INTRODUCTION
Self-performed tasks (SPTs) require participants to listen to action
phrases and then perform the actions. It has been demonstrated that SPTs
are remembered better than verbal tasks (VTs) in which participants simply
listened to action phrases (for review, see Engelkamp,
1998; Engelkamp & Zimmer, 1994). The
enactment effect has received considerable attention from memory
researchers, particularly because several variables known to affect recall
of verbal information exert little influence on SPT recall.
Larsson and Ronnberg (1987) found that
persons with traumatic brain injury showed better recall for SPTs than for
VTs, despite poorer overall recall in the persons with brain injury than
in control participants. Similarly, Mimura et al. (1998) applied the SPT paradigm to persons with amnesia
showing a severe deficit in explicitly remembering a prior event. Persons
with alcoholic Korsakoff's syndrome, as well as control participants
matched by age and educational background, were asked to learn action
phrases under SPT and under VT conditions. Although the persons with
Korsakoff's syndrome recalled and recognized fewer items than the
control participants, a comparable memory advantage of SPTs over VTs was
confirmed in both participant groups.
These findings are noteworthy from the perspective of rehabilitation
of persons with memory disorders including Korsakoff's syndrome.
Persons with Korsakoff's syndrome typically present with a chronic,
severe amnesic state following Wernicke's encephalopathy secondary to
thiamine deficiency. The most common underlying cause is chronic alcohol
consumption although nonalcoholic etiologies also are recognized (Kopelman, 1995). Despite their severely impaired
ability to retain and/or recall relevant information, persons with
Korsakoff's syndrome are known to consistently acquire new knowledge
to a limited extent by means of appropriate cognitive rehabilitation.
Specifically, encoding strategies such as the errorless learning approach
and the method of vanishing cues have been shown to improve learning of
new items (Komatsu et al., 2000). This
preservation is assumed to be dependent on intact memory systems.
Enactment during encoding is another cognitive intervention that may rely
on preserved memory systems in persons with Korsakoff's syndrome.
Such issues carry practical implications for effective memory
rehabilitation. This is because learning through enactment is expected to
remedy or alleviate memory impairment, which causes difficulty leading an
independent life. It remains unresolved, however, whether the memory
advantage of SPTs over VTs, found in persons with memory disorders, is
attributable to enactment during encoding, because there are only a few
studies assessing the enactment effect in such participants. Therefore, we
examined applicability of the SPT paradigm to memory rehabilitation for
participants with amnesia related to Korsakoff's syndrome.
Two experiments were designed to elucidate two points. The first
problem involved the role of real objects in SPT memory. The typical SPT
paradigm involves two different kinds of items, namely, those with real
objects, such as “put the matches in the box,” and those
without real objects, such as “raise your arm.” Yet in
previous studies (Larsson & Ronnberg, 1987;
Mimura et al., 1998), memory-impaired
individuals were tested for SPT memory exclusively by providing real
objects during enactment, whereas no real objects were presented under the
VT condition. The benefit from SPTs observed in memory-disordered persons
thus could be ascribed to selective use of real objects rather than
enactment during encoding. In Experiment 1, we explored whether SPT
advantage is observable in Korsakoff and control participants even when
enactment is performed without real objects.
The second problem emerges from a practical perspective. In the
typical SPT paradigm, participants are required to perform actions
described verbally by an experimenter. As pointed out by Lichty et al.
(1988), real-world performance of activities is
rarely dictated by another person, so the SPT paradigm is difficult to
apply to actual as opposed to experimental situations, such as in memory
rehabilitation. Engelkamp & Zimmer (1994)
has suggested that self-generation should be used by presenting objects
and asking participants to generate an appropriate action for the object.
Methodologically, the typical SPT paradigm and the Self-generation-SPT
paradigm each have specific advantages and disadvantages. The typical SPT
paradigm permits the experimenter to control the content of actions; in
the Self-generation-SPT paradigm, the actual actions rely on the
participant's associations, which may preclude meaningful comparison
with actions performed under different experimental conditions. The
enactment effect in the typical SPT paradigm is demonstrated as an
additional effect of performing an action, as compared with simply
listening to description of the action. Therefore, one should examine
whether persons with amnesia show an enactment effect not only in the
typical SPT paradigm but also in the Self-generation-SPT paradigm. In
Experiment 2 we addressed this issue.
EXPERIMENT 1
In Experiment 1, persons wth Korsakoff's syndrome and nonamnesic
alcoholic participants were presented with brief action phrases that
involved no extra-environmental objects. Under the VT condition,
participants only listened to descriptions of the actions and tried to
memorize them; in the SPT condition, they performed the described actions.
They then were tested by free-recall and recognition tests. If the memory
advantage of SPTs over VTs found in persons with amnesia (Mimura et al., 1998) was a result of enactment during
encoding, a similar effect would be expected to emerge from actions
without extra-environmental objects.
Another possibility, however, required consideration. According to a
multimodal encoding hypothesis (Backman et al.,
1986), performing actions leads to multimodal and contextually rich
experiences incorporating features such as color, texture, and shape. The
multimodal features are absent in VTs in which the presentation is
typically unimodal and the number of features is restricted to verbal
aspects of the actions. The presence of real objects during SPTs is
assumed to enhance multimodal encoding and thus to contribute to the
memory benefit from SPTs over VTs. Previous studies attempted to clarify
the effect of presenting real objects in subsequent SPT recall in
comparison with enactments without real objects. However, these results
were conflicting (e.g., Cohen et al., 1987;
Earles, 1996; Norris &
West, 1991). Engelkamp and Zimmer (1997)
demonstrated that although use of real objects during encoding improved
subsequent recall, this improvement was significantly independent of the
enactment effect. That is, introducing real objects enhanced not only the
recall of SPTs but also that of VTs; indeed, the positive effect was
considerably more pronounced in VTs than in SPTs. One therefore may argue
that the amnesic persons' memory advantage for SPTs (Mimura et al., 1998) can be attributable to an
experimental protocol in which real objects are presented in SPTs but not
in VTs.
Method
Design
The study design was a 2 × 2 mixed factorial. The
between-participants factor was group (persons with Korsakoff's
syndrome vs. nonamnesic alcoholics), and the within-participants
factor was study condition (SPTs vs. VTs).
Research Participants
Two groups of participants participated in the experiment. The first
group consisted of eight Japanese male persons with alcoholic
Korsakoff's syndrome who either reported periodically to a memory
clinic at Komagino Hospital, Tokyo, Japan, for follow-up examination or
who remained hospitalized at the same hospital. All persons had a history
of chronic alcoholism and were more than 1 year post-diagnosis of
Korsakoff's syndrome. They were disoriented to place and time, were
unable to recall day-to-day events, and showed varying degrees of
retrograde amnesia. The second group included eight Japanese male chronic
nonamnesic alcoholics who were recruited from among patients hospitalized
at the Alcohol Treatment Unit of the same hospital for the alcoholism
rehabilitation program. The structured clinical interview for
DSM-III-R (Spitzer et al., 1990) was
used to diagnose alcohol dependence. None of the alcoholic persons
evidenced any signs of other neurologic or psychiatric illness, or of
memory problems. Patients were excluded from the study if they had a
history of liver cirrhosis, diabetes mellitus, coronary artery disease,
cerebrovascular disease, or head injury. Patients with a history of other
drug dependence or of a major psychiatric illness (e.g., schizophrenia,
affective disorder) were also excluded. The persons with Korsakoff's
syndrome [Mean age (standard deviation, SD) + Mean education
(SD) = 53.8 (7.9) + 12.0 (2.9)] and nonamnesic alcoholic
participants [Mean age (SD) + Mean education (SD) =
51.1 (8.2) + 11.9 (2.6)] did not significantly differ in terms of
age, gender, ethnicity, or educational background.
Persons with Korsakoff's syndrome were evaluated with the Wechsler Adult Intelligence Scale–Revised
(WAIS–R) and Wechsler Memory Scale–Revised (WMS–R)
(Wechsler, 1987) as part of an initial
neuropsychological assessment. Mean scores (followed by SD) for
verbal and performance IQ on the WAIS–R were 95.6 (12.3) and 90.8
(14.3), and mean scores (SD) for attention/concentration and
delayed recall indices on the WMS–R were 88.6 (13.5) and 58.2
(18.1). None of the Korsakoff participants showed difficulty in
understanding auditory and/or visual instructions of the WAIS–R
and WMS–R. In addition, they were examined with subtests of the
Standard Language Test of Aphasia and the Standard Performance Test of
Apraxia (Japan Society for Higher Brain Dysfuntion, 1997, 2001, respectively).
None of the Korsakoff participants showed evidence of difficulty in
spontaneous speech, naming, or limb praxis.
Materials
Forty-eight action phrases were selected as stimulus items. Each item
consisted of a body-related action that did not involve any external
objects (e.g., raise your arm). These action phrases were divided into
four sets of 12 items (see Appendix). For each participant, two of the
four sets were presented as study lists under the SPT and VT conditions,
respectively; the remaining two sets were used as nonstudied distractors
for the recognition tests. The four sets of 12 items were rotated through
four experimental conditions orthogonally combining study condition (SPT
vs. VT) and item type (studied vs. nonstudied). Thus,
each set represented each of these combinations an equal number of times
across participants in each participant group. Four additional action
phrases involving no external objects were chosen as uniform practice
items presented at the beginning of the two study phases.
Procedure
Participants were tested individually. Each person participated in two
study-test sessions, separated by at least 1 week (mean interval: 3.3
weeks in persons with Korsakoff's syndrome and 3.0 weeks in
nonamnesic alcoholic participants). Each session corresponded to either of
two different study conditions: SPT or VT. The order of the SPT and VT
conditions between the two sessions was counterbalanced in each group of
participants. All participants completed both sessions. All stimulus items
were presented using an Apple Macintosh computer.
In the study phase, participants were instructed that a series of
action phrases would be presented auditorily and that memory tests
subsequently would be given. After two practice trials a total of 12 study
items were presented in random order at a rate of 8 seconds per item via
computer-generated speech. Under the VT condition, the participants were
required to listen to the action phrases and to memorize them. In the SPT
condition, they were asked to perform the described actions. After the
study phase, the test phase followed. A free-recall test then was
administered, in which participants were asked to recall as many phrases
as possible irrespective of presentation order. Following the completion
of the free-recall test, participants were given a recognition test in
which 12 study items and 12 distractor items were visually and auditorily
presented in random order.
Results
Free-recall data were scored on the basis of two different criteria:
strict and lenient. According to the strict criterion, only responses
identical to the corresponding action phrases in the study list were
accepted as correct. The lenient criterion additionally permitted synonyms
for the actions to be counted as correct. The strict and lenient criteria
were thought to reflect verbatim and gist recall performances,
respectively. Table 1 shows the mean proportions
of correctly recalled items. Data based on the strict criterion
demonstrated that recall performance of the nonamnesic alcoholic group
(M = .43) was higher than that of the Korsakoff group (M
= .24), and that both groups showed better recall for SPTs (M =
.43) than for VTs (M = .25). These observations were supported by
a 2 (group: Korsakoff vs. alcoholic participants) × 2
(study: SPT vs. VT) mixed ANOVA. A significant main effect of
group was present: F(1,14) = 7.84, MSE = .035,
p < .05, η2 = .36. A significant main effect
of study was noted as well: F(1,14) = 20.79, MSE = .179,
p < .001, η2 = .60. The interaction between
the two factors did not approach significance [F(1,14) =
1.40, observed power = .20]. A 2 × 2 ANOVA was also performed
to analyze recall data scored using the lenient criterion, resulting in
similar patterns of results: main effect of group, F(1,14) =
5.10, MSE = .031, p < .05, η2 = .16,
and main effect of study, F(1,14) = 15.25, MSE = .012,
p < .001, η2 = .52, both were significant
without interaction between them (F < 1, observed power =
.06).
Table 2 shows mean proportions of
“hits” and “false alarms” in recognition memory.
Based on the signal detection theory, d′ scores also were
calculated as a discrimination index, as shown at the bottom of Table 2. These d′ scores were submitted
to a 2 (group) × 2 (study) ANOVA. A significant main effect of group
was evident: F(1,14) = 12.04, MSE = .376, p
< .01, η2 = .46. This indicated that recognition
performance of the nonamnesic alcoholic group (M = 2.27) was
superior to that of the Korsakoff group (M = 1.52). The main
effect of study also was significant [F(1,14) = 17.44,
MSE = .369, p < .001, η2 = .56],
reflecting that both of the participant groups showed a recognition
advantage for SPTs (M = 2.34) over VTs (M = 1.45).
Interaction between group and study did not approach significance
(F < 1, observed power = .13).
Experiment 1: Recognition
Discussion
Results of Experiment 1 replicated previous findings concerning SPT
memory in persons with Korsakoff's syndrome (Mimura et al., 1998). As evaluated using free-recall
and recognition tests, the control group exhibited better memory
performance following SPTs than following VTs, demonstrating a typical
enactment effect. The enactment effect was also confirmed in the persons
with Korsakoff's syndrome, although their overall free-recall and
recognition performance was poorer than that in the nonamnesic alcoholic
group.
The possibility that the memory advantage of SPTs over VTs in
memory-impaired individuals might be attributable to presentation of real
objects rather than enactment has been difficult to rule out. In previous
studies (Larsson & Ronnberg, 1987; Mimura et al., 1998), memory-impaired individuals were
required to learn SPTs with real objects and VTs with no real objects. In
contrast, the materials used for SPTs and VTs in Experiment 1 were action
phrases involving no real objects; the result suggested that enactment
enhances recall and recognition performance by persons with
Korsakoff's syndrome, even if no real objects are presented.
EXPERIMENT 2
In Experiment 2 we further explored the enactment effect in persons
with Korsakoff's syndrome by adding a self-generation procedure to
the SPT condition. Participants under the Verbal generation-SPT condition
were presented with an object noun and were required to verbally generate
and then perform an action. Memory performance under the Verbal
generation-SPT condition was compared with that under a Read condition in
which participants were to memorize an object name by saying it aloud. As
described earlier, one of the problems concerning a Self-generation-SPT
paradigm is that the content of actions performed during the study depends
on the participant's generation, which impedes comparison of memory
performance across different experimental conditions. To parallel the
items to be remembered, we asked participants to remember the object nouns
rather than the names of actions. In addition to the Verbal generation-SPT
and Read conditions, another study condition, called the Verbal generation
condition, was designed to detect the contribution of verbal generation
without action to memory performance. The Verbal generation and Read
conditions were assumed to reflect qualitatively different types of verbal
encoding, deep conceptual elaboration and simple repetition,
respectively.
Verbally generating an action was expected to enhance memory
performance of persons with Korsakoff's syndrome. Generation and
enactment are assumed to rely on similar cognitive processes, since both
tasks require good item-specific encoding with little reliance on
relational encoding (Engelkamp & Zimmer,
1997). Persons with amnesia (Mimura et al.,
1998) and persons with Alzheimer's Disease (AD) (Hutton et al., 1996; Karlsson et
al., 1989; Lekeu et al., 2002) both have
been shown to benefit from SPT encoding. Persons with AD also exhibited
memory improvement following generation of a target (Fleischman et al., 1995; Multhaup
& Balota, 1997) or generation of a retrieval cue during
encoding (Lipinska et al., 1994). Persons with
amnesia therefore would be expected to benefit from generation during
encoding.
Investigating SPT memory of persons with Korsakoff's syndrome
under the Self-generation-SPT paradigm is of great significance from the
perspective of memory rehabilitation. In most real-world situations,
actions to be performed only rarely are predetermined by another person.
For this reason, the Verbal generation-SPT condition is considered a more
ecologically valid training situation than the standard SPT condition.
Moreover, both generating and performing an action phrase, compared with
solely listening to the phrase, would be expected to promote
patients' active participation in the learning environment. Wilson et
al. (1994) have argued that active participation
of amnesic persons is an important prerequisite for successful memory
rehabilitation.
To our knowledge, however, little is known about SPT memory in persons
with amnesia under the Self-generation-SPT paradigm. A hint may be found
in the results of a study by Nilsson and Cohen (1988), in which college undergraduates were given a
real object for which they were requested to generate an action and then
to perform it. The recall performance under the Verbal generation-SPT
condition was compared with that under the SPT condition without
generation. Analogously, a verbal generation effect was also examined
under VT conditions. Verbal generation was found to enhance the recall of
VTs; in contrast, no recall advantage of verbal generation was found for
SPTs. Nilsson and Cohen concluded that because sufficient item-specific
encoding had been induced by enactment under the SPT condition, little
room remained for further improvement when the verbal generation procedure
was added under the Verbal generation-SPT condition. One may argue that
the overall memory performance of persons with amnesia is decidedly poorer
than that of college undergraduates. Accordingly, room for further
improvement of memory performance may remain in persons with amnesia.
Persons with Korsakoff's syndrome thus were expected to show better
memory performance under the Verbal generation-SPT condition than that
under the Verbal generation or Read condition. In Experiment 2, we
examined whether enhancement by enactment was found under the Verbal
generation-SPT paradigm by comparing persons with Korsakoff's
syndrome, age-matched nonamnesic alcoholics, and young college students as
research participants.
Method
Design
The design was a 3 × 3 mixed factorial with group (Korsakoff
vs. nonamnesic alcoholics vs. young controls) as a
between-subjects factor and study (Verbal generation-SPT vs.
Verbal generation vs. Read) as a within-subjects factor. Each
participant was tested for both free recall and recognition.
Research Participants
Included in the experiment were 12 persons with Korsakoff's
syndrome, 12 chronic nonamnesic alcoholics, and 18 young adults. All
participants were native Japanese. Because the persons with
Korsakoff's syndrome and nonamnesic alcoholic individuals in
Experiment 2 were recruited at Komagino Hospital 3 to 9 months after the
completion of Experiment 1, these participants partially overlapped those
in Experiment 1; five of the persons with Korsakoff's syndrome in
Experiment 1 also participated in Experiment 2. No signs of additional
neurologic or psychiatric illness were present. Mean scores (SD)
of persons with Korsakoff's syndrome for verbal and performance IQ on
the WAIS–R were 92.6 (11.5) and 89.1 (12.8), and mean scores
(SD) for attention/concentration and delayed recall indices
on the WMS–R were 90.9 (11.5) and 55.9 (16.8). The nonamnesic
alcoholics were selected from the same population as those in Experiment
1, and two of them also had participated in Experiment 1. Persons with
Korsakoff's syndrome [Mean age (SD) + Mean education
(SD) = 54.3 (7.3) + 13.2 (2.8)] and nonamnesic alcoholics
[Mean age (SD) + Mean education (SD) = 52.5 (7.9) +
12.9 (2.6)] were comparable in age, gender, ethnicity, and
educational background. The young adults were undergraduate students from
Shinshu University (Mean age = 21.2 years) who received course credit for
their participation.
Materials
A total of 120 nouns designating an object served as stimulus items.
The mean frequencies of these nouns were 0.05 per million (SD =
0.03). The 120 items were divided into six sets of 20 items. For each
participant, three of the six sets were assigned as study lists, whereas
the remaining three sets were used as distractor items presented in the
recognition tests. Assignment of study lists and the assignment of study
conditions were counterbalanced across participants. Nine additional nouns
representing object names were chosen as standardized practice items
presented at the beginning of the three study conditions.
Procedure
Participants were examined individually. Each person participated in
three sessions, separated by at least 1 week. Mean intervals between the
first and second sessions and those between the second and third sessions,
respectively, were 3.2 weeks and 3.4 weeks in persons with
Korsakoff's syndrome and 3.1 weeks and 3.2 weeks in nonamnesic
alcoholic participants. The order of study conditions was counterbalanced
across participants. All participants completed all three sessions.
Each session consisted of study and test phases. In the Verbal
generation-SPT condition, participants were instructed to generate an act
that could be performed with the object whose name was presented, first by
describing this act aloud and then by performing the act. In the Verbal
generation condition, participants were instructed to generate an act that
they could perform with the object named, and then to speak the phrase
that they had generated, but were not requested to perform the act. In the
Read condition, participants were instructed to memorize an object name by
saying it aloud four times.
After the instructions, a study phase followed. Three practice items
were presented visually via an Apple Macintosh computer.
Participants in the Verbal generation-SPT and Verbal generation conditions
were required to generate the act, but the act was actually performed only
in the former condition. If they showed difficulty in generating an act
during the practice trials, they were given possible examples of action
verbs and were asked to actually perform one of the actions. A list of 20
study items was then visually presented at a rate of 10 seconds per item.
Pretest clinical observation established that none of the persons with
Korsakoff's syndrome had difficulties in speech/language or
ideomotor praxis. All participants successfully obeyed verb generation and
enactment commands in the Verbal generation-SPT and Verbal generation
conditions.
An immediate free recall test was given after presentation of each
list of items. A recognition test was presented as the final memory test.
A total of 24 test items including 12 study and 12 distractor items were
presented singly in random order. To avoid a ceiling effect, participants
in the young adult group were given a retention interval that lasted about
15 minutes, during which they played a computer game.
Results
Mean proportions of items recalled under the three study conditions
are shown in Table 3. A 3 (group: Korsakoff
vs. nonamnesic alcoholics vs. young controls) × 3
(study: Verbal generation-SPT vs. Verbal generation vs.
Read) mixed ANOVA demonstrated a significant main effect of group
[F(2,39) = 44.65, MSE = .036, p < .001,
η2 = .70] and also of study [F(2,78) =
7.68, MSE = .012, p < .001, η2 =
.17]. No interaction occurred between the two factors
[F(4,78) = 1.43, MSE = .012, observed power =
.43]. Post hoc comparisons using Fisher's least
significant difference (LSD) test with an alpha of .05 indicated that the
young control group (M = .49) recalled more than the nonamnesic
alcoholics (M = .25) or the persons with Korsakoff's
syndrome (M = .12), although the difference between the
nonamnesic alcoholics and the persons with Korsakoff's syndrome did
not reach significance. Post hoc analysis of the main effect of
study, using Fisher's LSD with an alpha of .05, demonstrated that
recall performance in the Verbal generation-SPT condition (M =
.38) was better than that in the Verbal generation and Read conditions
(Ms = .29).
Table 4 displays hit and false alarm rates,
together with d′ scores. Discrimination ability was
analyzed by entering the d′ scores into a 3 (group) ×
3 (study) ANOVA. Significant main effects were seen for group
[F(2,39) = 15.46, MSE = 1.120, p <
.001, η2 = .44], and also for study
[F(2,78) = 16.77, MSE = .309, p < .001,
η2 = .30]. Post hoc analysis using
Fisher's LSD with an alpha of .05 demonstrated that d′
scores of the young controls (M = 2.95) and nonamnesic alcoholics
(M = 2.80) did not differ, but were superior to the score in the
Korsakoff group (M = 1.74). In the Verbal generation-SPT
(M = 2.81) and Verbal generation (M = 2.70) conditions,
d′ scores did not significantly differ, but they exceeded
those for the Read condition (M = 2.16). No significant
interaction was detected between the two factors (F < 1,
observed power = .25), confirming that the advantage of the Verbal
generation-SPT and Verbal generation conditions over the Read condition
was observable across the three participant groups.
Experiment 2: Recognition
Discussion
Previous findings predicted that enactment and verbal generation might
additively enhance item-specific processing, leading to memory
improvement—especially in persons with Korsakoff's syndrome,
whose episodic memory is known to be extremely poor. Analysis of recall
performance demonstrated that verbally generating an action is not
sufficient; performing the action is necessary to produce an advantageous
effect over reading an object name. Our finding that a Verbal generation
effect failed to emerge for recall, even in young adults, disagreed with
that of Nilsson and Cohen (1988). Procedural
differences may have contributed to this inconsistency. In Experiment 2,
an object name was consistently presented to participants across all study
conditions, whereas in Nilsson and Cohen's experiment real objects
were presented under the SPT condition but not under the VT condition.
Also, in our Experiment 2, participants were asked to remember object
names rather than actions that could be performed with the objects; in
contrast, to-be-remembered items were action phrases in Nilsson and
Cohen's experiment in which the participants under the VT condition
were required to remember action phrases yoked to those under the SPT
condition.
Results showed a different pattern when they were assessed by
recognition memory. Across all participant groups, the typical generation
effect showed that the d′ score under the Verbal generation
condition was superior to that under the Read condition. Yet, further
improvement did not emerge when enactment in addition to generation was
introduced under the Verbal generation-SPT condition. Hit rates of
comparison groups under the Verbal generation condition revealed a ceiling
effect, indicating that further improvement is difficult even when
enactment is additionally required. In contrast, hit rates of persons with
Korsakoff's syndrome under the Verbal generation condition did not
reach a ceiling, but nevertheless the addition of enactment did not
enhance the hit rate (Verbal generation-SPT condition).
A conventional hypothesis premised that verbal generation and
enactment additively emphasize item-specific processing, which is regarded
as the critical determinant for successful recognition. Engelkamp (1998) has proposed a more advanced view of
item-specific encoding, where item-specific information should be
differentiated further depending on different subsystems involved. More
recently, functional brain imaging studies have suggested that object
concepts may be represented, in part, by distributed networks of discrete
cortical regions that parallel the organization of sensory (e.g., physical
form, color) and motor systems (Martin & Chao,
2001). Specifically, access to motion-related attributes is crucial
for identification of tools. In addition, motor-related brain areas appear
to be activated without actually using tools. Even generating action words
to tools (Grafton et al., 1997) or imaging
manipulating objects with the right hand (Gerardin et
al., 2000) also result in ventral premotor activation.
In Experiment 2, verbal generation of an action is assumed to
primarily activate the conceptual type of motor-related information;
performing an action is thought to lead to further activation of motor
programs, as well as activation of the word nodes and the action concept.
Addition of SPTs under the Verbal generation-SPT condition, as compared
with the Verbal generation condition, might make no further contribution
to conceptual motor-related processing, but still make a key contribution
to stable motor-specific processing. In free recall, participants can
benefit from enhancement of motor-specific encoding, which enables them
during retrieval to directly access the to-be-remembered items (Zimmer et al., 2000). In recognition, however,
conceptual motor-related processing may suffice, with actual
motor-specific information adding rather little.
GENERAL DISCUSSION
Results of the two experiments reported here suggest that persons with
Korsakoff's syndrome can benefit from enactment during encoding under
a variety of learning conditions. Whether the benefit from SPTs previously
found in persons with Korsakoff's syndrome (Mimura
et al., 1998) is attributable to enactment or use of real objects
during encoding has been unclear. In our Experiment 1, persons with
Korsakoff's syndrome and nonamnesic comparison participants
demonstrated a memory advantage of SPTs over VTs when action phrases used
in both tasks did not involve any extra-environmental objects. We
estimated effect sizes of SPT recall and recognition advantage over VT in
our Experiment 1 and those of Mimura et al.'s experiment (1998). Effect sizes (Cohen's d) were
calculated by dividing the mean differences (SPT − VT) by the pooled
standard deviations (Cohen, 1988). In our
Experiment 1, the effect sizes of SPT recall were 1.41 for persons with
Korsakoff's syndrome and 0.78 for the nonamnesic comparisons; almost
comparable values were obtained in Mimura et al.'s experiment
(d = 1.84 for Korsakoff; d = 0.76 for comparisons). In a
similar vein, the effect sizes of SPT recognition (d = 1.04 for
Korsakoff; d = 2.02 for comparisons) in our Experiment 1 were
also comparable to those (d = 1.02 for Korsakoff; d =
2.10 for comparisons) of Mimura et al.'s study. The analysis of the
effect sizes hence suggests that an enactment effect occurs to a similar
degree irrespective of the presence or absence of real objects during
encoding.
Experiment 2 sought to examine whether persons with Korsakoff's
syndrome exhibit an enactment effect under yet another condition in which,
preceding enactment during encoding, they were required to verbally
generate any action that could be carried out with the object designated
by its visually presented name. Compared with the Read condition, the
Verbal generation-SPT condition enabled persons with Korsakoff's
syndrome, as well as nonamnesic young and elderly comparison participants,
to improve free-recall of object names. When rich environmental support is
available, such as presentation of copy cues in a recognition memory test,
either generation or enactment alone appears to be sufficient to enhance
memory performance; in contrast, when environmental support is minimal, as
in a free-recall test, generation and enactment are thought to contribute
additively to memory improvement.
These findings carry practical implications for effective memory
rehabilitation in persons with amnesia. Because of impaired ability to
retain and/or recall relevant information, persons with amnesia are
known to have severe difficulty in adapting to environmental requests. The
domain-specific learning approach to memory rehabilitation, proposed by
Glisky et al. (1994), aims at teaching persons
with amnesia knowledge or skills useful in everyday life by exploiting
their residual memory abilities. Relative to explicit memory, implicit
memory is well preserved in amnesia. Interventions, believed to capitalize
on implicit rather than explicit memory, such as the method of vanishing
cues (Glisky et al., 1986) and errorless
learning (Baddeley & Wilson, 1994), can lead
to memory improvement, thereby alleviating the difficulties that patients
face within real-life situations. Enactment during encoding is likely to
be another intervention that relies on preserved memory in persons with
amnesia. Nilsson et al. (2000) demonstrated,
using positron emission tomography (PET), that the advantageous effect of
SPTs over VTs was attributable to activity in motor areas of the brain.
For this reason, persons with brain damage of differing etiologies but
with an intact motor system are expected to benefit from encoding support
in the form of enactment, as confirmed by the present study.
Recently, Kessels and de Haan (2003) have
conducted a meta-analysis on the treatment effects of two kinds of memory
rehabilitation techniques, errorless learning and the method of vanishing
cues, both of which are based on intact learning capacity in persons with
amnesia. As a result of the meta-analysis, which included eight studies
for errorless learning and three for the method of vanishing cues, a large
and statistically significant effect size (d = 0.87) was found
for errorless learning treatment, but no significant effect size
(d = 0.27) was demonstrated for the vanishing cues method. In our
Experiment 2, the effect sizes of Verbal generation-SPT recall advantage
over VT were 0.84, 0.62, and 0.76 for the young control, nonamnesic
alcoholic, and Korsakoff groups, respectively. These values can be
considered to be medium to large and to be close to the effect size of
errorless learning treatment.
Finally, four limitations of the present study are important to note.
First, our small sample size led to poor statistical power, which
precluded a firmer conclusion that no interactions took place between
participant group and study condition in either Experiment 1 or 2. Second,
no independent variables employed in the present study permitted persons
with Korsakoff's syndrome to fully compensate for their memory
deficit, but SPT encoding or its combination with verbal generation
enhanced memory performance in all groups to a similar degree. Third,
stimuli were presented only once during study, and only immediate recall
was examined. Future studies should investigate the durability of gains
from SPT encoding and verbal generation by using multiple learning trials
and longer delays between study and test. Fourth, from a rehabilitation
perspective, the memory advantage of SPTs found in the present study
should extend to real-world tasks, such as learning the layout of the
facility.
