Journal of Experimental Psychology: General
1984, Vol. 113, No. 4, 556-570.
Copyright 1984 by the
American Psychological Association, Inc.
Paired-Associate Learning and Priming Effects in Amnesia:
A Neuropsychological Study
Arthur P. Shirriamura
Larry R. Squire
Department of Psychiatry,
University of California
School of Medicine, La Jolla
Veterans Administration Medical Center,
San Diego
and Department of Psychiatry,
University of California,
School of Medicine, La Jolla
SUMMARY
Despite severe deficits of recall and recognition, amnesic patients can exhibit normal
priming effects. Amnesic patients have also been reported to perform well on tests of
paired-associate learning that involve related word pairs (e.g., TABLE-CHAIR). The
present study investigated the role of priming effects in paired-associate learning. Experiment 1 illustrated the distinction between the memory impairment of amnesic patients
and their intact priming ability. Amnesic patients were markedly deficient in learning
unrelated word pairs, despite exhibiting normal priming as measured by a word-completion test involving the same words. In Experiment 2A, amnesic patients showed good
paired-associate learning for related word pairs, though control subjects still performed
significantly better. In addition, the good performance by amnesic patients was shortlived, and performance fell to baseline after a 2-hr delay. Control subjects performed
well above baseline at all delay conditions. Experiment 2B showed that the forgetting of
related word pairs by amnesic patients followed the same time course as the decay of
word priming. Experiment 3 showed that amnesic patients were as good as control
subjects at learning related word pairs when incidental learning and test procedures were
used (a word-association test). The advantage of control subjects over amnesic patients
in Experiments 2A and 2s could therefore be attributed to the explicit learning instructions that are standard in paired-associate tests. Finally, Experiment 4 showed that
amnesic patients exhibited normal priming when they were asked to "free associate" to
words (e.g., CHILD) that were semantically related to previously presented words (e.g.,
BABY).
The results indicate that both priming effects and paired-associate learning of related
word pairs depend on activation, a process that is preserved in amnesia. Activation can
account for the findings of good performance by amnesic patients on tests of word
priming (Experiments 1 and 2s), related paired associates (Experiments 2A and 2s), and
word association (Experiments 3 and 4), Activation is a transient phenomenon presumed to operate on and facilitate access to preexisting representations. Control subjects
can establish new associations and can strengthen preexisting associations by engaging
processes that are impaired in amnesia. As a result, when explicit learning instructions
are used to test paired-associate learning of related word pairs, control subjects can learn v
better and can remember longer than can amnesic patients (Experiments 2A and 2s).
The process of activation, which like skill learning is intact in amnesia, is contrasted with
declarative memory, which is impaired. Declarative memory includes the facts and
episodes of everyday experiences and is available to conscious recollection. This distinction between activation and declarative memory illustrates the usefulness of neuropsychological approaches to questions about how memory is organized in the brain.
556
PAIRED-ASSOCIATE LEARNING AND PRIMING EFFECTS
One useful approach to the analysis of normal memory functions is to study amnesia
(for reviews, see Baddeley, 1982b; Cermak,
1982; Hirst, 1982; Squire, 1982b; Stern,
1981; Weiskrantz, 1982). Memory impairment can occur as a relatively selective neurological disorder; and, to the extent that this
impairment is selective, neuropsychological
investigations can reveal information about
the structure and organization of normal
memory. Because amnesia depends on neurological injury or disease in particular brain
regions, the same research program can rejate
the psychological study of memory functions
to neural studies of brain systems and can
make discussions of cognitive/neural mechanisms more specific and concrete.
Although amnesias of different etiologies
can differ from each other in certain respects
(Cohen & Squire, 1981; Huppert & Piercy,
1979; Moscovitch, 1982; Squire, 1981,
1982a), there are many similarities among
them. The most available and best-studied
example of amnesia is the Korsakoff syndrome, which develops after years of chronic
alcohol abuse and is characterized by bilateral symmetrical lesions along the walls of the
third and fourth ventricles, as well as by cortical atrophy and cerebellar damage. Although patients with Korsakoff syndrome
form a heterogeneous group and can show a
wide range of neuropsychological deficits in
addition to amnesia (Butters & Cermak,
1980; Talland, 1965; Zangwill, 1977), they
can be carefully screened to identify those
patients who have primarily a memory impairment that occurs out of proportion to
other cognitive deficits. Damage to the diencephalic midline is thought to.be responsible
for the memory impairment (Brierley, 1977;
Mair, Warrington, & Weiskrantz, 1982; Victor, Adams, & Collins, 1971).
This research was supported by a National Research
Service Award (MH08992) from the National Institute
of Mental Health (A.S.), the Medical Research Service of
the Veterans Administration, and National Institute of
Mental Health Grant MH24600 (L.S.).
We thank Joyce Zouzounis, Armand Bernheim; and
Patty Feldstein for research assistance.
Request for reprints should be sent to Larry R. Squire,
Department of Psychiatry (116), Veterans Administration Medical Center, 3350 La Jolla Village Drive, San
Diego, California 92161.
557
Severe loss of new learning capacity (anterograde amnesia) and loss of premorbid memory (retrograde amnesia) can occur in patients despite normal IQ scores, intact
language functions, and appropriate social
skills. Amnesic patients are unable to recall
or to recognize information that was presented only minutes ago, though working or
short-term memory, as measured by digit
span, is typically normal (Baddeley & Warrington, 1970; Drachman & Arbit, 1966).
Despite these memory impairments, amnesic patients can learn and can remember
perceptual-motor skills such as mirror
drawing (Milner, 1962) and pursuit-rotor
tasks (Brooks & Baddeley, 1976; Cermak,
Lewis, Butters, & Goodglass, 1973; Cohen,
1981; Corkin, 1968). Moreover, amnesic patients can acquire and can maintain in a nor^
mal fashion cognitive skills such as the ability
to read words from a mirror-reversed display
(Cohen & Squire, 1980) and the ability to<
solve the Tower of Hanoi problem (Cohen &
Corkin, 1981). These skills can be retained
for weeks or months (Brooks & Baddeley,
1976; Cohen & Squire, 1980) without patients having any conscious recollection of
having previously seen or used the test materials.
Another example of preserved memory
performance by amnesic patients is the phenomenon of priming. Warrington and
Weiskrantz (1968, 1970, 1974, 1978) originally developed a technique that sometimes
yielded normal performance in amnesic patients. Instead of requesting recall or recognition, subjects were cued by presenting partial
information of the to-be-remembered items.
For example, after studying a list of words,
subjects were presented with the first three
letters of each word. When cued in this way,
amnesic patients sometimes performed as
well as normal control subjects. Similar results were obtained by cuing subjects with
fragmented words or with degraded pictures
(Warrington & Weiskrantz, 1968, 1970).
This method, however, did not always yield a
selective improvement in amnesic patients
(Mayes, Meudell, & Neary, 1978; Morten,sen, 1980; Squire, Wetzel, & Slater, 1978;
Wetzel & Squire, 1982). These later findings
suggested that the cued-recall procedure does
not improve the memory of amnesic patients
558
ARTHUR P. SHIMAMURA AND LARRY R. SQUIRE
disproportionately but that it facilitates performance in all subjects to an equal extent.
Graf, Squire, and Mandler (1984) discovered an important variable that accounts for
the conflicting findings with the partial cuing
method. When subjects were instructed to
use three-letter word stems as cues to recall
previously presented words (cued-recall instructions), the cues improved recall in all
subjects, but amnesic patients still performed
more poorly than did control subjects. However, when subjects were instructed to use the
word stems to form the first word that came
to mind (word-completion instructions),
amnesic patients and control subjects performed the same. This word-completion effect decayed at a normal rate and disappeared
after a 2-hr delay in both amnesic patients
and control subjects.
The view that priming effects are preserved
in amnesia was first suggested by Rozin
(1976), who proposed a trace-activation process operating as a short-lived, "hot-tubes"
phenomenon. He stated that activation operates on already existing memory traces and
persists for at least several minutes (i.e.,
beyond the span of short-term memory). Although amnesic patients may have no conscious awareness of previously presented information, they can demonstrate activation
in word-completion tasks such as those reported by Graf et al. (1984) and Rozin and
Diamond (cited in Rozin, 1976). Jacoby and
Witherspoon (1982) found further evidence
to support this notion. Amnesic patients and
control subjects were asked to "name a musical instrument that employs a reed." They
were then asked to spell homophonic words
such as reed/read. Amnesic patients and
control subjects showed a similar bias toward
the spelling of previously presented words,
but amnesic patients were impaired in a recognition test involving the same words.
The skill learning and activation exhibited
by amnesic patients are similar in that they
do not require conscious recollection of past
learning sessions. Unlike conventional tests
of memory, where subjects are asked to determine whether a particular word or picture
was presented at a particular time, skill learning and activation can be demonstrated without subjects even knowing that their memory
is being tested. Subjects simply read words,
work at puzzles, complete word stems, spell
words, or identify degraded pictures. Amnesic patients are impaired when tasks require
what has been termed conscious recollection
or declarative memory.
These findings are important because they
point to a fundamental distinction between
two memory processes or systems. One system is impaired in amnesia and depends on
the integrity of the medial temporal and
diencephalic brain structures damaged in
amnesia; the other is spared in amnesia and
can operate normally in the absence of these
same brain regions. These facts of amnesia
have been interpreted in terms of a distinction between procedural memory and declarative memory (Cohen, 1981, 1984; Squire,
1982b) and have been incorporated into a
neuropsychological account of how memory
is organized in the brain (Squire, in press-a;
Squire & Cohen, 1984). Declarative memory
is available to conscious awareness and includes the facts and episodes of everyday experiences. Procedural memory is implicit
and is available only by engaging the specific
operations in which the memory is embedded. Declarative memory is created by adding new data structures. Procedural memory
is created by tuning, biasing^ or combining
preexisting elements. Similar distinctions
between two memory processes or systems
have been suggested by others (taxon vs.
locale, O'Keefe & Nadel, 1978; perceptual
vs. autobiographical memory, Jacoby &
Dallas, 1981; semantic memory vs. cognitive
mediation, Warrington & Weiskrantz, 1982;
skills vs. conscious recollection, Moscovitch,
1982; automatic vs. conscious recollection,
Baddeley, 1982a; habits vs. memories, Mishkin, Malamut, & Bachevalier, 1984; integration vs. elaboration, Mandler, 1979, 1980).
To specify more clearly what is preserved in amnesia, we considered the findings
for paired-associate learning in patients
with Korsakoff syndrome. Paired-associate
learning is ordinarily considered to be among
the most sensitive measures of memory impairment (Ericson & Scott, 1977; Squire, in
press-b). When randomly paired words are
used as stimuli, amnesic patients are strikingly impaired (Jones, 1974; Squire, in
press-b). Yet amnesic patients exhibit normal or close to normal learning when the
PAIRED-ASSOCIATE LEARNING AND PRIMING EFFECTS
stimuli are highly associated word pairs (e.g.,
TABLE-CHAIR) (Cutting, 1978; Winocur &
Kinsbourne, 1978; Winocur & Weiskrantz,
1976). Cutting (1978) showed that normal
performance occurred only when response
words were the most common associate of
the stimulus word and only when these pairs
were presented in blocked trials. Amnesic patients were impaired when response words
were less common associates (e.g., SOLDIER MARCH) and when the degree of association
was varied within a list (mixed trials).
The performance of amnesic patients in
tests of related paired associates suggests a
possible role of the trace-activation process
proposed by Rozin (1976). Instead of activating the representations of previously learned
words, a test of related paired associates activates preexisting semantic associations. Amnesic patients may benefit from an activation
process in tests of paired-associate memory,
and this process may be sufficient to produce
normal or near-normal performance. Amnesic patients may have available little declarative memory for word pairs but may be able
to produce the first associate that comes to
mind. As a result, they might perform nearly
as well as control subjects on tests involving
related word pairs. Furthermore, good performance on such tests should be apparent
only when short study-test delays are used
because activation could support good performance across short delays. At longer
delays, activation should have decayed, and
amnesic patients should be impaired. In contrast, control subjects can strengthen preexisting associations by invoking processes that
support declarative memory (e.g., organization, elaboration, reconstruction). As a result, they might perform even better than
amnesic patients at short study-test delays,
and their advantage over amnesic patients
should be even greater at longer delays.
Wickelgren (1979) explained in a slightly
different way why amnesic patients are so
good at learning highly related word pairs. He
made a distinction between vertical associations, which require the addition of new
nodes in an associative network, and horizontal associations, which require only the
strengthening of already existing nodes.
Wickelgren (1979) suggested that amnesic
patients are impaired at forming vertical as-
559
sociations but are not impaired at strengthening horizontal associations. Based on this
conceptualization, amnesic patients are impaired at learning unrelated word pairs because of a failure to form vertical associations. However, amnesic patients can
learn related word pairs because they can
strengthen horizontal associations in a normal way.
The present study asked whether the good
or normal performance of amnesic patients
at learning highly related word pairs can be
explained as an instance of the activation of
preexisting semantic associations. If this ability is an example of activation, then amnesic
patients should not be able to 16am and to
remember related word pairs in a normal
way, as suggesjed by Wickelgren (1979). Instead, good performance should be only
transient and should decay rapidly in the
same way that word completion has been
shown to decline during the hours after learning. In addition, if the learning of highly related word pairs can be explained by activation, then amnesic patients and control
subjects should exhibit substantial and
equivalent effects of word-pair presentation
when shown the first word in a pair and instructed simply to say the first word that
comes to mind (i.e., to free associate). Furthermore, if activation is important in the
learning of highly related word pairs, it
should be possible to produce substantial
priming in amnesic patients by presenting
only the response words initially and then by
simply asking for free associations to the
stimulus words.
The present study tested these ideas in a
series of experiments. Experiment 1 was designed to illustrate the memory impairment
in amnesic patients by showing a deficit in
paired-associate learning for unrelated word
pairs and by showing intact word-completion
ability for the same words. In Experiment 2A,
we tested the ability of amnesic patients to
learn highly associated word pairs, and we
determined how long after learning this ability persisted. In Experiment 2s, we compared
the persistence of paired-associate learning
with the decay of word completion. In Experiment 3, we tested paired-associate performance us^ng incidental, free-association instructions. Finally, in Experiment 4, we
560
ARTHUR P. SHIMAMURA AND LARRY R. SQUIRE
presented only the response words of highly
related word pairs (e.g., CHAIR) and asked
subjects to free associate to stimulus words
(e.g., TABLE).
Experiment 1
In Experiment 1, we examined word-completion ability and paired-associate learning
in amnesic patients and alcoholic control
subjects. Subjects studied unrelated word
pairs and then were asked to complete threeletter word stems to form words. The word
stems could be completed using stimulus
words from the study list. Immediately following the word-completion test, paired-associate memory was assessed. We predicted
that although paired-associate memory for
unrelated pairs would be imp%ired in amnesic patients, word-completion ability would
be intact.
Method
Subjects
1. Patients with alcoholic Korsakoff syndrome. The
group consisted of 5 men and 3 women with an average
age of 53.8 years and an average education level of 12.4
years. The average full-scale Wechsler Adult Intelligence
Scale-Revised (WAIS-R) IQ was 93.4 (range = 85103), and the average Wechsler Memory scale score was
78 (range = 64-93). Neuropsychological screening and
independent neurological examination indicated that
memory impairment was the only remarkable deficit of
higher cortical function. All patients could draw a cube
and a house in perspective, and none had aphasia or
apraxia. These patients have been studied as a group for
several years, and their memory impairment has been
documented in detail (Cohen & Squire, 1981; Graf et al.,
1984; Squire, 1981, 1982a; Zola-Morgan, Cohen, &
Squire, 1983).
2. Control group for patients with Korsakoff syndrome. Seven male alcoholics participated in this experiment. Their average drinking history was 18.1 years.
They had no history of head injury or liver disease, and
they had been free of alcohol for an average of 9.0 weeks.
They averaged 46.6 years of age and averaged 13.4 years
of education. Their average WAIS-R subtest scores
were 19.4 for Information (Korsakoff patients = 17.6)
and 47.4 for Vocabulary (Korsakoff patients = 43.7).
After all 12 word pairs were presented, the cards were
shuffled and were presented a second time. Following
this study phase, word-completion ability was assessed.
Subjects were given 18 three-letter stems and were asked
to add letters to form the first word that came to mind.
There were at least four different ways to complete each
stem. The words could be of any length; proper nouns
were not acceptable. No three-letter stem was repeated
during the experiment. Of these 18 stems, 12 could be
completed using stimulus words from the paired-associate test (e.g., subjects studied STAIR-DIAMOND and
were then asked to complete the stem STA
).
The other 6 stems were used to establish baseline performance, and they could be completed using stimulus
words from the unused set of word pairs. Immediately
after the word-completion test (2 min to 4 min after the
paired-associate study phase), memory for the word
pairs was tested by presenting the first word of each pair
and by asking subjects to recall the word that had been
presented with it. Finally, 6 additional word stems were
presented for the baseline measure, providing a total of
12 baseline responses. The probability of completing
these stems with stimulus words from the unused set
provided the measure of baseline guessing. Thus, one set
of 12 word pairs was used in the paired-associate and
word-completion tests, and another set was used to measure baseline word completion. The two sets were counterbalanced across these two conditions.
Results and Discussion
Shown in Figure 1 are the results of the
word-completion and paired-associate tests
(all statistical analyses are based on up < .01
significance level unless otherwise stated).
70
Study:
STAIR-DIAMOND (12 pain)
Complete: STA—
Test:
STAIR-
?
60
50
40
30
20
10
Materials and Procedure
Two sets of 12 unassociated word pairs were prepared
using random pairings of words from the A or AA categories of the Thorndike-Lorge (1944) word-frequency
norms. These word pairs were printed individually on
index cards. Subjects were given one set of 12 word pairs
and were instructed to read the words aloud and to try to
remember them as a pair. Subjects saw each pair for 3 s.
ALC
KOR
WORD COMPLETION
ALC
KOR
PAIRED-ASSOCIATE
TEST
Figure 1. Intact word-completion performance by amnesic patients despite impaired learning of unrelated
word pairs. (ALC = alcoholics; KOR = patients with
Korsakoff syndrome.)
PAIRED-ASSOCIATE LEARNING AND PRIMING EFFECTS
In the word-completion test, alcoholic con-"
trols and amnesic patients performed identically, F(l, 13) = 0.327, MSe = 4.55. In both
groups, stimulus words were used to complete the three-letter stems at a rate well
above the guessing baseline, F( 1,13) = 51.9,
MSe = 3.06. Baseline guessing was similar in
the two groups: controls = 20%, amnesic patients = 27%; f(13) = .99. In contrast to the
findings for word completion, there was a
large difference between groups in the
paired-associate test for unrelated word pairs,
t( 13) = 3.38. Alcoholic subjects recalled 40%
of the response words, whereas the amnesic
patients recalled only 2% of the words.
The results of Experiment 1 illustrated in a
single group of patients and with the same set
of words the two established features of amnesia that form the basis for the present
study: Although paired-associate memory
for unrelated word pairs was severely impaired in amnesic patients, the ability to activate previously presented words was intact.
Amnesic patients, like control subjects, exhibited a threefold increase in word activation over baseline guessing.
561
strengthen preexisting associations—an
ability presumed normal in amnesic patients
(Wickelgren, 1979)—then amnesic patients
should exhibit good paired-associate performance at all delay conditions.
Method
Subjects
1. Patients with alcoholic Korsakoff syndrome. (See
Experiment 1.)
2. Control group for patients with Korsakoff syndrome. Eight male alcoholics participated in this experiment. Their average drinking history was 20.2 years.
They had no history of head injury or liver disease, and
they had been free of alcohol for an average of 9.8 weeks.
They averaged 51.4 years of age and averaged 12.6 years
of education. Their average WAIS-R subtest scores
were 20.9 for Information (Korsakoff patients = 18.0)
and 51.0 for Vocabulary (Korsakoff patients = 43.8).
Materials and Procedure
Forty-eight highly associated wprd pairs were compiled using the word association norms of Jenkins
(1970). Response words were either the first, second, or
third highest associate. Some highly associated word
pairs from Winocur and Weiskrantz (1976) were also
included. The pairs were divided into four sets of 12.
Three of the four sets were used to assess paired-associate
memory after three delay conditions (0, J 0, or 120 min).
The remaining set was used to establish the guessing
Experiment 2A
baseline. In the study phase, each word pair was preIn this experiment, paired-associate mem- sented for 3 s, with instructions to read the words aloud
to remember them as a pair. Then the cards were
ory for highly related word pairs was assessed and
shuffled and were presented a second time. After a delay
at three different delay conditions after learn- of 0, 10, or 120 min, the first word of each pair was
ing (0, 10, and 120 min). Earlier studies presented, and subjects were asked to name the word
found that amnesic patients exhibit normal that had been paired with it, If subjects could not remember the word, they were asked to guess. All subjects
or near-normal paired-associate memory participated
in the three delay conditions, and each confor highly related word pairs when patients dition was administered on a separate day. The order of
are tested immediately after learning (Cut- the conditions was counterbalanced. The guessing rate
ting, 1978; Kinsbourne & Winocur, 1980; was assessed after the third delay condition by asking
Winocur & Kinsbourne, 1978; Winocur &' subjects to generate an associate to each stimulus word
the remaining set. All four sets of word pairs were
Weiskrantz, 1976). If the ability of amnesic from
counterbalanced across test conditions.
patients to remember highly related word
pairs is based on activation or priming, then
their ability to remember the word pairs
should be short-lived. Based on previous
studies of activation (Graf et al., 1984; Rozin
& Diamond, cited in Rozin, 1976), amnesic
patients should show complete forgetting
after a 2-hr delay. Control subjects, however,
should maintain a high level of performance
because their memory is based not only on
activation but also on declarative memory.
Alternatively, if memory for highly related
word pairs is based on the ability to
Results and Discussion
Shown in Figure 2 are the results of the
paired-associate test for highly related word
pairs by amnesic patients and alcoholic
control subjects. There was a significant
difference between subject groups, F(l,
14) = 28.7, MSe = 3.84. Newman-Keuls
analyses showed that the alcoholic controls
performed better than the amnesic patients at
all delay conditions. There was also a significant main effect of delay, F(3, 42) = 36.7,
562
ARTHUR P. SHIMAMURA AND LARRY R, SQUIRE
Experiment 2s
Study: TABLE-CHAIR (12 pairs)
Ten:
TABLE-
?
This experiment was identical with Experiment 2A except that a word-completion test
was given just prior to the paired-associate
test. If activation is responsible for both
word-completion and paired-associate performance in amnesic patients, then the forgetting curves for these two^tests should be
similar. Patients with Korsakoff syndrome
and alcoholic control subjects studied highly
related word pairs. After a delay condition of
0, 10, or 120 min, word-completion ability
was assessed for the first word in the pair by
giving a three-letter stem and by asking subjects to complete the stem with the first word
that came to mind.
ALC
g 40
£
30
Baseline
20
10
0 10
Method
120
DELAY (min)
Figure 2. Retention of highly related word pairs at three
intervals after learning by amnesic patients and control
subjects. (ALC = alcoholics; KOR = patients with Korsakoff syndrome.)
MSe = 3.95, but the Group X Delay interaction did not reach significance. Dunnett's
tests to compare treatment means with a
control condition showed that the alcoholic
subjects performed better than their baseline
guessing rate at all delay conditions. The amnesic patients, however, scored above their
baseline guessing rate only at the immediate
and 10-min delay conditions but did not
score above baseline after the 2-hr delay,
t(7) = 1 .26, p > . 1 . Finally, there was no significant difference between alcoholic subjects and amnesic patients in their baseline
guessing rates: 20% vs. 25.8%, t(l4) = 1.08,
These results support the hypothesis that
memory for highly related word pairs in amnesic patients is related to what has been
termed activation. In the immediate condition, amnesic patients showed a threefold increase in paired-associate memory over baseline guessing. Like word completion, this
ability decayed after a 2-hr delay. The control
subjects, however, performed well above
chance at all delay conditions. These results
suggest that control subjects can use elaborative encoding to strengthen preexisting associations, whereas amnesic patients can rely
only on a transient activation process.
Subjects
The same subjects who participated in Experiment 2A
participated in Experiment 2B.
Materials and Procedure
Forty-eight highly associated word pairs—all different from those used in Experiment 2A—were compiled
from Jenkins (1970) and Winocur and Weiskrantz
(1976). These pairs were divided into four sets of 12. The
design was similar to the one used in Experiment 2A.
Subjects studied 12 word pairs and were tested at one of
three delay conditions (0, 10, or 120 min). The only
difference between Experiment 2A and Experiment 2s
was that just prior to the paired-associate test, a wordcompletion test was given. For word completion, subjects were given three-letter stems and were asked to add
letters to form the first word that came to mind. There
were at least four different ways to complete each stem.
The words could be of any length; proper nouns were not
acceptable. No three-letter stem was repeated during the
experiment. At each test, 16 word stems were presented.
Twelve of the word stems could be completed using the
stimulus words from the paired-associate study list. The
remaining 4 word stems could be completed using stimulus words from the unused set, and they permitted an
estimate of baseline completion performance. Thus, a
total of 12 word stems (4 word stems per delay X 3
delays) were used to assess baseline guessing rates. The
four sets of word pairs were counterbalanced across conditions. ,
Results and Discussion
Shown in Figure 3 are the data for alcoholic subjects' and amnesic patients on the
word-completion and paired-associate tests.
There were no differences between alcoholic
subjects and amnesic patients on the word-
PAIRED-ASSOCIATE LEARNING AND PRIMING EFFECTS
Study:
STOVE-HOT (12 pairs)
Complete: STO—
Teit:
70
STOVE- ?
A. Word Completion
60
s
50
1C
S40
| 30
o
DC
° ^^r—
K 20
Baseline
10
I
120
I I
0' 10
DELAY (mini
100
B. Paired-Associate Test
90
80
70
6
°
so
40
i
30
Baseline
20
10
0 10
120
DELAY (mint
Figure 3. Word-completion performance (A) and
paired-associate performance (B) at three intervals after
studying highly related word pairs. (ALC = alcoholics;
KOR = patients with KorsakofF syndrome. The
paired-associate test followed word completion.)
563
The results of the paired-^associate learning
test (Figure 3, bottom half) replicated the
findings of Experiment 2A. The alcoholic
subjects showed better memory for the paired
associates than did the amnesic patients, F( 1,
14) = 34.2, MS; = 3.3. The amnesic patients
performed above the baseline guessing rate at
the immediate and 10-min delay conditions
but did not perform better than would have
been expected by guessing after a 2-hr delay.
The only difference between the paired-associate data in Figure 3 (bottom half) and the
paired-associate data from Experiment 2A
(Figure 2) was that the performance of both
groups was poorer in the immediate condition of Experiment 2s. This difference is
probably a result of introducing the wordcompletion test in Experiment 2B, which delayed the paired-associate test by a few minutes.
The findings of Experiment 2u give additional support to the hypothesis that activation is responsible for both paired-associate
learning and word-completion ability in amnesic patients. The decay rates were similar,
and amnesic patients did hot perform above
baseline guessing after a 2-hr delay in either
the paired-associate learning test or the
word-completion test. In the word-completion test, amnesic patients and alcoholic
control subjects performed similarly in
all respects. This finding replicated the wordcompletion data reported by Graf et al.
(1984). In the paired-associate tests used in
Experiments 2A and 2s, amnesic patients
performed well above chance at the 0-,and
10-min delay conditions, but their performance was still below that of control subjects.
Experiment 3
completion test (see Figure 3, top half). Both
groups performed above baseline guessing
rates at the immediate and 10-min delay
conditions, and both groups fell to baseline
guessing rates after a 2-hr delay. NewmanKeuls analyses showed no difference between
alcoholic subjects and amnesic patients at
any of the delay conditions and showed no
difference in their baseline guessing rates:
alcoholic subjects = 14%, amnesic patients = 22%;/(13) = .56.
The superior performance of the control
subjects at short delays in Experiments 2A
and 2s might be due to the fact that pairedassociate learning was accomplished under
explicit instructions first to learn the pairs
and later to recall words from the study list.
This method would permit control subjects
to rely not only on activation but also on their
intact ability to use declarative memory.
Thus, when instructions to remember are
given, contror subjects can use elaborative
memory strategies that are impaired in am-
564
ARTHUR P. SHIMAMURA AND LARRY R. SQUIRE
100
nesic patients. In Experiment 3, we presented
90
highly related word pairs in an incidental
learning paradigm. Subjects were shown
word pairs and were asked to rate each pair
according to the semantic relatedness between the two words in a pair. Subjects were
then given a word-association task in which
they said the first word that came to mind in
response to a list of cue words. Each cue word
was actually the first word of a pair used in the
rating task. If the results obtained in Experiments 2A and 2s were due to control subjects
using elaborative strategies under explicit instructions, then the incidental testing paraALC
KOR
ALC
KOR
digm used in Experiment 3 might be exPAIRED-ASSOCIATE
WORD ASSOCIATION
pected to lower the performance of control
TEST
(from Fig. 2)
subjects but not to affect the performance of
Figure 4. Comparison of word-association performance
amnesic patients.
Method
Subjects
1. Patients with alcoholic Kofsakoff syndrome. (See
Experiment 1.)
2. Alcoholic control subjects. Seven male alcoholic
subjects participated in this experiment. None of these
subjects had been tested in any of the previous experiments. These subjects had an average drinking history of
14.3 years with no history of head injury or liver disease.
They had been free of alcohol for an average of 8.7
weeks. They averaged 45.9 years of age and averaged
13.4 years of education. They obtained averages of 19.7
and 45.6 on the WAIS-R Information and Vocabulary
subtests, respectively.
Materials and Procedure
Two sets of 12 highly related word pairs that had been
used in Experiment 2A were used in this experiment.
The word pairs were presented in an incidental manner.
Subjects were shown a word pair and were asked to judge
how much the two words were related to each other on a
4-point rating scale, rating from highly related (I) to not
related (4). After all 12 word pairs were rated, they were
presented in a different order and were rated a second
time. Then, the subjects were told the following:
Now we will do a different task. I will say a word, and I
want you to tell me the first word that comes to mind.
Any word will do, just be sure that it is the first word
that comes to mind.
In this way, the subjects were encouraged to believe that
the test was one of free association. A total of 24 word
associations were requested. For 12 of these trials, the
cue words comprised the first word of each pair that had
been used in the rating task. The remaining 12 cue words
came from the unused set of word pairs and provided a
measure of baseline guessing. The two sets were counterbalanced between the study and baseline conditions.
(Experiment 3) and paired-associate learning (Experiment 2A) by patients with Korsakofi'syndrome (KOR)
and alcoholic control subjects (ALC). (The data for
paired-associate learning were taken from the 0-delay
condition of Figure 2.)
Results and Discussion
There was no difference in the ratings that
amnesic patients and control subjects gave
in response to the relatedness orienting
task: amnesic patients = 1.33, control subjects = 1.55; J(13) = .81. More important,
amnesic patients and alcoholic control subjects did not differ in their performance on
the word-association task, F(l, 13) = 0.55,
MSe = 2.71. Also, there was no difference in
baseline guessing: amnesic patients = 28%,
alcoholic controls = 23%; £(13) = .95. The
results from the word-association task used
here and the results from the paired-associate
test used in Experiment 2A are compared in
Figure 4. These two sets of data were obtained using the same list of word pairs and
the same learning-retention interval. Only
the instructions differed. The amnesic patients performed the same on the two tests,
but the control subjects performed better on
the test given in Experiment 2A, when they
were given explicit instructions to remember
the word pairs.
Experiment 4
The preceding experiments showed that
amnesic patients can be primed when re-
565
PAIRED-ASSOCIATE LEARNING AND PRIMING EFFECTS
r CORRECT
cently encountered material is directly preStudy:
BABY (12 wordi)
sented (e.g., presenting three-letter beginWord Auociation: CHILD?
60
Recall:
Study wordi?
nings of words or the first word of a highly
related pair). In fact, these findings, as well as
SO
previous demonstrations of preserved prim40
ing ability in amnesic patients, have all been
v^
s:
based on presenting material that was used
during study. Experiment 4 was designed to 130 determine if amnesic patients could also ex20 - Besellm
hibit priming when cued with semantic
associates—words related to previously pre10
sented words but words that had not been
0
directly presented themselves. In this experi0
120
ALC KOR
ALC KOR
DELAY (mint
0
120
ment, words were shown in an incidental
WORD ASSOCIATION
RECALL
learning paradigm, subjects rated how much
they liked the words. Then subjects were Figure 5. Semantic priming as measured by a word-assoshown semantic associates of the stimulus ciation test (left panel) and recall performance (right
words and were asked simply to free associate panel) by amnesic patients and control subjects at two
, to these words, The measure of interest was delays after studying a word list. (ALC = alcoholics;
the probability of producing a stimulus word KOR = patients with Korsakoff syndrome.)
when given its semantic associate. This
primacy and recency effects, 2 extra words were placed at
word-association task was given after a delay the
and end of each se^ of 12 words. After a
condition of 0 or 2 hr and was followed by a delaybeginning
condition of 0 or 120 min, the subjects participated
free-recall test. It was predicted that amnesic in a word-association task that was similar to the one
patients would show a preserved capacity for used in Experiment 3. The cue words actually consisted
semantic priming and that this ability would of stimulus words from the paired-associate test of Ex2B. For example, subjects rated how much they
decay after a 2-hr delay for both amnesic pa- periment
the word HOT and then were asked to say the first
tients and control subjects. The amnesic pa- liked
word that came to mind when given the word STOVE. In
tients should be impaired at free recall at both this case, the correct response was the word HOT. Both
delay conditions were given the same day, with a 10-min
delay conditions.
Hk
•M
•MB
=?
ffi
Method
Subjects
1. Patients with alcoholic Korsakoff syndrome. (See
Experiment 1,)
2. Alcoholic control subjects. Eight male alcoholic
subjects participated in this experiment. Four of them
had participated in Experiments 2A and IB, but none
had been tested for at least 2 months. The 8 subjects had
an average drinking history of 18.9 years with no history
of head injury or liver disease. They have been free of
alcohol for an average of .9.0 weeks. They averaged 46.6
years of age and averaged 13,4 years of education. They
obtained averages of 19.4 and 47.4 on the WAIS-R
Information and Vocabulary subtests, respectively.
Materials and Procedure
The stimuli consisted of 36 words that had been used
as response words in the paired-associate test of Experiment 2B. These words were divided into three sets of 12.
In the study phase, a word was presented on a card, and
subjects used a 5-point scale, ranging from dislike very
much (1) to like very much (5), to rate how much they
liked the word. After a set was presented; the cards were
shuffled, and the words were rated again. To minimize
rest period between conditions. The order of the delay
conditions was counterbalanced. Word associations
were also obtained for a third set of 12 cue words. These
cue words were associates of the unused set of words and
provided a measure of baseline guessing. Six of these cue
words were presented at each delay condition (0 or 120
min) and were randomly intermixed with the test cues.
Following the word-association task, subjects were asked
to recall the words rated in the liking test. The three word
sets were counterbalanced across the two delays and the
baseline condition.
Results and Discussion <
Amnesic patients and control subjects
did not differ in their responses to the liking
task: amnesic patients = 3.19, control subjects = 3.39; t(l4) = .79. Shown in Figure 5
are the data for the word-association and recall tests. Alcoholic controls and amnesic patients did not differ in the word-association
task, F(\, 14) - 0.016, MSf = 5.174. Both
the priming effect and baseline performance
were nearly identical. There was, however, a
significant effect of delay, F(2, 28) = 25.9,
MSe= 1.506. Dunnett's tests showed that
566
ARTHUR P. SHIMAMURA AND LARRY R. SQUIRE
performance in the immediate condition tributed to the explicit learning instructions
was better than baseline for alcoholic sub- given in standard paired-associate tests. Fijects (/[7] = 5.30) and for amnesic patients nally, Experiment 4 showed that amnesic pa(/[7] = 4.48). After a 2-hr delay, however, tients exhibited intact priming when preperformance was no better than baseline for sented with semantic associates—words
either group (p > . I). Baseline guessing per- related to previously presented words but
formance was identical for the two groups words that had not been directly presented.
In the present study and in a previous one
(23%).
On tests of free recall (see Figure 5), alco- .(Warrington & Weiskrantz, 1982), amnesic
holic subjects recalled more words than did patients performed well on tests with related
amnesic patients at both the immediate and word pairs, but their performance was signifithe 2-hr delay conditions, f(l, 14) = 39.4, cantly below normal levels. There have been
MSe = 3.25. There was a significant effect of several other reports that the performance of
delay, F(l, 14) =10.72, ATS.-1.68, but amnesic patients learning related word pairs
can be as good as normal performance (Cutthere was no Group X Delay interaction.
Thus, amnesic patients showed intact se- ting, 1978; Kinsbourne & Winocur, 1980;
mantic priming. Moreover, they should show Winocur & Kinsbourne, 1978; Winocur &
intact priming despite a severe impairment in Weiskrantz, 1976). An examination of these
free recall at both the immediate and the 2-hr four studies appears to account for the disdelay conditions. The priming effect decayed crepancy. In one study (Cutting, 1978), 10
after a 2-hr delay for amnesic patients and for highly related word pairs were presented, and
patients with Korsakoff syndrome and
alcoholic control subjects.
control subjects scored 91 % and 97% correct,
General Discussion
respectively. In this case, a statistical differMemory for related paired associates is a ence between amnesic patients and control
short-lived phenomenon in amnesic patients subjects might have been concealed by a ceiland can be explained by an intact, transient, ing effect.
activation process. Although amnesic paThe three studies by Winocur and his astients can activate preexisting associations, sociates (Kinsbourne & Winocur, 1980;
they lack the ability to establish new associa- Winocur & Kinsbourne, 1978; Winocur &
tions or to strengthen preexisting associa- Weiskrantz, 1976) used a design similar to
tions in a long-lasting way. Experiment 1 il- ours. Twelve related word pairs were prelustrated the inability of amnesic patients to sented once, twice, or four times. We preremember unrelated word pairs, despite in- sented 12 word pairs twice. The amnesic patact word activation as measured by a word- tients in their three studies performed
completion test. In Experiment 2A, amnesic similarly to the amnesic patients in our study.
patients showed good memory for related For example, when the lists were presented
word pairs at short study-test delays, though twice, the amnesic patients studied by Winocontrol subjects performed significantly bet- cur and Weiskrantz (1976, Experiment 1)
ter. In addition, the good memory of amnesic averaged 3.5 errors, and our amnesic patients
patients was short-lived, and performance averaged 3.6 errors (Experiment 2A, no
fell to baseline after a 2-hr delay. Control delay). The discrepancy in the two studies
subjects performed well above baseline at conies from the scores of the control subjects.
all delay conditions. Experiment 2a showed In their study, the control subjects averaged
that the forgetting of related word pairs fol- 2.2 errors, whereas our control subjects averlowed the same time course as the decay of aged only 0.7 errors. It should also be noted
word-completion performance. Experiment that in all of the published reports claiming
3 showed that amnesic patients performed intact paired-associate learning by amnesic
the same as control subjects when an inci- patients, the amnesic patients nevertheless
dental learning and test procedure was used. scored numerically lower than the control
Thus, the advantage of control subjects over subjects. Thus, it appears that at short retenamnesic patients at short delays could be at- tion intervals, amnesic patients show good
PAIRED-ASSOCIATE LEARNING AND PRIMING EFFECTS
paired-associate learning of related word
pairs, but their performance does not match
the level of control subjects.
The advantage of control subjects over
amnesic patients was found only with the
standard paired-associate learning instructions used in Experiments 2A and 2B. When
incidental learning and test instructions were
given (Experiment 3, see Figure 4), control
subjects and amnesic patients performed the
same. Specifically, the performance of amnesic patients was the same under both incidental and explicit instructions, but the performance of control subjects was raised by Using
explicit instructions. This finding is consistent with previous work showing that test instructions (explicit or incidental) can determine whether amnesic patients are impaired
or whether they perform normally (Graf et
al., 1984). Under incidental instructions, the
demand on declarative memory (i.e., organization and elaboration) is greatly reduced. In
this case, performance depends largely on activation. Under explicit instructions to learn,
control subjects gain an advantage because
their performance can depend on both declarative memory and activation.
One could suppose that the results reported here have been influenced by a tendency for control subjects tested under priming conditions to inhibit the production of
previously presented items, thereby lowering
their scores. If so, these findings would not
constitute evidence for preserved priming
ability in amnesia. One way to determine if
items are being inhibited in a priming task is
to determine if those items that a subject can
subsequently recall on a memory test are
items that are not produced in the priming
task. That is, compare the conditional probability of items being primed, given that they
were subsequently recalled, with the simple
probability of items being primed. Experiments 1, 2s, and 4 provide tests of this possibility because a given stimulus word was
tested in both the priming and the memory
tasks. As shown in Table 1, the conditional
probability of priming, given recall, was similar to the simple probability of priming. In
fact, in two of three comparisons, the proba*bility of priming, given recall, was slightly
higher than the simple probability of prim-
567
Table 1
Simple Probability (P) of Priming Versus
Conditional Probability of Priming, Given
Subsequent Recall
Experiment
P (Priming)
P (Priming|Recall)
1
.619
.354
.361
.770
.343
.415
2B
4
Note. In Experiments 1 and 2B, priming was tested by
word completion and was followed by paired-associate recall. In Experiment 4, priming was tested by word association and was followed by free recall. The scores refer
to the probability of response for control subjects, averaged
over the delay conditions for each experiment.
ing. A similar observation w,as reported by
Tulving, Schacter, and Stark (1982) when the
priming test preceded the memory test. Thus,
there was no evidence to suggest that control
subjects were reluctant to produce remembered items in the priming tests. They apparently followed instructions and produced the
first word that came to mind.
The results reported here are not completely consistent with the theoretical framework provided by Wickelgren (1979). One
prediction of his formulation is that amnesic
patients should not be deficient at strengthening already established associations (i.e.,
horizontal associations). Yet our findings in
Experiments 2A and 2B for amnesic patients
were as follows: Although memory for related word pairs can be strengthened to a considerable degree, performance fell to chance
after a 2-hr delay. Control subjects, however,
maintained a high level of performance at all
delays. As stated, Wickelgren's (1979) notion
of vertical versus horizontal associations is
not fully supported by these findings. What
have been termed horizontal associations
can apparently be strengthed only transiently.
Jacoby (1983; Jacoby & Dallas, 1981) presented a view that distinguishes between perceptual enhancement, which is thought to be
preserved in amnesic patients, and recognition memory, which is impaired. Perceptual
enhancement provides a perceptual basis for
remembering, that is, a fluency gained from
exposure to recently presented information.
This view was supported by word-comple-
568
ARTHUR P. SHIMAMURA AND LARRY R. SQUIRE.
tion and word-identification studies (Graf et
al., 1984; Jacoby & Witherspoon, 1982) in
which presentation of words biased performance on subsequent tests that involved the
same words. Experiment 4 of the present
study, however, demonstrated that amnesic
patients can be primed by presenting semantic associates of recently presented stimulus
words, that is, they can be primed by presenting cues that had never themselves been presented and for which there never was any
perceptual fluency. The activation process
that is preserved in amnesic patients is therefore not just perceptually driven. Activation
can occur by presenting information that is
semantically related to stimulus material;
thus, this process can also be conceptually
driven.
There are some similarities between the
priming tasks used here and others used previously (Feustel, Shiffrin, & Salasoo, 1983;
Jacoby & Dallas, 1981; McKoon & Ratcliff,
1979; Meyer & Schvaneveldt, 1971; Scarborough, Cortese, & Scarborough, 1977; Tulving et al., 1982). The lexical decision task is a
classic example. The reaction time to determine if a letter string is a word is faster when a
target string (e.g., NURSE) is preceded by a
semantically related priming stimulus (e.g.,
DOCTOR), this phenomenon was considered
to depend on a temporary activation process
that operates on preexisting elements in
memory (Anderson, 1976; Collins & Loftus,
1975; Loftus, 1973; Rozin, 1976). Some
studies have also reported that priming can
occur for both words and nonwords (Feustel
etal, 1983; Scarborough etal., 1977), a finding that raises questions about the nature of
the elements on which activation operates
(e.g., words, phonemes, or letters). In studies
where the time course of priming was examined, the phenomenon often decayed in the
minutes after word presentation and was
gone after a 2-hr delay (this study; Graf et al.,
1984; Loftus, 1973; Rozin & Diamond, cited
in Rozin, 1976). This time course has been
found for both normal subjects and amnesic
patients in tests of word completion and
word association and in the learning of related word pairs. It has also been suggested
that priming effects can occur over intervals
as long as a few days or a week (Jacoby, 1983;
Scarborough et al., 1977; Tulving et al.,
1982). More studies are needed to determine
if these long-lasting effects are similar to or
different from the relatively transient effects
observed here and in other studies. One important question is whether amnesic patients
can also exhibit long-lasting priming effects.
It is useful to compare the' priming effects
observed here with skill learning, which is
also preserved in amnesia. Both priming and
skill learning are presumed to operate on
preexisting elements in memory, and both
can be exhibited without subjects knowing
that their memory is being tested. In addition, the knowledge demonstrated by subjects in tests of priming and skill learning can
be expressed only in performance. The contents of memory are not known to the subject
explicitly and are not accessible to conscious
awareness. For these reasons, both priming
and skill learning have been considered to be
examples of knowing how or procedural
memory (Cohen, 1984; Squire & Cohen,
1984). Knowing-how is considered to be tied
to and expressible only through the activation of the particular processing structures or
procedures engaged by learning tasks.
There are, however, some differences between the activation of previously presented
material and skill learning. Activation exerts
transient effects, but a newly acquired skill
can be retained for months (Cohen & Squire,
1980; Kolers, 1976). Moreover, activation
operates on specific, preexisting elements or
processing structures, whereas skill learning
involves the tuning and sequencing of many
elements into a specific procedure. One func-;
tion of activation may be to increase accessi-1
bility to the particular elements needed for
skill learning to develop. In this way, transient activation may be a necessary and facilitating step in skill acquisition.
In summary, the process of activation can
account for findings of good performance by
amnesic patients on tests of word completion, word association, and related paired associates. Activation is a transient phenomenon that operates on and facilitates access to
preexisting knowledge structures. Activation
is preserved in amnesia and is therefore independent of the brain regions damaged in amnesia. Activation stands in contrast to the severe impairment of recall and recognition
that is the hallmark of the amnesic disorder.
PAIRED-ASSOCIATE LEARNING AND PRIMING EFFECTS
569
eral medial temporal lobe excision. Neuropsycholo.This impairment has been described as a failgia, 6, 255-265.
ure to acquire declarative memory. The disJ. (1978). A cognitive approach to Korsakoff's
tinction between activation and declarative Cutting,
syndrome. Cortex, 14, 485-495.
memory demonstrated here helps to charac- Drachman, D. A., & Arbit, J. (1966). Memory and the
terize how memory processes are organized
hippocampal complex. Archives of Neurology, 15,
52-61.
in the brain and illustrates the usefulness of
neuropsychological approaches to the study Ericson, R. C., & Scott, M. L. (1977). Clinical memory
testing: A review. Psychological Bulletin, 84, 1130of cognitive function.
1149.
References
Anderson, J. R. (1976). Language, memory, and
thought. Hillsdale, NJ: Erlbaum.
Baddeley, A. D- (1982a). Domains of recollection. Psychological Review, 89, 708-729.
Baddeley, A. D. (1982b). Implications of neuropsychological evidence for theories of normal memory. In
D. E. Broadbent & L. Weiskrantz (Eds.), Philosophical Transactions of the Royal Society of London (Vol.
298, pp. 59-72). London: The Royal Society.
Baddeley, A. D., & Warrington, E. K. (1970). Amnesia
and the distinction between long- and short-term
memory. Journal of Verbal Learning and Verbal Behavior, 9, 176-189.
Brierley, J. B, (1977). Neuropathology of amnesic states.
In C. W. M. Whitty & O. L. Zangwill (Eds.), Amnesia
(pp. 199-223). London: Buttersworths.
Brooks, D. N., & Baddeley, A. (1976). What can amnesic
patients learn? Neuropsychologia, 14, 111 -122.
Butters, N., & Cermak, L. S. (1980). Alcoholic Korsakoff's syndrome: An information processing approach
to amnesia. New York: Academic Press.
Cermak, L. S. (1982). Human memory and amnesia.
Hillsdale, NJ: Erlbaum.
Cermak, L. S., Lewis, R., Butters, N., & Goodglass, H.
(1973). Role of verbal mediation in performance of
motor tasks by Korsakoff patients. Perceptual and
Motor Skills, 37, 259-262.
Cohen, N. J. (1981). Neuropsychological evidence for a
distinction between procedural and declarative knowledge in human memory and amnesia. Unpublished
doctoral dissertation, University of California, San
Diego.
Cohen, N. J. (1984). Preserved learning capacity in amnesia: Evidence for multiple memory systems. In L.
Squire & N. Butters (Eds.), The neuropsychology of
memory (pp. 83-103). New York: Guilford
Press.
Cohen, N. J., &Corkin, S. (1981). The amnesic patient,
H. M.: Learning and retention of a cognitive skill.
Society for Neuroscience Abstracts, 7, 235.
Cohen, N. J., & Squire, L. R. (1980). Preserved learning
and retention of patternjanalyzing skill in amnesia:
Dissociation of knowing how and knowing that.
Science, 210, 207-210.
Cohen, N. J., & Squire, L. R. (1981). Retrograde amnesia and remote memory impairment. Neuropsychologia, 19, 337-356.
Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological
Review, 82, 407-428.
Corkin, S. (1968). Acquisition of motor skill after bilat-
Feustel, T. C., Shiffrin, R. M., & Salasoo, A. (1983).
Episodic and lexical contributions to the repetition
effect in word identification. Journal of Experimental
Psychology: General, 112, 309-346.
Graf, P., Squire, L. R., & Mandler, G. (1984). The information that amnesic patients do not forget. Journal of
Experimental Psychology: Learning, Memory, and
Cognition, 10, 164-178.
Hirst, W. (1982). The amnesic syndrome: Descriptions
and explanations. Psychological Bulletin, 91, 435460.
Huppert, F. A., & Piercy, M. (1979). Normal and abnormal forgetting in organic amnesia: Effects of locus of
lesion. Cortex, 15, 385-390.
Jacoby, L. L. (1983). Perceptual enhancement: Persistent effects of an experience. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 9,
21-38.
Jacoby, L. L., & Dallas, M. (1981). On the relationship
between autobiographical memory and perceptual
learning. Journal of Experimental Psychology: General, 110, 306-340.
Jacoby, L. L., & Witherspoon, D. (1982). Remembering
without awareness. Canadian Journal of Psychology,
32, 300-324.
Jenkins, J. J. (1970). 1952 Minnesota word association
norms. In L. Postman & G, Keppel (Eds.), Norms of
word association (pp. 9-32). New York: Academic
Press.
Jones, M. K. (1974). Imagery as a mnemonic aid after
left temporal lobectomy: Contrast between materialspecific and generalized memory disorders. Neuropsychologia, 12, 21-30.
Kinsbourne, M., & Winocur, G. (1980). Response competition and interference effects in paired-associate
learning by Korsakoff amnesics. Neuropsychologia,
75,541-548.
Kolers, P. A. (1976). Reading a year later. Journal of
Experimental Psychology: Human Learning and
Memory, 2, 554-565.
Loftus, E. F. (1973). Activation of semantic memory.
American Journal of Psychology, 86, 331-337.
Mandler, G. (1979), Organization and repetition: Organizational principles with special reference to rote
learning. In L.-G. Nilsson (Ed.), Perspectives in memory research (pp. 293-327). Hillsdale, NJ: Erlbaum.
Mandler, G. (1980). Recognizing: The judgment of previous occurrence. Psychological Review, 87, 252271:
Mair, W. G. P., Warrington, E. K., & Weiskrantz, L.
(1982). Memory disorder in Korsakoff's psychosis: A
neuropathological and neuropsychological investigation of two cases. Brain, 102, 749-783.
Mayes, A. R., Meudell, P. R., & Neary, D. (1978). Must
570
ARTHUR P. SHIMAMURA AND LARRY R. SQUIRE
amnesia be caused by either encoding or retrieval disorders? In M. M. Gruenberg, P. E. Morris, & R. N.
Sykes (Eds.), Practical aspects of memory (pp. 712719). London: Academic Press.
McKoon, G., & Ratcliff, R. (1979). Priming in episodic
and semantic memory. Journal of Verbal Learning
and Verbal Behavior, 18, 463-480.
Meyer, D. E., & Schvaneveldt, R. W. (1971). Facilitation
in recognizing pairs of words: Evidence of a dependence between retrieval operations. Journal of Experimental Psychology, 90, 227-234.
Milner, B. (1962). Les troubles de la memoire accompagnant des lesions hippocampiques bilaterales
[Memory impairment accompanying bilateral hippocampal lesions]. In Physiologie de I'hippocampe.
Paris: Centre National de la Rechesche Scientifique.
Mishkin, M., Malamut, B., & Bachevalier, J. (1984).
Memories and habits: Two neural systems. In J. L.
McGaugh, G. Lynch, & N. M. Weinberger (Eds.),
The Neurobiology of Learning and Memory (pp. 65 77). New York: Guilford Press.
Mortensen, E. L. (1980). The effects of partial information in amnesic and normal subjects. Scandanavian
Journal of Psychology, 21, 75-82.
Moscovitch, M. (1982). Multiple dissociations of function in amnesia. In L. S. Cermak (Ed.), Human memory and amnesia (pp. 337-370). Hillsdale, NJ: Erlbaum.
O'Keefe, J., & Nadel, L. (1978). The hippocampus as a
cognitive map. London: Oxford University Press.
Rozin, P. (1976). The psychobiological approach to
human memory. In M. R. Rosenzweig & E. L. Bennett (Eds.), Neural mechanisms of learning and memory (pp. 3-46). Cambridge, MA: MIT Press.
Scarborough, D. L., Cortese, C, & Scarborough, H.
(1977). Frequency and repetition effects in lexical
memory. Journal of Experimental Psychology:
Human Perception arid Performance, 3, 1-17.
Squire, L. R. (1981). Two forms of human amnesia: An
analysis of forgetting. Journal of Neurosciences, 1,
635-640.
Squire, L. R. (1982a). Comparisons between forms of
amnesia: Some deficits are unique to Korsakoff syndrome. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 560-571.
Squire, L. R. (1982b). The neuropsychology of human
memory. Annual Review of Neuroscience, 5, 241273.
Squire, L. R. (in press-a). The neuropsychology of memory. In P. Marler & H. Terrace (Eds.), The biology of
learning. Berlin: Springer-Verlag.
Squire, L. R. (in press-b). The neuropsychology of memory dysfunction and its assessment. In I. Grant & K.
M. Adams (Eds,), Neuropsychological assessment of
neuropsychiatric disorders. New York: Oxford University Press.
Squire, L. R., & Cohen, N. J. (1984). Human memory
and amnesia. In J. L. McGaugh, G. Lynch, & N. M.
Weinberger (Eds.), The Neurobiology of Learning and
Memory (pp. 3-64). New York: Guilford Press.
Squire, L. R., Wetzel, C. D., & Slater, P. C. (1978). Anterograde amnesia following ECT: An analysis of the
beneficial effect of partial information. Neuropsychologia, 16, 339-347.
Stern, L. D. (1981). A review of theories of human amnesia. Memory & Cognition, 9, 247-262.
Talland, G. (1965). Deranged memory. New York: Academic Press.
Thorndike, E. L., & Lorge, I. (1944). The teacher's word
book of 30,000 words, New York: Columbia University, Teacher's College Press.
Tulving, E., Schacter, D. L., & Stark, H. A. (1982). Priming effects in word-fragment completion are independent of recognition memory. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 8,
336-342.
Victor, M., Adams, R. D., & Colling G. H. (1971). The
Wemicke-Korsakoff syndrome. Philadelphia: Davis.
Warrington, E. K., & Weiskrantz, L. (1968). A neW
method of testing long-term retention with special reference to amnesic patients. Nature, 217, 972-974.
Warrington, E. K., & Weiskrantz, L. (1970). The amnesic syndrome: Consolidation or retrieval? Nature, 228,
628-630.
Warrington, E. K., & Weiskrantz, L. (1974). The effect
of prior learning on subsequent retention in amnesic
patients. Neuropsychologia, 12, 419-428.
Warrington, E. K., & Weiskrantz, L. (1978). Further
analysis of the prior learning effect in amnesic patients. Neuropsychologia, 16, 169-177.
Warrington, E. K., & Weiskrantz, L. (1982). Amnesia:
A disconnection syndrome? Neuropsychologia, 20,
233-248.
Weiskrantz, L. (1982). Comparative aspects of studies of
amnesia. In D. E. Broadbent & L. Weiskrantz (Eds.),
Philosophical Transactions of the Royal Society of
London (Vol. 298, pp. 97-109). London: The Royal
Society.
Wetzel, C. D., & Squire, L. R. .(1982). Cued recall in
anterograde amnesia. Brain and Language, 15, 7081.
Wickelgren, W. A. (1979). Chunking and consolidation:
A theoretical synthesis of semantic networks, configuring in conditioning, S-R versus cognitive learning, normal forgetting, the amnesic syndrome, and the
hippocampal arousal system. Psychological Review,
86, 44-60.
Winocur, G., & Kinsbourne, M. (1978). Contextual cueing as an aid to Korsakoff amnesics. Neuropsychologia, 76,671-682.
Winocur, G., & Weiskrantz, L. (1976), An investigation
of paired-associate learning in amnesic patients. Neuropsychologia, 14, 97-110.
Zangwill, O. L. (1977). The amnesic syndrome. In C. W.
M. Whitty & O. L. Zangwill (Eds.), Amnesia (2nd ed.)
(pp. 104-117). London: Buttersworth.
Zola-Morgan, S., Cohen, N. J., & Squire, L. R. (1983).
Recall of remote episodic memory in amnesia. Neuropsychologia, 21, 487-500.
Received March 17, 1984
Revision received June 1, 1984