Copyright 1986 by the American PsyehologicaJ Association, Inc.
0278-7393/86/$00.75
Journal of Experimental Psychology:
Learning, Memory, and Cognition
1986, Vol. 12, No. 1,66-71
Optimizing Cue Effectiveness: Recall of 500 and 600
Incidentally Learned Words
Timo Mantyla
University of Umea, Umea, Sweden
In three cued recall experiments, prerequisites to optimal memory performance of large amounts of
verbal materials were examined. Practically perfect recall of 500 and 600 words was obtained when
effective retrieval cues were provided at test. The method used to demonstrate this was to instruct the
subjects to define their own retrieval cues by generating properties or features to each word presented.
At an unexpected recall test, these self-generated properties were presented as cues, and the subjects
were instructed to recall the previously presented items. Cue effectiveness was manipulated by varying
amount of retrieval information, type of cues, and retention interval. Distinctiveness and compatibility
of retrieval cues are proposed as two necessary prerequisites to perfect recall performance.
for pictorial materials in the study of Shepard (1967) are the
most commonly cited ones, it should be observed that relatively
high levels of recognition were obtained for large amounts of
verbal materials as well.
Corresponding studies reporting high recall, in contrast to
recognition, of a large number of words are, however, practically
nonexistent. To the best of my knowledge, the only exception in
this respect is the study presented by Wallace, Turner, and Perkins
(1957). In this study, subjects learned lists of words by forming
a mental image connecting the members of each pair. Subjects
began with lists of 25 pairs and worked up to through lists of 50,
75, 100, 300, 500 and finally to lists of 700 pairs of words. Following a single presentation, recall of one of the members of
each pair was cued with the other member of that pair, resulting
in nearly perfect performance for all lists. However, in a recent
study, Senter, Richter, Wilson, and Clements (1982) tried to replicate the results of Wallace et al., but failed in the sense that the
two subjects participating in the experiment did not even manage
to complete the whole experimental task involving 700 pairs.
The subjects obtained recall performance of about 25% and 75%,
respectively, for the 350 pairs they managed to go through during
study. Thus, the conclusion to be drawn on the basis of the Senter
et al. study is that there are no reliable data reporting near perfect
recall of large sets of words.
The well-known studies of Shepard (1967), Standing (1973),
and Standing, Conezio, and Haber (1970) have shown extremely
high recognition-memory performances of large amounts of
pictorial information. Shepard (1967) presented subjects with
600 pictures for a few seconds each and then asked them which
member of the various test pairs had been among the 600 items.
More than 95% of the pictures presented previously were correctly recognized. At the time of publication, this result was found
surprising by memory researchers, who were attuned to the rather
meager ability of humans to remember nonmeaningful material
such as nonsense syllables and isolated digits (Loftus, 1982).
Although these studies are spectacular, it should, however, be
pointed out that good memory performance per se is not an
exceptional phenomenon in the memory literature. Mnemonic
devices have been known to western culture for thousands of
years (Yates, 1966), and several experiments have demonstrated
that the capacity of human memory can be increased by using
specific mnemonic systems (see, e.g., Bellezza, 1981) or some
other forms of recoding procedures (Ericson & Chase, 1981;
Miller, 1956). Furthermore, a number of studies have shown that
persons with different skills are able to attain good memory performances in domains where they are experts (e.g., Charness,
1979; Chase & Simon, 1973; Egan & Schwartz, 1979), and there
is also a variety of demonstrations concerning people with exceptional memories (e.g., Hunt & Love, 1972; Luria, 1968).
The purpose of the present study was to investigate whether
good recall performance of a large number of words can be obtained under conditions where subjects are not instructed to use
mnemonic devices. Thus, the subjects of the present study had
no previous experience of memory experiments and, in addition,
the three experiments to be reported here were carried out under
incidental learning conditions. The aim was to have subjects process words in such a way as to produce good retrieval cues.
In order to demonstrate high levels of recall, two characteristics
of cue effectiveness were conceived as prerequisites to good
memory performance. First, in order to be effective, a retrieval
cue has to be compatible with the memory trace in the sense of
describing some of the features or attributes (Underwood, 1969)
that constitute the functional information encoded (Tulving &
Thompson, 1973). An underlying assumption of this notion is
The study of Shepard (1967) is, however, exceptional in the
sense that good recognition memory performance was obtained
in spite of the fact that (a) a large amount of information was
presented on a single trial, (b) the subjects were not experts, and
(c) no specific mnemonic systems were used. Although the data
This study was supported by a grant from the Swedish Council for
Research in the Humanities and Social Sciences. The author is grateful
to L.-G. Nilsson for comments and criticism throughout the course of
this work and to H. Roediger, R. Crowder, and G. Loftus for their reviews
of an earlier version of this article.
Correspondence concerning this article should be addressed to T.
Mantyla, Department of Psychology, University of Umea, Radhusesplanaden 2, S-90247 Umea, Sweden.
66
67
OPTIMIZING CUE EFFECTIVENESS
that the meaning of a word is not a fixed entity but a function
of the context of the task and the rememberer's previous knowledge. If a word is encoded according to some specific meaning,
then, those retrieval cues that represent this meaning of the word
are the most effective ones, that is, they represent a compatible
description of the functional information.
Second, distinctiveness of retrieval cues was conceived as another prerequisite to good recall performance. That is, in order
to optimize cue effectiveness, the number of target items subsumed by a retrieval cue should be minimized. In line with the
cue-overload theory (e.g., Watkins & Watkins, 1975), the probability of obtaining effective retrieval cues was expected to increase as a function of the number of retrieval cues provided,
that is, additional retrieval cues were assumed to aid recall over
a single cue. Distinctiveness, however, was not denned in an absolute sense, but rather relative to some context or particular
background (e.g., Jacoby & Craik, 1979) and to the rememberer's
specific knowledge base. If a retrieval cue is distinctive in the
sense that it represents features of the information encoded that
differentiate this specific encoding from preexperimental encodings (Eysenck, 1979), then good recall would be expected.
item, and at test, these self-generated properties were given as
cues. In addition, the subjects were instructed to generate either
one or three properties for each presented word. Best recall performance of a large number of words was expected to be obtained
when sets of three self-generated properties were provided at test.
Finally, the retention interval was manipulated within subjects:
Tests were given immediately, after 1 day, after 2 days, and after
7 days. The manipulation of retention interval was considered
in part a control for the possibility that the recall performance
could be explained in terms of the general cuing power of the
properties generated by the subjects. Namely, if the properties
generated at study were such good descriptions that they would
enable the construction of the correct word at test without any
preceding study, then the recall level should not decrease as a
function of increasing retention interval.
The two characteristics of cue effectiveness, originating from
the encoding specificity principle (Tulving & Thompson, 1973)
and from the levels of processing framework (e.g., Lockhart,
Craik, & Jacoby, 1976), were proposed as prerequisites to good
recall performance. If both compatible and distinctive retrieval
cues are constructed and later presented at test, good recall of a
large number of words would be expected. According to this
perspective, the main practical problem is construction of retrieval cues that fulfill these prerequisites. However, the effectiveness of retrieval cues depends on a combination of several
factors, and an experimenter cannot determine a priori the most
compatible and distinctive retrieval cues. Only the learner knows
to what aspects of the information presented he or she attends
and what are the most distinctive properties of that information.
If a subject is instructed to describe a word to be remembered
according to his or her idiosyncratic associations, then this selfgenerated description should serve as a very effective retrieval
cue. Hence, an excellent experimental situation for obtaining
high levels of recall should be one where the subjects are allowed
to produce their own retrieval cues.
Subjects. Four undergraduate students at the University of Umea
served on a voluntary basis. None of the subjects had any previous experience in memory experiments. They were paid the equivalent of $4
per hour for their participation.
Materials. The information to be remembered consisted of 504 randomly selected Swedish nouns (Allen, 1970) with the only restriction
being that no synonyms were included. The words were also randomly
divided into six sublists of 84 items each.
Procedure. The subjects participated in the experiment for three consecutive days. Two sublists of 84 words were presented on each of the 3
days. The subjects were tested in groups of two persons, and the order of
sublists was randomized across each pair of subjects. On the first day, the
subjects were instructed to generate one property to each of the 84 items
of the first sublist and three properties to each of the 84 items of the
second sublist. That is, the subjects were instructed to generate one or
three words that according to subjects' own experience constituted an
appropriate description of the target item. Three words (banana, freedom,
and tree) were given as practice to determine if each subject understood
instructions. The subjects were not informed of the subsequent recall
test.
The items were displayed on a television monitor, driven by a Luxor
ABC/80 microcomputer, at a rate of 20 s per item. The subjects were
asked to write down each property word in a booklet with one and three
words, respectively, on each page. The subjects were also instructed to
avoid such properties for which the item to be remembered was one
component (e.g., primadonna as a property of madonna) or for which
the property was a component of the item (e.g., Donna). On the third
day, immediately after the presentation of the final sublist, each subject
was presented with 252 randomly selected pages of the booklet, that is,
42 pages with one and three self-generated properties, respectively, from
each of the three days. The subjects were not informed as to whether a
certain property or a set of properties was generated during the first,
second, or third day. The subjects were instructed to recall the words to
which they had generated the properties by writing down each word on
the appropriate page of the booklet. The order of presentation of test
cues was random and different for each subject. Recall rate was subject
paced, and the test phase took about 90 min. After the recall test, the
subjects were asked if they were interested in participating in another
experiment one week later. Thus, the subjects were not informed of the
final recall test where the remaining 252 cues were presented. At this test
the subjects were again instructed to recall the appropriate words by
using their own properties as retrieval cues.
Mantyla and Nilsson (1983) instructed a group of subjects to
write down three properties which according to the subject's
own experience described the meaning of each word presented.
During an unexpected recall test given later, subjects were presented with their own sets of descriptions in a random order and
were instructed to recall the words presented previously by using
these self-generated descriptions as retrieval cues. The performance level was remarkably high; on the average, more than
96% of a 30-word list was correctly recalled. The Mantyla and
Nilsson (1983) study also showed that neither the mere generation
of properties by itself nor the general cuing power of the selfgenerated properties given to other subjects accounted for the
high recall level. Instead, a distinctive and compatible match
between encoded traces and cues seemed to determine the effectiveness of the descriptions as retrieval cues.
The essence of the experimental procedure used here was
identical to that in the Mantyla & Nilsson (1983) study. That is,
subjects were instructed to generate properties to each presented
Experiment 1
Method
68
TIMO
Results and Discussion
The recall scores, expressed as mean proportions of words
correctly recalled, are based on a strict scoring. Namely, with
the exception of the plural and definite form, only those responses
that were identical to the corresponding target word were accepted
as correct. Furthermore, only those properties that did not, in
some form, include the original item were accepted as proper
retrieval cues. For example, if the word Donna was given as a
property to the presented item madonna, the response of this
cue was omitted. The mean percentage omitted responses was,
however, only about 5%. Furthermore, although the second recall
test included all the remaining 252 items, only the recall performance of the words presented during the third day was calculated, that is, the interval between study and test was exactly
1 week. (The differences in performance between these 3 days
were, however, not significant).
The individual recall scores were first transformed by using
an arc sin transformation (Kirk, 1968) and then subjected to a
2 X 4 analysis of variance (ANOVA) with number of properties
(one vs. three) and retention interval (immediate, 1 day, 2 days,
and 7 days) as within-subjects factors. The ANOVA yielded significant main effects of number of properties, F\l, 3) = 221.53,
MSe = 1.778, p < .01, and retention interval, F\\ 9) = 4.58,
MSe = .244, p < .05. The interaction between these two variables
did not reach statistical significance. The mean proportions of
words correctly recalled are presented in Table 1.
As can be seen in Table 1, high levels of recall were obtained
when three self-generated properties were presented as retrieval
cues immediately after the generation phase. The data indicate
also that the level of recall is directly related to the amount of
retrieval information provided. Three properties produced nearly
40% higher recall level than one property as a cue.
The manipulation of retention interval was conceived as a
control for the possibility that the recall performance could be
explained in terms of the self-generated properties per se. The
results of the present study indicate that the high recall level
cannot be explained as totally due to confabulation, because the
recall performance decreased approximately 30% when the retention interval increased from immediate to 7 days. However,
about 60% of the items were correctly recalled when three properties were presented 1 week later. Thus, there is a possibility
that the properties corresponding to these items were such powerful descriptions that correct recall was obtained due to guessing.
If so, the actual cuing power of the self-generated properties is
only about 30% in the immediate-recall condition. Thus, it could
be argued that the high recall performance demonstrated could
be obtained by merely presenting the properties without any
preceding study phase.
Experiment 2
The purpose of the second experiment was, first, to rule out
this alternative explanation by providing a baseline performance
against which the obtained recall levels could be evaluated, and
second, to replicate the findings of Experiment 1.
In one condition of the experiment, subjects were presented
with 504 words with the instructions to generate their own properties, and then at an unexpected test, subjects were presented
MANTYLA
Table 1
Mean Proportion of Words Correctly Recalled as a Function
of Number of Properties and Retention Interval
Retention interval
Condition
Immediate
1 day
2 days
7 days
3 properties
1 property
.915
.534
.784
,465
.710
.312
.602
.262
with their own or another person's properties as retrieval cues.
Because self-generated properties were assumed to be the best
descriptions of the information encoded, it was expected that
self-generated properties as retrieval cues would produce significantly higher levels of recall than properties generated by someone else. Furthermore, in another condition, subjects were presented with properties generated by other subjects and were instructed to construct the target items without any preceding study.
Thus, the purpose of this condition was to examine if the properties generated were such powerful descriptions of the target
items that good performance could be obtained by merely presenting the properties.
Method
Subjects and design. Sixteen undergraduate students at the University
of Umea participated in the experiment. They were paid the equivalent
of %4 per hour for their participation. The subjects, who had no experience
in memory experiments, were randomly assigned to four groups with
2 females and 2 males in each. The design of the experiment consisted
of two conditions, study versus no study, with two groups in each. Subjects
in the study condition generated, on three consecutive days, either one
or three properties to each presented word. During an unexpected recall
test, these two groups of subjects were given both their own properties
and those generated by someone else as retrieval cues. Thus, number of
properties (one vs. three) was manipulated between subjects, whereas
type of properties (own vs. someone else's) and retention interval (immediate, I day, 2 days, and 7 days) were manipulated within subjects.
Subjects in the no-study condition were not presented with the target
items at all: Subjects were either presented with 504 single properties or
with sets of 3 properties that the subjects in the study condition had
previously generated.
Materials. The information to be remembered consisted of those 504
words used in Experiment 1. The words were randomly divided into
three sublists of 168 items.
Procedure. The two groups of subjects in the study condition were
presented with 168 items on each of the three consecutive days. They
were tested individually, and the subjects were not informed of the forthcoming recall test. The words were displayed on a monitor of a computer
terminal (Digital/VT220 driven by PDP 11/44), and the subjects were
instructed to type their properties by means of the keyboard of the terminal. All the subjects had some previous experience in typing. Three
words were given as practice to determine whether the subjects understood
instructions and how to use the keyboard. The order of sublists was random
for each subject. The presentation rate was subject-paced, and each of
the three study sessions took about 40 or 90 min when one or three
properties, respectively, were generated to each of the 504 words. On the
third day, immediately after presentation of the final sublist, the subjects
were presented with 84 single properties or 84 sets of 3 properties from
each of the 3 days. An equal number of these 3 X 84 cues consisted of
subjects' own properties and of those generated by other subjects in the
OPTIMIZING CUE
study condition. The properties, or the sets of properties, were displayed
on the monitor of the terminal, and the order of presentation was random.
The subjects were instructed to recall the corresponding target words by
using the keyboard of the terminal. After the recall test, the two groups
of subjects in the study condition were asked if they were interested in
participating in another experiment 1 week later. Thus, the subjects were
not informed of the final recall test where the remaining 252 properties
or sets of three properties were given, and the subjects were instructed to
recall the appropriate words by using their own and someone else's properties as retrieval cues.
The two groups of subjects in the no-study condition were given either
the 504 single properties or the 504 sets of 3 properties that the subjects
in the study condition had previously generated. The subjects were informed that the properties to be presented constituted descriptions of
words generated by some other subject. They were also told that each
property or a set of 3 properties described a single noun. As in the study
condition, the properties were presented by means of a computer terminal,
and the subjects used the keyboard of the terminal for their responses.
Presentation rate was subject-paced, and the test took about 90 min.
Results and Discussion
The recall scores are based on the same scoring criterion as
in Experiment 1. The data of the two conditions, that is, study
vs. no study, were analyzed separately. When one or three properties were presented as cues without any preceding study phase,
the mean proportions of correctly generated target words were
.047 and . 174, respectively. Thus, these data indicate that the
high degrees of recall obtained in Experiment 1 clearly result
from retention and cannot be explained in terms of the general
cuing power of the properties per se in generating correct guesses.
The recall performance of the subjects in the study condition
is presented in Table 2. These data are similar to those of Experiment 1 in the sense that very good recall performance was
obtained when three self-generated properties were presented as
cues immediately after study. However, the present study also
demonstrated that self-generated retrieval cues were much more
effective than those generated by someone else. As can be seen
in Table 2, the difference in immediate recall between self-generated and others1 is more than 35% when three properties were
presented, and nearly 50% when only one property was presented
as a cue. Thus, even when the subjects had been presented with
the target items during study, the effectiveness of the properties
generated by another person as retrieval cues was clearly lower
than that of the self-generated properties.
Furthermore, the data indicated that forgetting to self-generated retrieval cues was greater than to those generated by someone
else. The mean recall performance decreased as a function of
retention interval approximately 30% when the subjects were
presented with their own properties. However, when someone
else's properties were presented as cues, a smaller decrease in
performance was observed (12% for 3 cues, 5% for I cue). Thus,
the data indicate that self-generated properties are powerful cues,
but their effectiveness seems to depend on contextual factors.
When a cue is presented immediately after study, the meaning
of that cue is represented in the same sense as when the cue was
generated, namely, the functional cue is a compatible description
of the information encoded. However, at a delayed test, the
meaning of the same nominal cue can be represented differently
due to the fact that the internal and/or external context is different.
69
EFFECTIVENESS
Table 2
Mean Proportion of Words Correctly Recalled as a Function
of Type of Properties. Number of Properties, and
Retention Interval
Retention interval
Condition
3 properties
Self-generated
Someone else's
1 property
Self-generated
Someone else's
Immediate
1 day
2 days
7 days
.911
.548
.726
.423
.654
.435
,649
,428
.607
.113
.421
.112
.305
.089
,207
,062
No
study
.174
.047
The recall scores of the subjects in the study conditions were
arc sin transformed (Kirk, 1968) and subjected to a 2 X 2 X 4
ANOVA with number of properties as a between-subjects factor
and type of properties and retention interval as within-subjects
factors. The ANOVA yielded significant main effects of number
of cues, F{\, 6) = 24.30, MSe ^ 2.844, p < .01, type of cues,
FX\, 6) = 58.85, MS< = 1.921, p < .01, and retention interval,
F{3, 18) = 26.15, MSt = .235, p < .01. A significant interaction
effect between retention interval and type of cues was also obt a i n e d , ^ , l8) = 21.51,MS e = .102, p<. 01. No other effects
reached statistical significance.
Taken together, the data of the present experiment indicate
that the high degrees of recall obtained here and in the previous
experiment cannot be explained in terms of the general cuing
power of the properties generated. It was also shown that the
self-generated retrieval cues, in contrast to those generated by
someone else, are the best cues.
Experiment 3
The purpose of the third experiment was to replicate Experiments 1 and 2 by presenting a still larger number of words. In
contrast to the previous experiments, 600 words were presented
in a single study session, and at the unexpected recall test immediately following the generation phase, the subjects were instructed to recall all the items presented at study. Thus, the proportions of recall were based on the total number of words to
be remembered, and not on small subsets of items as in the
previous experiments. Furthermore, distinctiveness of retrieval
cues was manipulated between subjects by instructing the subjects
to generate either one or three properties to each presented word.
Method
Subjects. Eight undergraduate students at the University of Umea
participated in the experiment on a voluntary basis. None of the subjects
had any prior experience in memory experiments. They were paid the
equivalent of $4 per hour for their participation.
Materials. The information to be remembered consisted of 600 randomly selected nouns including those 504 items used in Experiments 1
and 2. The 600 words were randomly assigned to four sublists of 150
words each.
Procedure. The subjects were informed of the nature of the generation
phase including the fact that the experiment would take about 6 or 7 hr
70
TIMO MANTYLA
and that the task was relatively demanding. However, the subjects were
not informed of the forthcoming recall test. The subjects were randomly
divided into two groups with 2 females and 2 males in each. The items
were displayed on a television monitor, by means of a Luxor ABC/80
microcomputer, at a rate of 20 s per word. The subjects were instructed
to generate either one or three properties to each word presented, and to
write down their properties in a booklet with a single property or a set
of three properties on each page. The total of 600 items was presented
in sublists of 150 words with a rest period of 5-10 min following each
sublist. Thus, the generation phase took about 4.5 hr including the rest
periods. At the unexpected recall test, the subjects were presented with
their self-generated properties and were instructed to recall the 600 words
to which they had generated the properties now being presented. The
order of the presentation was random and unique for each subject. The
recall rate was subject-paced and the test phase took approximately 2 hr.
Results and Discussion
The recall scores are based on the same scoring criterion as
in Experiments 1 and 2.
As can be seen in Table 3, the recall performance was found
to be extremely high when three self-generated properties were
presented as retrieval cues. The mean percentage of correctly
recalled words was more than 90%. The corresponding recall
performance was about 62% when one self-generated property
was presented as a cue. As demonstrated by a / test, this difference
in performance was statistically significant, f(6) = 4.95, SE =
A 19, p < .01. Thus, the data presented here have again demonstrated good recall performance of a large number of words.
Even though the experiment was very demanding, that is, it took
almost 7 hr and the subjects were required to recall all of the
600 words, more than 90% of these items were correctly recalled
in the three cues condition, and surprisingly high degrees of recall
were obtained even in the single cue condition where such good
performance was not even expected.
General Discussion
The main purpose of the present study was to demonstrate
that good recall performance of a large number of words, encountered on a single trial, can be obtained even when no specific
mnemonic devices are used. Cue effectiveness was considered as
a prerequisite to good recall performance: In order to be effective,
a retrieval cue has to represent a compatible and distinctive description of the information encoded. These two characteristics
of cue effectiveness, that is, compatibility and distinctiveness,
were defined relative to the rememberer's previous knowledge
and the specific context in which the information to be remembered was encountered.
The present study demonstrated that the presentation of sets
of three self-generated properties as retrieval cues immediately
after the study phase resulted in correct recall for more than
90% of the target items. The effect of the self-generated cues
declined, however, as a function of increasing retention interval.
This decrease in performance reflects a loss in compatibility between the functional cue and the information encoded due to
the different contextual factors of the delayed test. This notion
implies that good delayed recall could be obtained if the original
encoding context somehow could be reinstated during testing.
Reddy and Bellezza (1983) have demonstrated that the level of
recall is directly related to the degree to which the contextual
Table 3
Recall Performance as a Function of Number of Properties
Subject
Condition
1
2
3
4
M
3 properties
1 property
.932
.635
.872
.692
.900
.623
.915
.515
.905
.616
information produced during encoding is reinstated. They used
a method where the overt verbalization produced by the subjects
themselves during the study of a list of words was recorded. At
recall, one group of subjects was given transcripts of their study
vocalizations, and a second group of subjects was given transcripts
of some other subject's vocalizations. The results showed that
presence of cues vocalized by the subjects during presentation
facilitated recall. More than 90% of a 40-item list was correctly
recalled. However, when a different encoding context (someone
else's utterances) was presented at recall, then only about 36%
of the items were correctly recalled.
The results of the present study also demonstrated that selfgenerated properties were more effective than those generated
by someone else. Thus, the method used here enables subjects
to use their own idiosyncratic and well differentiated representations of the information to be remembered in order to construct
effective retrieval cues. As experts, for example, in chess or bridge
can demonstrate exceptional memory performances in domains
of their expertise, the subjects of the present study, as "experts"
of their own lives, were able to use their rich and idiosyncratic
knowledge base in order to construct effective retrieval cues and,
hence, to demonstrate extremely high degrees of recall.
As a concluding remark, it should be pointed out that although
extremely high degrees of recall were obtained in the present
study, the paradigm used here is not proposed as the optimal
method for obtaining high recall performances. For example,
due to the fact that the subjects were instructed to verbalize their
properties by writing them down, a number of distinctive properties of the information to be remembered were probably excluded. The present study, however, has shown that by fulfilling
at least some of the necessary prerequisites, surprisingly high
degrees of recall can be obtained. The main focus of the present
study was an important but often neglected component of the
"theorist's tetrahedron" (Jenkins, 1979)—the rememberer. In
most traditional memory experiments, the subjects are instructed
and required to process information according to certain rules
and restrictions defined by the experimenter. These experiments
have produced valuable pieces of knowledge, but it can still be
argued that our understanding of memory and remembering can
be extended by instructing and encouraging subjects to process
information according to their own cognitive structures. Especially in situations where the information to be remembered has
an abstract/symbolic character, the rememberer's knowledge and
specific skills are natural and powerful resources.
References
Allen, S. (1970). Frequency dictionary of present-day
Stockholm: Almqvist & Wiksell.
Swedish.
OPTIMIZING CUE EFFECTIVENESS
Bellezza, F. C. (198 i). Mnemonic devices: Classification, characteristics,
and criteria. Review of Educational Research, 51, 247-275.
Chamess, N. (1979). Components of skill in bridge. Canadian Journal
of Psychology, 33. 1-16.
Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive
Psychology, 4, 55-81.
Egan, A. E., & Schwartz, B. J, (1979). Chunking in recall of symbolic
drawings. Memory & Cognition, 7, 149-158.
Ericson, A., & Chase, W. G. (1981). Skilled memory. In J. R. Anderson
(Ed.), Cognitive skills and their acquisition (pp. 141-189). Hillsdale,
NJ: Erlbaum.
Eysenck, M. W. (1979). Depth, elaboration, and distinctiveness. In L. S.
Cermak & F. I. M. Craik (Eds.), Levels of processing in human memory
(pp. 89-118). Hillsdale, NJ: Erlbaum.
Hunt, E., & Love, T. (1972). How good can memory be? In A. W. Melton
& E. Martin (Eds.), Coding processes in human memory (pp. 237260). Washington, DC: Winston.
Jacoby, L. L.,& Craik, F. I. M. (1979). Effects of elaboration of processing
at encoding and retrieval: Trace distinctiveness and recovery of initial
context. In L. S. Cermak & F. I. M. Craik (Eds.), Levels of processing
in human memory (pp. 1-21). Hillsdale, NJ: Erlbaum.
Jenkins, J. J. (1979). Four points to remember: A tetrahedral model of
memory experiments. In L. S. Cermak & F. I. M. Craik (Eds.), Levels
of processing in human memory (pp. 429-446). Hillsdale, NJ: Erlbaum.
Kirk, R. E. (1968). Experimental design procedures for the behavioral
sciences. Belmont, CA: Wadsworth.
Lockhart, R. S., Craik, F. 1. M., & Jacoby, L. (1976). Depth of processing,
recognition, and recall. In J. Brown (Ed.), Recall and recognition (pp.
75-102). New York: John Wiley & Sons.
Loftus, G. R. (1982). Picture methodology. In C. R. PufT(Ed.), Handbook
of research methods in human memory and cognition (pp. 257-285).
New \brk: Academic Press.
Luria, A. R. (1968). The mind of a mnemonist. New York: Avon.
71
MantylS, T., & Nilsson, L.-G. (1983). Are my cues better than your cues?
Uniqueness and reconstruction as prerequisites for optimal recall of
verbal materials. Scandinavian Journal of Psychology, 24, 303-313.
Miller, G. A. (1956). The magical number seven plus or minus two. Psychological Review, 63, 81-97.
Reddy, B. G., & Bellezza, F. S. (1983). Encoding specificity in free recall.
Journal of Experimental Psychology. Learning, Memory, and Cognition,
9, 167-174.
Senter, R. J., Richter, D. Q , Wilson, Q, & Clements, D. (1982). Wallace,
Turner and Perkins revisited. Bulletin of the Psychonomic Society, 20,
187-179.
Shepard, R. N. (1967). Recognition memory of words, sentences and
pictures. Journal of Verbal Learning and Verbal Behavior, 6, 156-163.
Standing, L. (1973). Learning 10,000 pictures. Quarterly Journal of Experimental Psychology, 25, 207-222.
Standing, L., Conezio, J., & Haber, R. N. (1970). Perception and memory
for pictures: Single trial learning of 2560 visual stimuli. Psychonomic
Science, 19, 73-74.
Tulving, E., & Thompson, D. M. (1973). Encoding specificity and retrieval
processes in episodic memory. Psychological Review, SO, 352-373.
Underwood, B. J. (1969). Attributes of memory. Psychological Review,
76, 559-573.
Wallace, W. H., Turner, S. H., & Perkins, C. C. (1957). Preliminary studies
of human information processing. Unpublished manuscript (Signal
Corps Project No. 1320c), University of Pennsylvania, Institute for
Cooperative Research, Philadelphia.
Watkins, O. C , & Watkins, M. J. (1975). Buildup of proactive inhibition
as a cue-overload. Journal of Experimental Psychology: Human Learning and Memory, 104, 442-452.
Yates, F. (1966). The art of memory. London: Routledge and Kegan Paul.
Received September 10, 1984
Revision received April 12, 1985 •