia status because, on the one hand,
some normal elderly have decreased free recall and, on the other hand, some
demented patients have free recall that is within normal limits because the
range of free recall in nondemented elderly is so large. Total recall may be a
better indicator of dementia status than free recall because normal elderly,
but not demented patients, display nearly perfect total recall of all 16 items
regardless of their free recall.
FIGURE 2 Verbal learning shown by free recall (open circles) and total recall (closed
circles) by a nondemented 89-year-old man (Panel A), a nondemented 87-year-old
woman (Panel B), and an 86-year-old woman with AD (Panel C).
GENUINE MEMORY DEFICITS IN DEMENTIA 23
was a main effect of trials, F(2, 96) = 6.27, p < .01, and an interaction be-
tween trials and group, F(2, 96) = 16.85, p < .01. Post hoc comparisons
using Scheffe's test showed that nondemented participants retrieved more
items on Trials 2 and 3 than they did on Trial 1, indicating that learning did
occur after Trial 1. Additional post hoc comparisons showed that demented
patients displayed the same depressed level of free recall across all trials, indi-
cating the absence of learning.
Total recall provided a higher estimate of the amount of information each
group was able to encode and retrieve. The number of items in total recall
was significantly higher for nondemented than demented participants (15.7
vs. 7.3), F(l,48) = 86.7, p < .01. There was a marginal effect of trials, F(2,
96) = 3.77,p < .05, and no interaction between group and trials (F< 1). Al-
though the possibility of learning by normal elderly participants was pre-
cluded by nearly perfect recall from Trial 1, their recall of nearly all items
shows that this procedure results in extremely effective cued recall. Although
demented patients were not limited by ceiling effects, they dispayed no evi-
dence of learning across trials.
Not only is the total recall of demented patients depressed, but it is also
very inconsistent. Although they recalled as many as 10 different items over
trials, they recalled only 4.5 items consistently on all three trials.
To evaluate the relative contribution of free and cued recall to overall per-
formance, two measures were computed for each participant: the proportion
of items retrieved by free recall over trials and the proportion of the re-
maining items retrieved by cued recall. Nondemented participlants retrieved
52% of the items by free recall and retrieved 96% of those remaining by cued
recall. Demented patients, in contrast, retrieved 10% of the items by free re-
call and succeeded in retrieving only 42% of the remaining items by cued re-
call. These differences were confirmed by ANOVA that revealed a group ef-
fect, F(l, 48) = 143.6,/? < .01, an effect due to free versus cued recall, F(l,
48) = 248.8, p < .01, and an interaction between group and type of retrieval,
F(l,48) = 7.85,/? < .01. Demented patients retrieved proportionately fewer
of the nonrecalled items by cued recall than nondemented participants did,
showing that effective cued recall can be used to identify genuine memory
deficits in the elderly.
In an earlier report by Buschke (1984), four mild AD patients showed cued
24 GROBER AND BUSCHKE
Factor Analysis
TABLE 3
Varimax-Rotated Component Loadings for Free and Total Recall Measures Computed
Separately for Patients With Dementia and for Normal Elderly
mented group. Varimax-rotated factor loadings are given in the left half of
Table 3. Factor 1 was defined by the free recall variables. Total recall varia-
bles loaded on Factor 2 along with the Blessed score. A similar separation be-
tween free and total recall emerged in the analysis of the data from the nor-
mal elderly group. The factor loadings for the three factors that emerged are
given in the right half of Table 3. Factor 1 was also defined by free recall mea-
sures, whereas total recall measures were now split across two factors. TR1
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
and CTR loaded on Factor 2, presumably reflecting the fact that in the case
of the normal elderly the value of CTR was determined by TR1. The Blessed
score also loaded on Factor 2. Factor 3 was defined by TR2 and TR3, pre-
sumably because both represent additional learning beyond Trial 1.
The separation of free recall and total recall measures in the factor analysis
suggests that the two variables measure different aspects of learning and
memory. Furthermore, given that the Blessed score is an index of dementia
status and that impaired total recall is an index of genuine memory impair-
ment, their association in the factor structure of both groups suggests that
such genuine memory impairment is the kind of memory deficit observed in
dementia. The same factor structure has emerged in a cross-validation study
(Grober, Buschke, & Crystal, 1987).
Recognition
The first three columns of Table 4 present the proportion of correctly classi-
fied targets (hit rate) and the proportion of related and unrelated foils incor-
rectly classified as targets (false alarm rates) for normal elderly and demented
patients. The normal elderly group recognized more targets and incorrectly
classified fewer foils than demented patients. Apart from these differences,
both groups made more errors by failing to recognize targets than by incor-
rectly classifying foils as targets and made more false alarms when the foil be-
longed to the same category as the target than when the foil belonged to a dif-
ferent category.
Using hit and false alarm rates, a bias-free measure of sensitivity (d) was
computed separately for related and unrelated foils for each participant by
the application of signal detection theory (Swets, 1973). These measures are
shown in the last column of Table 4. An ANOVA performed on the d values
indicated that recognition was better for the nondemented than the demented
group (3.3 versus 1.7), F(l, 43) = 27.5, p < .01. and better for unrelated
than related foils (2.6 versus 2.5), F(l, 43) = 7.61, p < .01. The interaction
between group and relatedness was not significant, F(l, 43) = 2.29, p > .10.
In an additional analysis, recognition of specific items was compared to to-
tal recall of the same items on the last recall trial. Of the 11 nondemented
participants who made any errors at all on recognition, all 11 displayed rec-
ognition failure of one word that they had recalled. Only 2 nondemented par-
26 GROBER AND BUSCHKE
TABLE 4
Recognition Scores for Normal Elderly and for Patients With Dementia
False Alarms a
Hits Related Unrelated Related Unrelated
Normal elderly .93 .02 .01 3.29 3.36
(.10) (.04) (.03) (.61) (.55)
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
ticipants failed to recall an item that they had recognized. Demented patients
made both types of errors more often. That participants failed to recognize
words they had recalled suggests that recognition may not be the best indica-
tor of those items that have been stored.
Prediction
the free recall measure. An optimum cut score or referent value, computed
for each measure to separate cases from noncases, was determined by
maximizing the sum of hits and correct rejections. Maximizing the sum of
sensitivity and specificity resulted in the same cut scores.
Table 5 contains the sensitivity, specificity, and positive and negative pre-
dictive values at the optimum cuts for each measure. Sensitivity and specific-
ity were highest for CRT and sumTR, intermediate for sumFR, and lowest
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
for d'R. To evaluate the utility of the measures for discriminating between
demented and nondemented persons in an unselected population, it is neces-
sary to compute their predictive values (Galen & Gambino, 1975). Total re-
call measures performed perfectly in identifying persons with dementia (true
positives), sumFR performed less well, and d'R performed least well of all.
The likelihood of identifying persons without dementia (true negatives) was
highest for total recall measures and lower for sumFR and d'R. The use of ei-
ther of the total recall measures resulted in the misclassification of only 1 de-
mented patient, whereas the free recall measure misclassified 2 normal eld-
erly participants and 3 demented patients. Recognition misclassified more
people than free recall did, including 10 of the 25 nondemented participants
and 6 of the 25 demented patients.
The relatively low predictive value of d'R in the present study differs from
the results of Branconnier et al. (1982) in which d' was shown to have sub-
stantial predictive value due primarily to a high false alarm rate (.48) among
demented patients. Unlike the present study, the memory items in the
Branconnier study consisted of instances from a single semantic category. To
eliminate the possibility that recognition of targets could be made on the ba-
sis of category information alone, the foils consisted of other instances from
the same category. Distinguishing targets from foils under these conditions
requires that sufficient item-specific information be encoded about each tar-
get to avoid confusing them with foils. Because demented patients are limited
in their ability to generate distinctive encodings (Martin & Fedio, 1983;
Wilson, Bacon, Kramer, Fox, & Kaszniak, 1983), a high false alarm rate
would be expected when targets and foils belong to the same semantic cate-
gory, accounting for the results obtained by Branconnier et al. (1982).
TABLE 5
Sensitivity, Specificity, and Predictive Value of Total Recall, Free Recall,
and Recognition Measures
DISCUSSION
The objective of this study was to identify and accurately measure memory
deficits in demented patients. Because learning and memory depend on other
cognitive processes that are often impaired in dementia, accurate assessment
requires the use of controlled learning conditions that induce appropriate se-
mantic processing and provide for effective cued recall. It is only under these
conditions that genuine memory deficits due to impairment of specific mem-
ory processes can be distinguished from apparent memory deficits due to im-
pairment of other cognitive processes.
Description of Deficits
The first major finding was that cued recall provides a more accurate meas-
ure of memory than free recall does because it reveals learning not shown by
free recall. This was true for both normal elderly and demented patients. The
advantage of cued recall even without controlled learning was first demon-
strated by Tulving and Pearlstone (1966) for young adults who recalled
nearly twice as many items by cued recall than by free recall alone. A smaller
cued recall advantage has been documented for normal elderly adults
(Drachman&Leavitt, 1972; Laurence, 1967; Miller, 1975; Perlmutter, 1979;
Rabinowitz & Ackerman, 1982; Smith, 1977). In the absence of recall of all
or nearly all items as found in the present study, it is not clear whether the
items that were not retrieved by cued recall were not available in memory or
whether the experimental conditions were not adequate to obtain cued recall
of other items available in memory. Maximally effective cued recall is needed
for complete and accurate assessment of learning and memory. When condi-
tions designed to promote effective cued recall were used in the present study,
normal elderly participants were able to recall all or nearly all 16 items pre-
GENUINE MEMORY DEFICITS IN DEMENTIA 29
sented for learning even though they were able to free-recall only 50% of
them. This is an indication of how effective cued recall can be when done in
accordance with the principles developed by Tulving and his colleagues.
Cued recall after controlled learning is so effective that some nondemented
elderly participants from the present study showed intact cued recall of a
48-item list (Buschke & Grober, 1986). Other nondemented elderly partici-
pants who had intact cued recall of 16 items showed decreased cued recall of
48 items. This shows that genuine memory deficits not due to other cognitive
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
deficits can also occur in nondemented elderly and provides an empirical ba-
sis for the diagnosis of "age-associated memory impairment" (Crook et al.,
1986), previously known as "benign senescent forgetfulness" (Krai, 1962).
Although it is not clear why recall of a 48-item list revealed genuine memory
impairment that was not apparent in recall of a 16-item list, for the purpose
of identifying genuine memory impairment in dementia, a 16-item list is
sufficient.
In the present study, demented patients, like the nondemented partici-
pants, benefited from the addition of cued recall. They retrieved four times
as many items by cued recall than by free recall, although they were still
unable to retrieve half the items when cued recall was added. The cued recall
advantage demented patients showed was larger than that reported in earlier
studies by Davis and Mumford (1984), Diesfeldt (1984), and Miller (1975).
One third of the participants in the present study were unable to free-recall
any items, but they were able to recall nearly 5 items on each trial
when cues were added. In other words, effective cued recall allows the meas-
urement of memory in patients who no longer can retrieve items by free recall
alone. Thus, in addition to providing a more accurate estimate than free re-
call of the amount of information stored by demented patients, effective
cued recall provides a measure that can be used to follow patients as the dis-
ease progresses.
Not only were some items never retrieved at all by demented patients, but
items that were retrieved on one trial often were not retrieved on the next
trial. This pattern of depressed and inconsistent total recall suggests that
genuine memory impairment in dementia reflects the disruption of more than
one memory process. The finding that some items were never recalled at all
suggests an encoding deficit whereby insufficient information about the item
is stored in the memory trace (Martin, Brouwers, Cox, & Fedio, 1985). When
this trace is retrieved, there is too little information in it to uniquely specify
the item, and so it may be confused with other semantically similar items.
Such an encoding deficit might help to account for the intrusions made by
AD patients (Fuld, Katzman, Davies, & Terry, 1982).
The inconsistency of total recall, on the other hand, shows that the mem-
ory impairment in dementia is not limited to encoding deficits but may in-
volve retrieval deficits as well. That patients were not always successful in
30 GROBER AND BUSCHKE
Identification of Dementia
The second major finding of the study was that decreased cued recall despite
effective processing discriminated elderly persons with dementia from those
without dementia better than free recall or recognition did. Seventy-five per-
cent of the variation in dementia status was explained by total recall, with
free recall and recognition explaining less. Only 1 person was misclassified
when total recall measures were used, but 5 and 16 persons, respectively, were
misclassified when free recall or recognition were used.
The fact that free recall did not misclassify even more patients may be due
in part to the composition of the dementia group. Ten of the patients had a
severe dementia with very limited free recall, which fell far below the cut
score. Fifteen patients had mild to moderate dementia shown by Blessed
scores of 16 or less. Four of them (26%) were misclassified by free recall but
were correctly classified by total recall. Because the population for which de-
mentia screening is intended is similar to the mild dementia group, the exclu-
sive use of free recall would be expected to lead to more misclassifications
than were obtained in the present study. In a cross-validation study, 7 of 13
(54%) mildly to moderately impaired patients (shown by Blessed scores of 16
or less) were misclassified by free recall (Grober et al., 1987). This means that
free recall alone will fail to identify some demented patients who could be
correctly identified by cued recall.
Total recall, unlike free recall, was below the cut score for each mildly im-
paired patient in both the present and follow-up studies, indicating that total
recall is a better measure for identifying dementia, particularly for patients in
the early course of the disease. The utility of total recall for identifying de-
mentia was further supported by the results of a factor analysis in which total
recall measures loaded with the Blessed score.
Because memory impairment is an essential criterion for the diagnosis of
dementia (APA, 1980; McKhann et al., 1984), we propose that a first step in
identifying dementia should be to identify genuine memory impairment. The
present study shows that when memory is measured under conditions that in-
duce appropriate semantic processing and provide for effective cued recall of
a 16-item list, it is possible to distinguish between elderly persons who have
apparent memory deficits and elderly persons who have genuine memory
GENUINE MEMORY DEFICITS IN DEMENTIA 31
deficits. Those persons with apparent memory deficits may have age-
associated memory impairment (Buschke & Grober, 1986). Those who show
genuine memory deficits on a 16-item list are likely to be demented because
dementia is the major cause of this kind of memory impairment in the eld-
erly. It appears that other causes of genuine memory impairment sufficiently
severe to produce decreased total recall of a 16-item list are much less com-
mon in the elderly. Anyone with such genuine memory impairment should be
evaluated further to determine whether they meet other criteria for the diag-
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
nosis of dementia.
Not only might the current version of controlled learning function to iden-
tify dementia, but it may be useful in performing the complementary func-
tion of ruling out dementia. That is, persons who show preserved memory
under conditions that induce semantic processing and provide for effective
cued recall are unlikely to be suffering a dementia. It is more likely that they
are suffering from age-associated memory impairment, a disorder for which
diagnostic criteria have recently been proposed (Crook et al., 1986). Of the 20
nondemented participants from the present study who survived the next
2 years, 18 underwent a complete reevaluation. Cognitive status was un-
changed in 17 of them who as a group recalled an average of 15.8 items per
trial at follow-up. Only 1 (Participant 2 in Table 1) of the 18 (5.5%) devel-
oped dementia during the follow-up period, which is compatible with the
base rate of dementia for this age group (Katzman, 1986; Rocca, Amaducci,
& Schoenberg, 1986). It is noteworthy that she was 1 of 2 nondemented par-
ticipants with a Blessed score of 5, which is in the range at risk for dementia,
and that her sum of total recall in the first year was 44, which is just above the
cut score of 43. At follow-up, her sum of total recall was 35, now clearly be-
low the cut score. Although cognitively normal elderly persons with intact to-
tal recall need to be followed for a longer period of time to determine if they
continue to show preserved memory, the results so far suggest that good per-
formance on controlled learning may be useful in ruling out dementia. This
would be important because normal elderly persons who suffer from age-
associated memory impairment are concerned that their forgetfulness may
presage a dementia.
Although the present study used a combination of free recall and cued re-
call to assess memory, the use of free recall is not necessary because it has al-
ready been shown that items that can be free-recalled can also be retrieved by
cued recall (Macht & Buschke, 1984). Thus, it would be sufficient for the pur-
pose of identifying genuine memory impairment to rely on cued recall alone.
Impaired free recall, however, may be useful in identifying age-associated
memory impairment or impairment of other cognitive processes that may re-
quire further evaluation.
Although controlled learning may be useful as a screening test for demen-
tia, several caveats are in order. First, it is necessary to validate the cut score
32 GROBER AND BUSCHKE
ACKNOWLEDGMENTS
This research was supported by U.S. Public Health Service Grants AGO-
3949, NS-03356, HD-01799, and NS-19234.
The authors thank the residents, staff, and administration of Morningside
House Nursing Home for making this study possible; Shereen Bang, Rita
Scott, and Lila Sherlock for evaluating the patients; and Miriam Aronson,
GENUINE MEMORY DEFICITS IN DEMENTIA 33
REFERENCES
American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
Davis, P. E., & Mumford, S. J. (1984). Cued recall and the nature of the memory disorders in
dementia. British Journal of Psychiatry, 144, 383-386.
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.
Drachman, D. A., & Leavitt, J. (1972). Memory impairment in the aged: Storage versus retrieval
deficit. Journal of Experimental Psychology, 93, 302-308.
Eysenck, M. W. (1974). Age differences in incidental learning. Developmental Psychology, 10,
936-941.
Eysenck, M. W., & Eysenck, M. C. (1979). Processing depth, elaboration of encoding, memory
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
Perlmutter, M. (1978). What is memory aging the aging of? Developmental Psychology, 14,
330-345.
Perlmutter, M. (1979). Age differences in adults' free recall, cued recall, and recognition. Jour-
nal of Gerontology, 34, 533-539.
Perlmutter, M., & Mitchell, D. B. (1982). The appearance and disappearance of age differences
in adult memory. In F. I. M. Craik & S. Trehub (Eds.), Aging and cognitive processes: Vol. 8.
Advances in the study of communication and affect (pp. 127-144). New York: Plenum.
Poon, W. (1985). Differences in human memory with aging: Nature, causes and clinical implica-
tion. In J. E. Birren & K. W. Shaire (Eds.), Handbook of the psychology of aging (2nd ed.,
Downloaded by [Memorial University of Newfoundland] at 18:08 03 August 2014
Cognition, 6, 590-598.
Wilson, R. S., Bacon, L. D., Kramer, R. L., Fox, J. H., & Kaszniak, A. W. (1983). Word fre-
quency effect and recognition memory in dementia of the Alzheimer type. Journal of Clinical
Neuropsychology, 5, 97-104.