A Brief History of Memory Research: Background: Associationism
A Brief History of Memory Research: Background: Associationism
A Brief History of Memory Research: Background: Associationism
ww....
.
1
1
:
:
::
..
:. .
:. .. .. ..
Background: Associationism
connecting together in memory simple ideas
based on sensations that are experienced
conPsychology as a discipline developed out of tiguously in time and/or space. The
memory philosophical discussions regarding the na- that sensory quality or event A was
experiture of the mind and mental life. The study of enced together with, or immediately
precedmemory and learning arose from philosophi- ing, sensory quality or event B is
recorded in cal questions regarding how people come to the memory bank as an
association from idea know things about their world. Learning is as- a to idea b.
Reviving these associative sesuredly the primary way we acquire knowl quences from
memory (when recurrence of edge, and remembering is a primary means by event A
makes us think of event B) is the prewhich people support knowledge claims, as sumed
method by which people's past experiwhen a witness in court asserts "I remember
ences cause their later thoughts to progress seeing Jones with a revolver in his hand."
from one idea to the next. This basic notion
Philosophical speculations about learning can be elaborated to account for the way
huwere prominent among advocates of empiri- mans develop coordinated expectations
about cism, which is the view that sensory experi- properties of objects, expectations
about ences are the only ultimate source of knowl- causal sequences of events,
predictions about edge and truths about the world (contra innate future events,
explanations of how or why ideas or religious authorities). People's ideas something
came about, and plans of action deabout the world are alleged to derive from signed to
bring about particular outcomes, sense impressions either as simple copies or These
are basic abilities of the mind. as combinations of simple ideas. Objects such
Throughout the seventeenth to nineteenth as oranges, dogs, and houses are allegedly
centuries, empiricist philosophers such as constellations of many sensory qualities (e.g.,
John Locke, John Stuart Mill, and Thomas the color, shape, taste, and texture of an or-
Brown speculated about various factors that ange).
might affect the degree or strength of particuThe empiricist program required some lar
associations (Warren, 1921). They recogmeans for learning these constellations. Thus
nized that associations would vary in their was introduced the fundamental theory of as-
strength according to the vividness or distincsociation by contiguity (Warren, 1921).
Com- tiveness of the original experience, its duraplex ideas are allegedly formed in the
mind by tion (study time), its frequency (repetitions),
.
..
...
....
:
4
STUDY OF MEMORY
and its interest for the observer. Revival of as- (Darwin, 1859). Human learning was
seen as sociations from memory was hypothesized to an adaptive mechanism that
evolved over milvary with the resemblance of the stimulating lions of years throughout
the animal kingdom cue to the memory, the recency of the experi- by small variations
and minor accretions in ence, the coexistence of fewer alternative asso- the neural
hardware that carries out the variciates to the cue (called "interference"), and ous
learning tasks with which organisms are "temporary diversities of state" (intoxication,
confronted. This "biological continuity” view delirium, depression). Such conjectures
have justifies the many comparative studies by psygenerated much experimental
research on chologists of behavioral adaptation and learnlearning and memory, and
every learning the ing in lower animals. Since animals do not ory deals with these
factors in some way talk, those studies led in turn to a strong be(Bower & Hilgard,
1981).
haviorist orientation toward learning. Thus,
The scientific investigation of association learning came to be viewed as a change in an
formation began with the work of a German sci- organism's behavioral dispositions in
particuentist, Hermann Ebbinghaus, whose pioneering lar situations (S-R habits) as a
result of its exresearch (with himself as sole subject) was pub- periences. It was
recognized, of course, that lished in his treatise On Memory in 1885. Dis the responses
may be complex skills and the cussion of his work will be postponed in order stimuli
may be those stemming from a comto examine briefly another major influence on plex
environment, including intricate and studies of learning-namely, the doctrine of be-
subtle social situations. haviorism, which became wedded for many Behaviorist
approaches to learning were years to the doctrine of associationism.
greatly encouraged around the turn of the twentieth century by the pioneering
studies of
conditioned reflexes by the Russian
physioloBehaviorism and
gist Ivan Pavlov (1927) and by early studies
of "trial-and-error" (instrumental) learning
by S-R Psychology
Edward Thorndike (1898, 1903), an influential
educational psychologist in America. This
beThe Behaviorist Philosophy
haviorist orientation was promulgated by
(like DAX, QEH) to provide himself with test, and relationships among several
sets of learning materials of homogeneous difficulty, materials being learned. As
variables have thus avoiding the variability of familiar words been isolated and
studied, a huge backlog of or prose. He taught himself by studying serial
empirical information has accumulated about lists of 6 to 20 syllables, reading
them aloud in how humans learn in these situations. And sequence in pace with
a metronome and then many theoretical hypotheses have been protrying to
recite the series from memory. The posed and tested to integrate and account for
serial list was his analog of the associative the evidence surrounding specific
topics. chain of ideas about which philosophers had speculated.
Analysis of Laboratory Rote
Ebbinghaus introduced many important
Learning Tasks ideas and
methods (see the Ebbinghaus symposium published in the July 1985 issue
(vol. The rote learning tradition was established ume 11) of the Journal of
Experimental Psy- around the intensive study of three different chology:
Learning, Memory, and Cognition). kinds of learning paradigms-serial
learning, He measured the difficulty of learning a list by paired-associate
learning, and perceptual-mothe number of study trials required for him to
tor skill learning. We will briefly characterize attain one errorless recitation
of it. He noted each of these learning tasks and a few of their how difficulty
increased disproportionately findings. with the length of the list being
learned. He introduced the idea of measurable "degrees of
Serial Learning learning" (or
forgetting) by noting the savings in relearning a list he had learned earlier. The
The task Ebbinghaus used is called serial percent savings was the difference in
trials for learning, an analog of learning the alphabet or original learning (say, 9
trials) minus those learning to put letters in sequence to spell a needed for later
relearning (say, 3) divided by word: the subject learns to output in a specithe
original learning trials (so, (9-3)/9 = fied order a small set of temporally ordered,
67%). Using this measure, he was able to plot discrete items (letters, nonsense
syllables, his famous forgetting curve relating percent written or spoken words,
pictured objects, savings to retention interval. This curve (fig- sentences).
Subjects are asked to remember ure 1.1) showed very rapid losses over the first
both the items and their serial order. Retrieval few hours or days, with more
gradual but may be tested by asking subjects either to resteady decline over
subsequent days, weeks, produce (recall) all items in the order preand months.
Ebbinghaus also found that for- sented, or to recall what item followed a
spegetting of a list decreased with multiple re- cific cued item, or to reconstruct
the presented learnings of it, that overlearning increased re- order when given
the items (on flashcards) in tention, and that widely distributed study scrambled
order. In some experiments, a numtrials (say, 1 per hour) were more effective ber
of series are presented only once for recall than closely packed trials (say, 1 per
minute) (e.g., for measuring the immediate memory for long-term retention.
span). In other experiments, the same
items Ebbinghaus's new paradigm (adults learn- may be presented many times
in the same oring lists of nonsense materials) defined a task der for repeated
study and test trials to examin which a multitude of variables can be de- ine
accumulative learning. fined and their influences on "remembering" Studies of
serial learning have uncovered behaviors observed. The phenomena that he
many facts, Increasing the study trials and discovered, his ideas, and his
methods cast a time per item increases learning; increasing long shadow
throughout the twentieth century the time subjects are given to anticipate the of
research on human memory. Subsequent re- next successor in the series
improves their search has invented several other paradigms performance. While
making the items very and teased out many variables that determine similar to
one another (e.g., XON, NEH, XEH, memory performance in these settings. The
NOH) improves their recallability, this simimemories established can be tested by
either larity creates many confusion errors about recall, recognition, and
reconstruction, or by a their ordering. A robust finding is that items variety of
indirect measures. The nature of the at the beginning and end of the list are
easier materials can be varied, as can their mode of to learn than items in the
middle (see figure presentation, strategies subjects use in study. 1.2), a fact that
has provoked many explanaing them, expectations regarding the memory tory
attempts (Johnson, 1991).
JWUNALLA. L
I ." TW. ARAW
L
.
T
12
-
..
.-..
.
-
-
-
-
-
IMMEDIATE RECALL
20 MIN.
SAVINGS
I HR.
409HR.
9 MR.
oted
0
12
Figure 1.2 Predicted and observed relative serial position error curves for 8-, 11-, and 14-item
lists. Data from Hovland (1940). The fit to the data is provided by Johnson's theory of relative
distinctiveness of different serial positions. (From figure 1 of "A distinctiveness model of serial
learning" by G. J. Johnson (1991). Psychological Review, 98, pp. 204-217, Reprinted by
permission.)
1
2
3
4
5
6
7
9
10
11
12
13
14
SERIAL POSITION
-
-
Conjectures regarding the effective stimu- ones for which subjects can easily
discover lus for the next response in the series have meaningful relationships.
This issue will be also provoked much research. The natural hy- revisited later.
pothesis is that the series is learned by chaining together a set of pairwise
associations, so
Skills that the series A-B-C-D is
stored as the unordered pairs C-D, A-B, B-C. Thus, item C is Many categories of
human skills have been both a response to its preceding cue, B, and in studied
systematically since the turn of the turn a cue for its successor, D. However, pre-
century (for reviews, see Rosenbaum, 1991; dictions from this pairwise chaining
hypothe- Schmidt, 1988). They vary greatly and include sis frequently fail,
suggesting that some more (1) perceptual motor skills such as typing and
abstract "mental slots" for relative position telegraphy, athletic skills such as
diving and (“first-A, second-B, ... last-D”) may be shooting basketball, musical
skills such as viothe effective cues for the ordered associations. lin and piano
playing, and laboratory tasks
such as pursuit rotor tracking and
reversePaired-Associates Learning
8
STUDY OF MEMORY
PETIT
100 T
recall threshold
recognition threshold
ASSOCIATIVE STRENGTH
o
i
Ź
6
3 4 5 TRAINING TRIALS
7
8
...
.
.
10
STUDY OF MEMORY
that such mediators serve as only a temporary Implicit in this emphasis on the
association crutch during early learning; as practice pro- value of to-be-learned material
was an imporceeds and the direct king-table association is tant underlying idea-namely,
that people repeatedly reinforced, it is retrieved more learn most easily by relating the
new material quickly, thus short-circuiting (and beating out) to things they already know.
They look for the the mediating associative chain that then "mediators" mentioned
previously. They try drops away.
to transform (recode) the to-be-learned mateA
second aspect of this generalization-dis- rial into something familiar or close to
somecrimination analysis is that after learning- thing that is meaningful to them. Thus,
the which, remember, occurred by inhibiting the nonsense syllable MIK is recoded as
"MIKE preexisting and generalized responses-the with the E missing" and JQA as "the
initials passage of a retention interval allows the inhi- of John Quincy Adams" (see
Montague, 1972). bition of errors to decay, thus leading to the Learners then remember
their recoding, and partial recovery of errors over time. Therefore, convert it back into
the to-be-remembered sealthough once inhibited, error responses such ries when they
are asked to recall. as throne-king may recover in strength over a This observation led
different investigators delay interval, creating forgetting via errors of in three different
directions. One direction commission. While this hypothesis (due to was to account for
the kind of mediators subGibson, 1940) explains forgetting in terms of jects came up
with; for example, Montague the familiar conditioning concept of spontane- and Kiess
(1968) showed that learning of a ous recovery, it must be admitted that it fares paired
associate was highly predictable from poorly against the evidence (see Underwood, the
ease with which adults came up with a 1961). For example, contrary to implication,
mediator for it; and Prytulak (1971) predicted once learned, high-similarity
paired-associate the memorability of different nonsense trilists are not forgotten at a
higher rate than low grams by how complicated were the "mental similarity lists.
steps" subjects required to convert them into familiar words. A second and predominant
di
rection for researchers was to abandon use of
Materials Variations
nonsense materials (as too variable) for study
ing elementary learning and move on to
studThe foregoing illustrates that even in Ebbing- ies of learning of already meaningful
materials haus's earliest experiments, it was clear that such as words, sentences, texts,
pictures of learning depended greatly on the nature of the common objects, and video
events; this prematerials—whether numbers, nonsense sylla- dominant trend accounts
for the fact that nonbles, words, meaningful sentences, coherent sense-syllable learning
studies practically disprose, or poetry. These vary greatly in their fa- appeared in the
memory literature after the miliarity and meaningfulness for subjects for late 1960s. The
third direction for researchers example, in how much prior experience sub was to
examine carefully how past learning jects have had with such materials and how was
brought to bear upon a person's current rich the network is of associations subjects
learning. These were called studies of "transhave surrounding the units to be learned.
fer" of past knowledge—a topic to be dis
Researchers in the Ebbinghaus tradition cussed briefly now. soon noticed that even
their favored nonsense syllables differed greatly in how many words and other
associations they evoked among learners, and that syllables evoking more asso
Transfer of Knowledge ciations were
learned more easily. To deal with this variability, researchers tested many By
assumption adult learners always come subjects to tabulate "association norms" or into
a given learning situation with consider"meaningfulness norms" for collections of able
knowledge, learning strategies, and spenonsense syllables (Glaze, 1928), other non-
cific associations that they use as best they can sense materials (consonant trigrams
such as to optimize performance on the given task. XMQ; Witmer, 1935), and words
(Noble, 1952; These transfer effects include both general Noble & Parker, 1969). These
norms were used methods of attack for solving particular learnto compose lists of
materials that were of spec ing problems and more specific associations ified
associative difficulty,
among units employed in the new task.
.
!! .
. " TITA
Nonspecific Transfer
useful techniques for learning educational ma
terials (e.g., history or science
textbooks), inGeneral (nonspecific) transfer includes strate- cluding how to
analyze texts for important gies for studying particular kinds of materials, ideas
and their relationships, memorize, orgaselecting discriminating cues from the
nomi- nize one's study time, and recite and review nal stimuli, composing
mediators for particu- the material (see, e.g., Dansereau, 1978; Mayer, lar
materials, optimizing use of immediate 1987; O'Neil, 1978). memory, and
adjusting to the temporal pacing of the study and test trials in the laboratory
Specific Transfer task. An
example of such general transfer effects is demonstrated in an experiment
by The second class of transfer studies examine Thune (1951). His
college-student subjects the influence of specific prior associations, learned
three new unrelated lists of 10 paired and these are of two types. One type
of study adjectives on each of 5 successive days, each examines subjects'
preexisting associations list practiced for 10 trials. The average correct
learned over a lifetime, and investigators then responses over the first 5
trials on each list (in notice how these operate in transfer to new figure 1.5)
reveals a swift rise in performance learning situations-understanding, of
course, across the three new lists within each day (re- that these past
learnings may vary widely flecting "warm up" adjustments, along with a
across subjects owing to their differing experislower improvement across
days (reflecting ences. An example here is the use of familiar "learning to
learn" lists of this kind). Many words rather than unfamiliar letter strings as
studies of this general kind have examined the learning materials. The
perceptual system dibenefits of explicitly teaching subjects various vides or
segregates stimuli into discrete mnemonic devices (Bower, 1970a), as well
as groups or "chunks," and looks for a match in
i
egen
WSPS.ME
---
--
--
--
اﻟﻠﯿﻞ
-
ﻻﻻﻻﻻﻻﻻﻻﻻﻻﻻﻻااا
1 2 3 Day 1
1 2 3 Day 2
1 2 3 Day 3
1 2 3 Day 4
1 2 3 Day 5
Figure 1.5 Trends in nonspecific transfer with practice. Three paired
associate lists were learned for 10 trials on each of 5 days; average
number of correct responses over first 5 trials on each list is plotted.
Contrast the large gains within sessions with the smaller ones between
sessions. (From L. E. Thune (1951), "Warm-up effect as a function of level
of practice in verbal learning." Journal of Experimental Psychology, 42, pp.
250–256. Reproduced by permission.)
12
STUDY OF MEMORY
memory to these perceptual chunks. A chunk stimulus are closely related in meaning.
Havis defined as a familiar collection of more ele- ing learned prince-throne, for
example, submentary units that have been interassociated jects would be quick to learn
prince-chair, and stored in memory repeatedly and that act presumably because the first
associate is used as a coherent, integrated group when retrieved as a mediator to aid
learning of the second (Miller, 1956; Simon, 1974). The brain's mech- pair. anism for
learning begins its operation of The topic of transfer is far richer and more association
formation) by linking together complex than this brief discussion can indielementary
units into larger chunks. The dif cate. For example, psychologists have examference
can be seen in comparing, for exam- ined the generally positive transfer that occurs ple,
letter strings such as IBF versus FBI ver- when adults learn two computer programming
sus FIB: the first is three chunks, the second a languages or two computer text editors
(see meaningful but unpronounceable chunk, and Singley & Anderson, 1989). In those
situations the third a pronounceable and meaningful positive transfer appears to arise
owing to simword. A simple rule is that novel material will ilar problem-solving steps that
are utilized in be remembered more easily the fewer chunks the two domains, despite
differences in decontained in it. This, of course, requires a suit- tails. For example, the
general plan for how to ably educated person who has the requisite fa- "cut and paste"
text must logically be nearly miliarity with the verbal units.
the same for every text editor (so that that
plan Beyond examining these preexisting lan transfers), whereas the specific key
strokes reguage chunks, however, investigators may aim quired may differ somewhat in
the two edito study the composition of transferable habits tors, causing minor negative
transfer at that and their mode of operation by explicitly level. teaching subjects prior
associations and then noticing how these affect later learning of re
Forgetting lated material. Different
arrangements can produce either positive, negative, or zero transfer The
stimulus-response approach easily acon a second learning task,
commodates three basic reasons for forgetting Classic cases of negative transfer arise
associations after they have been learned. The when subjects must learn a new
arbitrary re first reason is simply autonomous decay in the sponse to an old stimulus.
For example, hav- strength of the S-R association due to physing already learned who
is married to whom iological and metabolic processes that cause (e.g., Bill-Sally and
Dan-Ruth), subjects then progressive erosion of the synaptic changes in learn new
pairings as couples divorce and re- the brain that had encoded the original associmarry,
requiring new associations such as ation. The second reason for forgetting is
perBill-Jane and Dan-Sally. This situation pro formance loss due to a type of stimulus
generduces large negative transfer relative to a con- alization: the training cue (or total
stimulus trol condition, because the prior specific asso- complex) is allegedly
progressively altered beciations compete with and interfere with the tween training and
testing owing to natural subject's quickly learning the new associa- forces, so that the
full stimulus complex is not tions. The old associations keep coming to reinstated at
testing, resulting in progressively mind and blocking retrieval of the new associ- poorer
retrieval of the original association. As ations. The effect here is similar to that in the a
hypothetical example, suppose that the full earlier illustration of king-throne interfering
training complex were composed of the simulwith remembering king-table. The amount
of taneous cues A, B, C, D, all associated with negative transfer is greater the more the
first response R. If only cue A were present later learned associations are aroused by
the second (because B, C, D are altered, missing, or forgotstimuli. Thus, if the
word-stimuli in list 2 ten), then A is unlikely to retrieve the whole were close in meaning
to the stimuli in list 2, associative complex, so response R will not be then the earlier
associations will generalize to recalled. The B, C, D cues in this example the second-list
stimuli, creating more negative might refer to any of a diverse set of what are transfer
the greater their similarity. In con- called "contextual stimuli”-internal postrast, transfer is
rather different when the rela- tural, sensory stimuli plus emotional and astionship of the
nominal response terms in the sociative responses of the subject, as well as successive
pairs is varied. When words are environmental cues besides those experimentused as
paired associates, positive transfer is ers explicitly recognize in their description of
observed when the successive responses to a the stimulus situation.
-
-
A third reason in S-R theory for forgetting by prior learning of a first list
(A-B). The conis that other associations learned before or trol subjects in
this case learn only the A-C after the target association in question may list.
come over time to compete with, displace, or Many facts are known about
forgetting block out and interfere with retrieval of the caused by such
interference (see Postman, target material. This happens, for example,
1971). First, the stronger the training of the when people have trouble
remembering the target response compared to the competing asname of a
woman's second husband because sociates, the better it will withstand
interferher first husband's name is quite familiar and ence from other
associates. Second, the it keeps coming to mind and getting in the stronger
or more numerous are the interfering way. Another example is that people
often associates (e.g., multiple lists such as A-C, Amisremember attributes
of a person they met D, A-E learned before or after the A-B target earlier by
intruding those appropriate to the list), the greater the decrement in recall of
the ethnic stereotype that they had assigned to target associates.
Examples of these relationthat person.
ships are shown in figure 1.6 (from Briggs, The evidence for each of these
three causes 1957): in this study, after 2, 5, 10, or 20 trials of forgetting is
simply overwhelming. Through- of second-list training, subjects were asked
to out the history of research on forgetting, at give the first response that
each "A" stimulus tempts have been made to eliminate theoreti- brought to
mind. The panels of the figure cally one or another factor, or to explain one
show the data for subjects who received 2, 5, factor as really due to
another (e.g., decay 10, or 20 trials of original learning (OL) in figeliminated
in favor of interference—see Mc- ure 1.6 on A-B. The more A-C trials that
accuGeoch & Irion, 1952; or decay eliminated in mulated, the more A-C
came to the fore; but favor of progressive drift over time in the re- the more
original learning (A-B) trials, the trieval context--see Estes, 1955). Because
all longer A-B predominated before losing out. three factors are usually
highly correlated, Another fact about interference is that a these attempts at
colonization are typically in- longer interval or gap between learning of the
sufficient, often requiring assumptions that two lists reduces the extent to
which they inare equivalent to the other factors to account terfere with one
another. Furthermore, as the for the full range of data.
interval from the second-learned
list to testing The interference factor has been most often is lengthened,
proactive interference increases studied in the laboratory. In the paradigm
whereas retroactive interference decreases. most easily studied, subjects
learn two or Such results are explicable by the ideas that more lists of
associations bearing particular re- all associative strengths decay, and that
comlationships to one another. To illustrate with peting associates
suppressed during learning paired associates, let A-B denote a generic list
recover somewhat as time passes and their of many pairs, where A and B
denote the ge- suppression dissipates. neric stimulus and response terms,
respec Interference effects are not solely a function tively. In retroactive
interference studies, re- of overt intrusion errors of the explicit comcall of a
first learned list of homogeneous A peting responses. The subject may
think of the B pairs is substantially impaired after subjects competing
response but recognize it as not the learn a second list with the same
stimuli but requested target-response, so withhold it (a different responses
(denoted generically as A- process called "list discrimination"). On other C).
The amount of retroactive interference is occasions, interference arises as
a total blockassessed by comparison to a control condition ing or inability of
the subject to think of any in which subjects learn the A-B list but then
relevant response (an "omission" error). Thus, rest or engage in a
distracting task for an for example, if after learning two lists in an
Aequivalent time before receiving a recall test B, A-C relationship subjects
are asked to recall with their A-B list. This A-B, A-C arrangement both
responses to the stimulus terms, they can is the same as that noted earlier
that produces recall both responses to some extent, but pronegative
transfer in learning the second asso- gressively more C's and fewer B's, the
greater ciation. Indeed, across conditions, negative the amount of training
they have just received transfer in second-list learning correlates on the
second (A-C) list (Barnes & Underhighly with the amount of later forgetting
pro- wood, 1959). duced by interference. Proactive interference Although
interference effects have been ilconditions are the converse, examining the
lustrated with paired-associate learning, the decrement in second-list (A-C)
recall caused basic ideas apply to analyses of forgetting in
..
.
...
.
..
...
.
....
.
..
.
..
***
A
.
..17.0,
.
www
A
14
STUDY OF MEMORY
OL - 2
OL-5
IIIIIIO
A-8
A-8
kes o
10
I
NUMBER OF TRIALS A-C
Figure 1.6 Relative frequencies of the originally learned response (A-B) and the newly
learned response (A-C) during learning of the new response. The four graphs are from
four different groups of subjects given 2, 5, 10, or 20 trials of original learning (OL).
(From G. E. Briggs (1957), "Retroactive inhibition as a function of original and
interpolated learning." Journal of Experimental Psychology, 53, p. 64. Reprinted with
permission.)
all learning situations such as serial learning, cation) theory, with its concepts of encoding free
recall, memorizing addition and multipli- and decoding messages, and the concept of cation
tables, and remembering in which of "limited capacity," which applied to any commultiple lists
(or contexts) particular items oc- munication channel (for a review, see Garner, curred. It also
applies to forgetting sentences, 1962). The metaphor was that the human perparagraphs, and
stories when similar concepts ceptual system analyzed the stimulus array, are involved (see
Anderson, 1995).
extracted information from it in the form of neural codes (e.g., in visual or auditory
modal
ities), and that these codes entered into mental
Cognitive Psychology
programs designed either to store them in
brief short-term memory that could hold unat- Karl Pribram (1960), and Cognitive
Psychology tended speech sounds for several seconds be- by Ulric Neisser
(1967). These inspirational fore attention could be switched over to deal books
provided arguments and exciting new with them.
agendas for an "information processing" view
of humans. People were seen as taking
inforComputer Models
mation into a perceptual system, selectively
of Cognition
attending to parts of it, encoding or transform
ing it for use by their cognitive abilities,
storA second impetus to cognitive psychology ing it in memory, and later retrieving it
from was the work of Allen Newell and Herbert Si- memory when an appropriate plan
and remon (1961, 1972), who constructed computer trieval cue were activated. The
analysis of perprograms designed to simulate details of the ception, attention,
immediate sensory memothought processes that people go through as ries, short-term
memory, and the structure of they solve various kinds of problems (e.g., long-term
memory became prominent topics. logic proofs, analogies, chess playing, intelli- Since
S-R theory had very little to contribute gent search). Such simulation programs to these
topics, the major developments in clearly had to have very large memories of memory
theory began breaking away from the structured knowledge about a given domain S-R
perspective. (such as rules and moves of chess), have a means for putting new
information into the
Critiques of S-R Theory system (e.g.,
an opponent's chess moves), have a means for reasoning about the current situa
Concurrent with these developments from tion by manipulating symbolic expressions in
communication engineering and artificial ina central processor (or "short-term memory'),
telligence were a series of effective critiques of and have some means for reporting out
the S-R theory as a means for dealing with com"actions" (moves) the program had
decided to plex varieties of human cognition. The critake. These activities are easily
viewed as tiques attacked the S-R accounts of motor analogous to activities of the
human mind. Si- skills (Lashley, 1951), thinking (Bruner, Goodmon and Newell noted
that the stimulus-re now, & Austin, 1956; Newell, Shaw, & Simon, sponse analyses of
problem solving were of no 1958; Newell & Simon, 1961), pattern recognihelp
whatsoever in deciding how to structure tion (Neisser, 1967; Selfridge & Neisser, 1960),
the knowledge base of the programs in order memory (Anderson & Bower, 1973; Asch,
to carry out efficient retrieval, or how to rea- 1969; Bartlett, 1932; Bower, 1970c; Tulving
& son effectively to achieve specific goals. Along Madigan, 1970), and language (Fodor,
Bever, & with other researchers in artificial intelli- Garrett, 1974; Chomsky, 1959, 1972).
Even in gence, they proceeded to create a number of the traditional S-R domain of
conditioning, computer programs that, by manipulating new investigations of higher
cognitive prosymbols, displayed an impressive amount of cesses in nonhuman animals,
especially simulated "intelligent behavior" of a sort for higher primates and dolphins,
were introducmerly carried out only by humans (see an ing major revisions in
conceptualizations of early account by Feigenbaum & Feldman, cognitive processes
(and learning) in lower 1963). This included not only reasoning pro- organisms (e.g.,
Roitblat, Bever, & Terrace, grams but also learning programs (the EPAM 1984).
program of Feigenbaum, 1959; Simon & Fei- One of the most well-known critiques was
genbaum, 1964; Richmond & Simon, 1989) that by the linguist Noam Chomsky (1959,
and many others. For reviews, see the hand- 1972) in his review of B. F. Skinner's book
books by Barr and Feigenbaum (1981a, 1981b) Verbal Behavior (1957). In his book,
Skinner, and Cohen and Feigenbaum (1981),
the radical arch-behaviorist, attempted an S-R Memory theorists were inspired by such
account of language acquisition and use. developments to step outside the strictures of
Chomsky argued that Skinner's account (and S-R theory and to begin to theorize about
per- by implication, all S-R accounts) was doomed ceptual and attentional processes,
as well as to superficiality since it ignored the obvious strategies and knowledge
structures that were cognitive processes that intervene between a learned and used to
acquire new knowledge. situation and language utterances. He argued The "call to
revolution" was urged in two im- further that extrapolations of terms such as portant
books: Plans and the Structure of Be- "stimulus," "response," "reinforcement," and
havior by George Miller, Eugene Galanter, and "discrimination" from their
animal-laboratory
16
STUDY OF MEMORY
setting (the "Skinner box") to interpret lan- bled orders, and repeatedly recalled. A
freguage behavior were vague, metaphorical, and quent finding is that items presented
at the beno better than common sense. Moreover, ginning ("primacy") and end
("recency") of Chomsky argued strongly that the simple the list are typically recalled
earlier and more ideas of S-R associationism were inadequate often than items
presented in the middle of in principle to account for the way children the list (see figure
1.7). Longer lists and learn their language from experience. Children shorter study times
per item cause poorer learn nonconscious linguistic rules that are far overall recall,
except for the final ("recency") more abstract and complex than warranted by items of
the list. These effects are obvious in the language samples they hear, especially the the
recall curves of figure 1.7. realistic samples containing their many errors, Free recall
was difficult for S-R theories to back-ups, false starts, and ungrammatical ut explain,
since no specific stimulus appeared terances (see Chomsky, 1968, 1972). Chomsky to
be involved and subjects were not obvisimilarly argued that S-R concepts were not
ously making overt responses as they were exrich enough to adequately explain the
com- posed to the study list. Moreover, subjects' plexities of humans' abilities to
comprehend performance seemed to reveal a variety of "orand generate sentences in
context.
ganizational processes" at work-a fact emChomsky's critique and those of his associ-
phasized by Tulving (1968, 1962). For examates established a new direction for the
study ple, when multiple study and recall trials are of language; the new field of
psycholinguistics given with the same list of supposedly unredrew its ideas and
inspiration from new lin- lated words, subjects' improving recall is usuguistic theories
rather than S-R psychology ally accompanied by increasing stereotypy or (Fodor,
Bever, & Garrett, 1974; Miller & Mc- consistency in what items they recall together Neill,
1969). In turn, psycholinguistics had a as clusters (Tulving, 1962). The clusters are
ofvery strong influence on ideas of memory re- ten idiosyncratic groups of 3 to 7 list
words searchers who recognized that some day they among which a subject finds some
kind of would have to understand in detail how and meaningful relationship. With
training, these what people learn from language inputs- subjective clusters grow longer
(include more which, after all, is the medium through which items) and become more
stable. Recall can be most education proceeds.
substantially increased by using words in the
list that have strong prior connections (e.g.,
inPopularity of Different
stances of a taxonomic category), since
subMemory Tasks
jects are likely to discover these inter-item re
lationships and use them for organizing
their As noted, experiments within the S-R para- recall (see also Bower, 1970c;
Mandler, 1967). digm intensively investigated serial learning Such results suggest that
learners are not the and paired-associate learning. With the advent passive tabula rasa
(blank slate) assumed in of the cognitive approach to memory, some traditional
associationism; rather, they are what different questions about memory were very active
in using what they already know asked, leading to different memory paradigms to
search for meaningful relationships among that became popular because they provided
a the learning materials that they can utilize to means for studying different kinds of
memory ease their memorization task. processes. These experimental arrangements
Another impressive fact about free recall is were free recall, context judgments about
that subjects know far more of the list words memorized events, and indirect memory
tests, than they are able to recall when given the along with a renewed emphasis on
short-term general request to "recall the words of that list memory. We shall discuss
these newly popu- you studied." When provided with more spelar memory tasks.
cific associative cues (e.g., a category name),
subjects now recall many list items they
forFree Recall
merly failed to retrieve (Tulving & Pearlstone,
1966; Tulving & Psotka, 1971). In an
impresIn a free recall task, following presentation of sive tour de force regarding the
power of rea set of discrete experiences (words, pictures trieval cues, Mäntylä (1986)
had subjects write of common objects, actions), subjects are 3 meaningful associates to
each of 600 target asked to recall them in any order that they words, with no instructions
for remembering. choose for convenience. The set of items may Seven days later they
received a recall test. be repeatedly presented, often in new scram- When given none of
their associates as cues,
A BRIEF HISTORY OF MEMORY RESEARCH
17
1.00
40-1
PROBABILITY OF RECALL
30-1
fall bear.com
20-1
00 L
0
ILIUIUUUUUUUUUUUUUUUUU 5 10 15 20 25 30 35 40
SERIAL POSITION
Figure 1.7 Serial position curves for free recall of lists of 10, 15, 20, 30, or 40
words each presented for 1 or 2 seconds of study (e.g., 15-2 denotes 15 words
presented at a 2-second rate). (From B. B. Murdock, Jr. (1962), "The serial
position effect of free recall." Journal of Experimental Psychology, 64, pp.
482-488. Reprinted by permission.)
they recalled only 6% of the target words; in the earlier, specified list (called a
"recogniwhen given their 3 associates as cues, they re- tion memory" test). If items
were presented in called 65% of the 600 target words. Mäntylä several different lists,
subjects may be asked and Nilsson (1988) produced even more dra- to indicate in
which list, if any, the test item matic recall (94% at 3 weeks), even without was
presented. Items may be presented several instructions for memorizing, by asking sub-
times in each of two or more distinguishable jects to produce unique and/or distinctive
as lists, and subjects later may be asked to indisociates during original exposure to the
target cate how often a test item was presented in words. The results illustrate the
differing each list. Subjects may also be asked to recall power of different retrieval cues
(viz., one's aspects of the "source"—who said the word, associates versus "the list
studied"). It sug- where or when it was presented, in what mangests that when a
memory is claimed to have ner of voice (male or female speaker), or charbeen
forgotten, the claim should be limited to acteristics of its visual presentation (type font
the particular kind of retrieval test that failed. of word, size, colored vs. black-and-white
If one retrieval cue fails, perhaps a more po- slides). Some experimenters refer to all
these tent set of cues could be found that would re judgments as "source memory,"
whereas othvive the temporarily inaccessible memory. ers distinguish "recognition
memory” judg
ments (which could be based on an undiffer
Context Judgments
entiated feeling of recent familiarity of an
from Memory
item) from the other explicit source judg
ments, In this paradigm, having been
presented with one or more sets of discrete items, subjects are Source Memory. Source
memories are labopresented with a copy of the old items (and ratory analogs of people
remembering where perhaps similar new ones) and asked to re- and from whom they
learned certain informatrieve and judge something about that item's tion. The difficulty
these performances raise context of earlier presentation. The judgment for an S-R
theory is that in no case are subjects may be whether or not the item was presented
overtly responding to these aspects of the ex
18
STUDY OF MEMORY
perience as they are studying the items. cific time, place, and mode of presentation (its
Rather, it seems that the perceptual system is "context," as he says). As evidence,
Hintzman almost automatically recording something like found that subjects were often
able to retrieve sensory images that are retrieved later to sup- some contextual details
about each presentaport such judgments.
tion of an item, almost in the manner of a vidA basic finding of source memory is that
eotape recording. For example, subjects might remembering which source provided
particu- experience presentation of a long list in which lar information grows worse the
more similar 30 unrelated words are presented once mixed are the salient properties of
the several in among 15 presented twice and 10 presented sources. For example,
people are more likely four times, all scattered randomly throughout to confuse in
memory which of two unfamiliar a 100-presentation list. Upon later testing with speakers
said a particular message if the the test word table, a hypothetical subject speakers look
and sound alike. Source confu- might be able to report correctly that "the sions also
increase the less associated are the word table was presented twice: its first
precontents of the messages with their respective sentation was in the first quarter of
the list and sources. For example, readers are more likely was spoken in a male voice;
its second presento ascribe their memory of an alien abduction tation was in the third
quarter of the list and it story to having read it in the National En- was presented visually
in red Gothic letters." quirer than in the more credible New York While such contextual
details are generally Times.
poorly recalled (as would be expected by the A
second fact is that material people imag- similarity-based interference they suffer), their
ine or produce themselves (e.g., as associates moderate accuracy indicates that
subjects are to something they saw) can be confused later often successfully storing
detailed records of in memory with something they actually saw several different
experienced presentations of or heard. This discrimination of "my internal the same
word. This in turn supports the mulvs. his external" event improves the more tiple-trace
over the strengthened single-trace thinking or cognitive effort subjects engage in
account of how repetition works in memory. while generating the imagined scene origi-
With higher frequencies (10 or more), individnally. This is likely due to the fact that the
ual memories of a given item's presentations imagined scene was stored along with a
record greatly interfere with one another, and subof the cognitive operations needed to
create jects appear to shift to a "frequency estimathe scene. Later retrieval of these
traces helps tion" rather than "memory counting" strategy people judge the memory as
an imagined one for making judgments of these higher frequen(see Johnson & Raye,
1981; Johnson, Hash- cies. troudi & Lindsay, 1993).
Another interesting type of contextual judgment is the relative recency of
an event. For
example, the subject might be asked to
estiFrequency Judgments. Frequency judgments mate how many items back in a series
(or how ask subjects how often a given word or picture long ago) the word table (vs.
prince) had been was presented in a list. Hintzman (1976; presented. In general,
accuracy of judgments Hintzman & Block, 1971) has brought these declines as a power
function of intervening frequency-in-context judgments to bear upon items or time
elapsed; and discrimination of a fundamental issue in memory theory which of two items
was more recent follows a namely, how multiple repetitions of the nomi- logarithmic
function—that is, discrimination nally same item are treated by the memory varies with
the ratio of their recency differsystem. The traditional theory is that different ence
divided by the recency of the more recent presentations of, say, the word table simply
one (Yntema & Trask, 1963). A multiple-trace strengthen the one association of that
memory theory also provides a better account of how unit to the list context, and that
later judg- repetitions of an item affects its apparent rements of its presentation
frequency could be cency judgment (Flexser & Bower, 1974). derived from the strength
of this one association.
The alternative view favored by Hintzman Recognition Memory. In recognition
memory is that each presentation of the same nominal experiments, subjects judge
whether or not a stimulus establishes a new memory trace, test item was explicitly
presented on a list each associated with its accompanying spe- studied earlier. A basic
fact is that access to an
A BRIEF HISTORY OF MEMORY RESEARCH
19
Pogg-
am-bn hann
es
item's memory requires that subjects interpret The primed word is also more likely to be
gen(or "encode") the item at test in roughly the erated as a member of a category (list
some same manner as they did when it was origi- birds) or as an associate to a related
cue nally studied. This basic fact has been demon- ("Perched in a tree is a red-breasted
strated for a variety of stimuli that have ambig - "). In such indirect tests, the subject
uous (multiple) interpretations, including may produce the primed item with a fresimple
drawings (the duck-rabbit picture; the quency far above an unprimed baseline, and
young wife-old crone picture), naturalistic yet have no subjective awareness of
rememsounds (trotting horses vs. tap dancers), and bering any specific episode from
his or her polysemous words (kitchen TABLE vs. math past. The nature of indirect
memory tests and TABLE). The point is that the nominal stimu- their relation to direct
tests are topics of great lus during the test must arouse the same per- interest in current
research (see, e.g., Bower, cept or meaningful interpretation it did during 1996; Tulving
& Schacter, 1990), and several study in order for it to make contact with the chapters of
this handbook will touch on these memory trace of the original event. This is the issues.
sense of the "encoding specificity principle" proposed by Tulving (1983; Tulving &
Thom- Short-Term Memory Models son, 1973).
Tulving believes the principle applies far In the early 1960s, considerable interest
arose more widely than with just ambiguous words in the study of immediate (or
"short-term") or pictures, since the context of presentation memory. While the limits of
immediate memcan activate somewhat different properties ory and the "memory span"
had been known and associations to thousands of words or pic- since Ebbinghaus (and
described by William tures that are otherwise unambiguous. For ex- James, 1890), the
novel, startling fact observed ample, people think of different aspects of a by Lloyd and
Margaret Peterson (1959) was piano if they speak about tuning it versus lift- that people
very rapidly forget a few unrelated ing it. The aspect aroused by the word during letters
or words (such as TGK or glass, pen, its original context of presentation can then wood)
that they have just read if they are disserve later as the more potent retrieval cue for
tracted and occupied with another task for just re-arousing that memory.
a few seconds. Figure 1.8 illustrates this rapid
loss as subjects were engaged for 1 to 18
secIndirect Memory Tests
onds in a simple subtraction task (subtracting
successive 3's from a starting number)
before The tasks above are called "direct" tests since being asked to recall the three
items they had subjects are asked explicitly to try to remem- just read. Early
investigators had apparently ber an earlier episode, perhaps to recall sev- overlooked
this rapid fall-off in memory for eral features of the context surrounding the trivially small
amounts of material because presented event. These tests may be con- they typically
had used serial lists well betrasted with indirect tests of memory in which yond subjects'
capability for perfect immedisubjects answer general knowledge questions ate recall.
Such results suggest a short-term that presumably do not refer to recent, specific
memory of extreme fragility, lasting only a few experiences (Richardson-Klavehn &
Bjork, seconds after the subject's attention is drawn 1988; Schacter, 1987). Yet it can be
shown that elsewhere. This rapid loss may be contrasted recent experiences affect the
speed and con- to the long-term memories for information tent of what is retrieved from
one's general people have stored in their knowledge base. knowledge.
The next decade in memory research was The prototypic example is repetition prim-
filled with many studies of such short-term ing wherein a word (or picture of a common
memories: many different paradigms were inobject) is accessed more readily if it is a
repeti vestigated, and many variables that controlled tion of one recently presented.
Such priming short-term retention were investigated. As evi(say, of robin) shows up in
facilitated percep- dence accumulated, several formal theories tual identification (seeing
it more accurately were proposed to explain the interaction bein a degraded or quick
flash), in lexical deci- tween short-term and long-term memory. sions (that robin is a
word, but ribon is not). These built on the earlier proposals of Broadand in fragment
completion ("Give the first bent (1958), Bower (1964), and Waugh and word you can
think of to complete r_bi_") Norman (1965) (reviewed in Murdock, 1974).
ESAPARE
..
. AVERDAD
E
20
STUDY OF MEMORY
Recall probability
LLLLLLLLL
TTTTTTTTTT,
3
IIIII 6 9 12 15 18
Recall interval (sec)
Figure 1.8 Correct recall of 3 letters as a function of a retention interval filled with a
"counting backwards" distraction task. (From L. R. Peterson and M. J. Peterson (1959),
"Short-term retention of individual verbal items,” Journal of Experimental Psychology,
58, pp. 193– 198. Copyright 1959 by the American Psychological Association.
Reprinted by permission.)
The most popular model of this period was transfers into the more durable long-term
that proposed by Atkinson and Shiffrin (1968, memory information about the
associations 1971) because it was the most explicit and was being studied. Information
transferred to longapplied to the widest range of results.
term memory can also be forgotten, but at a
much slower rate than the information in A
Popular Model of
STM. A to-be-remembered item will be even
tually displaced or lost from STM since newly
Short-Term Memory
arriving information enters STM and bumps
Figure 1.9 shows the general structure of the out the earlier items. The greater the
demands three memory stores postulated: a sensory
or difficulty of the interpolated task, the more
store that held briefly glimpsed (or heard) rehearsal suffers and so the poorer the reten
pes for a couple of seconds before their tion of the target items. traces decayed. Stimulus
traces in the sensory store that are attended to, identified, and re
Benefits of the Short-Term coded are thereby
entered into a short-term
Memory Model memory (STM). The STM
is of limited capac
ity, holding at most only a few items (e.g., 4 Several features of such models should be
emwords or 2 paired associates). If the items are phasized. First, they permit flexibility in the to
be learned, the central-control processes initial encoding of the input coding. For norinitiate a
plan for memorizing, such as covert mal speakers, many to-be-remembered stimuli rehearsal
(silently going over the words). The will be encoded by their names or descriptors. model
assumes that each rehearsal cycle also Thus, later confusions in recall of letter or
A BRIEF HISTORY OF MEMORY RESEARCH
21
Short-term memory
24
STUDY OF MEMORY
types, and routine, stereotyped actions
(e.g., Concepts and Schemas
making a telephone call, visiting the
dentist, As noted, language understanders require a buying cinema tickets,
buying groceries at a huge network of concepts and their interrela- check-out
counter, attending a classroom lections. Psychologists have speculated exten-
ture). Each stereotyped activity is composed of sively about the nature of
concepts and how an ordered sequence of actions to achieve they might be
structured in human memory. some goal, and it "runs off" in behavior (and
Concepts are obviously represented in terms in recall chunks) more or less
automatically. of their associated properties, perceptual fea- People use their
schema knowledge not only tures, intended uses or functions, and their re- to
guide their actions but also to understand lations to other concepts. While an
early view those of others (as viewed or described in narwas that concepts
should be represented in ratives). Through experience people acquire memory
as a set of defining features that are schemas regarding novel activities (e.g.,
rollerall necessary and jointly sufficient to recog blading, talking on cellular
phones). They also nize instances (e.g., the concept of a triangle), acquire
schemas for familiar narrative forms, the more comprehensive view that has their
coherence (e.g., the hero acts in order to emerged proposes that naturalistic
concepts resolve some problem), and the genre of differare better represented in
memory as probabi ent narratives (e.g., cowboy Westerns, science listic
prototypes (like a summary). Different fiction, parodies). It is often claimed that
peofeatures of the prototype are more or less diag- ple understand events (or
their narration) by nostic, and features that are not at all diagnos being able to
explain them by fitting them into tic may nonetheless be so characteristic of the
familiar schemas-familiar abstract patterns class that they are used for rapid
identification of relationships. Thus, movie viewers underof instances (see Smith
& Medin, 1981; stand immediately why a sheriff in a Western Rosch & Mervis,
1975). If concepts are proba- movie jumps on his horse to chase bank robbilistic
prototypes, different examples will bers: his motivations flow from his role and
vary in how closely they match the prototype, his plans along with constraints
imposed by and hence vary in how "good" (or typical) an his limited resources.
example they are of the concept. For example, A final point is that these
structured schea robin is a very typical bird, a stork somewhat mas can be
viewed as simply large clusters of less, and an ostrich or chicken a very poor ex-
elementary associations (or expectations) such ample of a bird.
as those studied in the laboratory. When a In
addition to such object concepts, seman- subject hears a sentence asserting new relatic
memory includes many concepts about re tionships (e.g., "President Clinton is now
vislations such as spatial prepositions (on, in, iting South Africa"), memory theorists can
around, left of) and actions. Actions are typi- think of this as the person recording into
cally represented as verbs in English, and have memory a novel set of associations that
the associated frames or schemas that stipulate subject did not have
before-associations roles for participants in the action scenario. which can be used to
answer questions Verbs like buy, rent, or steal denote schemas ("Where's Clinton
now?") and to draw inferthat put together concepts in a certain sce- ences using other
knowledge (e.g., "He's probnario. So, "John bought Bill's car" means that ably not
speaking at the U.N. today"). Bill transferred ownership of a car to John, The new
wrinkle that is added to tradiand John transferred ownership of something tional
associationism is that the associations valuable (probably money) to Bill in exchange
are labeled according to their logical type or for his car.
schematic role (see Anderson & Bower, 1973: Schemas (or frames) are good
candidates Norman & Rumelhart, 1975). To illustrate, cafor how long-term memory
might be struc nary is associated with the concepts of bird, tured (Rumelhart & Ortony,
1977). Schemas yellow, legs, and Tweety, yet people apprecicapture clusters of
organized expectations and ate the different relationships-namely, that a represent
abstract knowledge about some do- canary is a kind of bird, yellow is a property main.
Schemas can be of any "size" or "grain," of its appearance, legs are a part of canaries,
and some schemas (e.g., eyes and mouth) can and Tweety is an example of a canary.
These be embedded within other schemas (human kinds of relations are appreciated as
rapidly as face). Common examples of schemas come the items themselves. Moreover,
relations are from common objects (e.g., birds, horses, cars, useful for carrying out
directed search and refaces, rooms), ethnic and personality stereo- trieval operations,
as when people are asked
..
...
4
....
..
A BRIEF HISTORY OF MEMORY RESEARCH
25
to "List some parts of canaries.” So, in more manner are stored in an "imagery' memory recent
models of memory, the associations are store that is more durable than the verbal assumed to
be recorded into memory along memory store. Although the two coding syswith an attached
relational label specifying its tems are closely interconnected (so people can type (see, e.g.,
Anderson & Bower, 1973; Nor translate between them), memory traces estabman & Rumelhart,
1975).
26
STUDY OF MEMORY
WET
Tesch-Romer, 1993; Rubin, 1995), restaurant used to behave (or their former
opinions) to waiters who remember large number of meal be more consistent
with their opinion of today orders (Ericsson et al., 1993), and repertory (Ross,
1989). These and other tendencies sugactors who memorize and must rapidly
re- gest that social, motivational, and personality trieve lengthy roles in several
plays (Noice, factors play a significant role in the way mem1991, 1992). These
studies often provide valu- ories are altered over time. More studies of able
insights into the means by which people this kind can be expected as the field
aduse limited resources very efficiently in order vances and makes better contact
with other to surmount limitations of memory. However, branches of psychology,
Banaji and Crowder (1989) have argued that the "everyday memory" movement
has turned up very few novel phenomena or principles
References that had not already been
known for many Anderson, J. R. (1978). Arguments concerning years based on
laboratory research. (For op- representations for mental imagery. Psychoposing
arguments, see articles in the American logical Review, 85, 249-277. Psychologist,
January 1991 (volume 46, Issue Anderson, J. R. (1983). The architecture of #1], pp.
16-48.)
cognition. Cambridge, MA: Harvard
UniOne especially fascinating aspect of the ev. versity Press. eryday-memory
movement is its investigation Anderson, J. R. (1995). Learning and memory: of
autobiographical memory (e.g., Rubin, An integrated approach. New York: John 1986;
Conway, 1992; Conway & Bekerian, Wiley. 1987; Conway & Rubin, 1993). By collecting
Anderson, J. R., & Bower, G. H. (1973). Huand classifying phenomenological reports,
man associative memory.
Washington, psychologists have constructed theories of
D.C.: D. H. Winston. how people
structure in memory large seg
Asch, S. E. (1969). Reformulation of
the probments of their life, how they use salient
lem of association. American Psychologist,
28
STUDY OF MEMORY
Bartlett, F. C. (1932). Remembering: A study
in experimental and social psychology.
Cambridge: Cambridge University Press. Bergson, H. (1911). Matter and memory. Lon-
don: Allen & Unwin. Bower, G. H. (1964). A descriptive theory of
memory. Speech presented at the Second International Inter-Disciplinary Conference
on Learning, Remembering, and Forgetting, Princeton, New Jersey. (Published in D. T.
Kimble (Ed.) (1967), The Organization of Recall (pp. 112–185). New York:
New York Academy of Sciences. Bower, G. H. (1970a). Analysis of a mne
monic device. American Scientist, 58,
496-510. Bower, G. H. (1970b). Imagery as a relational
organizer in associative learning. Journal of Verbal Learning and Verbal Behavior, 9,
529-533. Bower, G, H. (1970c). Organizational factors
in memory. Cognitive Psychology, 1,
18-46. Bower, G, H. (1972). Mental imagery and asso
ciative learning. In L. Gregg (Ed.), Cognition in learning and memory (pp. 51-88).
New York: John Wiley. Bower, G. H. (1996). Reactivating a reactiva-
tion theory of implicit memory. Conscious
ness & Cognition, 5, 27–72. Bower, G. H., & Hilgard, E. R. (1981). Theo
ries of learning (5th ed.). Englewood Cliffs,
NJ: Prentice-Hall. Briggs, G. E. (1957). Retroactive inhibition as
a function of original and interpolated learning. Journal of Experimental Psychol.
ogy, 53, 64. Broadbent, D, E. (1958). Perception and com
bijulish
ity for processing information, Psychologi-
cal Review, 63, 81-97. Miller, G, A., Galanter, E., & Pribram, K. H.
(1960). Plans and the structure of behav
ior. New York: Holt, Rinehart & Winston. Miller, G. A., & McNeill, D. (1969). Psycholin
guistics. In G. Lindzey & E. Aronson (Eds.), The Handbook of social psychology (pp.
666-794) 2nd ed., vol. 3; Reading,
MA: Addison-Wesley. Montague, W. E. (1972). Elaborative strategies
in verbal learning and memory. In G. H. Bower (Ed.), The psychology of learning and
motivation: Advances in research and theory (pp. 225-302). New York: Aca
demic Press, Montague, W. E., & Kiess, H. O. (1968). The
associability of CVC pairs. Journal of Experimental Psychology Monograph, Sup
plement 78, no. 2, part 2, 1-23. Murdock, B. B., Jr. (1962), The serial position
effect of free recall. Journal of Experimen
tal Psychology, 64, 482–488. Murdock, B. B., Jr. (1974). Human memory:
Theory and data. Potomac, MD: Erlbaum. Neisser, U. (1967). Cognitive psychology.
New
York: Appleton-Century-Crofts. Neisser, U. (1978). Memory: What are the im
portant questions? In M. M. Gruneberg, P. E. Morris, & R. N. Sykes (Eds.), Practical
aspects of memory (pp. 1-24). London: AC
ademic Press. Neisser, U. (Ed.). (1982), Memory observed:
Remembering in natural contexts. San
Francisco: Freeman, Newell, A., Shaw, J. C., & Simon, H. A.
(1958). Elements of a theory of human problem-solving. Psychological Review, 65,
151-166. Newell, A., & Simon, H. A. (1961). Computer
simulation of human thinking. Science,
134, 2011-2017. Newell, A., & Simon, H. A. (1972). Human
problem-solving. Englewood Cliffs, NJ:
Prentice-Hall. Noble, C. E. (1952). An analysis of meaning.
Psychological Review, 59, 421-430. Noble, C. E., & Parker, G. V. C. (1969). The
Montana scale of meaningfulness (m). Psy