Dual Coding Theory
Jiyeon Ryu, Tingling Lai, Susan Colaric
Joanne Cawley, Habibe Aldag
Educational Psychology 523
February 29, 2000
The role of imagery in human cognitive processing has sometimes been
ignored in favor of an emphasis on verbal processing. But in experiments
conducted during the last three decades, Allan Paivio has carefully developed
an empirical research agenda that has led to a theory that places equal
importance on both verbal and non-verbal processing. The Dual Coding Theory
assumes there are two cognitive subsystems, one specialized for the representation
and processing of nonverbal objects and the other specialized for dealing
with language. The theory has been applied to many cognitive phenomena
including mnemonics, problem-solving, concept learning, and language. This
paper will explore the development of the theory, explain the major components
of the theory with a specific focus on its use in problem solving, discuss
neuropsychological evidence, present criticisms of the theory, examine
the applications of the theory and end with a brief discussion of research
needed in the future.
Development of the Theory
Paivio began his graduate studies at McGill University in 1954 working
with Dr. Wallace Lambert in the area of bilingual processing. As graduate
students often do, Paivio borrowed from his advisor's research interests
to conduct an experiment as part of his coursework. In this initial study
Lambert and Paivio (1956) examined the difference between the adjective-noun
order and the noun-adjective order as indicators for recall. The basis
for this study was a disagreement with the associative habit view of language
behavior that stated that the sequences of nouns and modifying adjectives
would be learned more easily when the adjectives are listed ahead of the
nouns rather than after them. They believed the noun-adjective pairing
would be more effective as demonstrated in image mnemonics techniques being
used as a memory aid in public speaking. This was in fact true, leading
to the development of the conceptual peg hypothesis of image-noun associative
memory – the idea that nouns, since they are more concrete then adjectives,
serve as memory "pegs" for associated items in memory.
This study laid the basis for Paivio's research agenda in the 1960s;
"the initial purpose of my research program in the early 1960s was to identify
the effective variables, concentrating first on the attributes of the words
that served as conceptual pegs and next on the thought processes they arouse,
which together might explain why they help us to remember new information" (Paivio, 1991a). A study conducted in 1963
(Paivio, 1963) confirmed noun-adjective
pairs were learned more easily in the noun-adjective order than the reverse
and that recall increased when the nouns were concrete rather than abstract.
This was explored further in his 1965 study in which he found learning
benefited more from noun concreteness when the concrete nouns served as
stimuli rather than responses for their associated adjectives. It was at
this time that he hypothesized that the imagery of the noun was the salient
variable in mediating recall; "concrete nouns are superior to abstract
nouns in their capacity to elicit sensory images, and that imagery can
mediate the formation of an associative connection between members of a
pair (p. 37)". Paivio and Yarmey (1966) sought to verify this hypothesis
through the collection of data on participants' use of imagery, verbal
and other strategies during the learning of the paired words. They found
that self-reported imagery use was related to higher recall.
The next step in the research agenda was to eliminate other attributes
which had the potential to be relevant in aiding recall (Paivio, 1968).
Such other factors were identified as measures of imagery and concreteness,
familiarity, distinctiveness, interest value, verbal associative meaningfulness
and associative reaction time. Subjects were asked to scale 96 nouns on
30 variables and a factor analysis was performed to examine word attributes
and verbal learning. Memory scores were correlated with different attributes
of the nouns. Measures of concreteness and imagery (which were highly correlated
with each other) were the strongest correlates of noun recall, especially
with pair associates when the word attributes were varied on the stimulus
side of pairs. Rated ease of imagery was "the best single predictor of
recall scores for words, especially when the words served as stimulus members
in paired associated learning" (Paivio, 1991b).
In a further effort to distinguish between imagery and verbal processes,
Paivio and Yuille (1969) designed a series of studies to test the effects
of imagery instructions, item attributes, and reported learning strategies.
They concluded, "The common effective process was indeed imagery, which
was generated to word pairs during learning, and reactivated by a concrete
stimulus word during recall so that the response word could be retrieved
from the image." (Paivio, 1991a, p. 5).
The results from this series of studies confirmed the conceptual peg
hypothesis of paired associate memory tasks, generalized the hypothesis
to other memory tasks and lead to the development of the Dual Coding Theory
in which images and verbal representations exist in two separate systems
that complement and supplement each other in developing memory.
Theory and Assumptions
Dual Coding Theory represents a set of assumptions, hypotheses and the
implications about the origins and structural and functional properties
of representational systems. It is based on the general view that cognition
consists of representational systems called verbal and nonverbal systems.
These are specialized for dealing with environmental information in such
a way as to serve functional and adaptive behavioral goals.
Although the theory does not deny the human brain as a product of a
biological evolution with species-specific functional properties, the main
focus of the theory is based on specific experiences. The functionality
of useful content of language and imagery can vary enormously, depending
on specific experiences of each individual.
The representational systems constitute a person's perceptual, affective,
and behavioral knowledge; in other words,
internal/mental representations have their developmental origin in perceptual,
motor and effective experience.
According to Paivio, "Human cognition is unique in that it has become
specialized for dealing simultaneously with language and with nonverbal
objects and events. Moreover, the language is peculiar in that it deals
directly with linguistic input and output (in the form of speech and writing)
while at the same time serving a symbolic function with respect to nonverbal
objects, events and behaviors. Any representational theory must represent
this dual functionality" (1986, p. 53).
Paivio refers to the nonverbal (symbolic) subsystem as the imagery system
because its critical functions include the analysis of scenes and the generation
of mental images (both functions encompassing other sensory modalities
such as environmental sounds, feel of object in addition to visual). The
verbal system deals with language specialized representations which includes
auditory, visual words, end writing patterns of these words.
Symbolic and sensorimotor systems
Symbolic representations retain the properties of different sensory
and response modalities. The verbal-nonverbal symbolic distinction is assumed
to be orthogonal to the sensory modalities. For example, if the object
is a word in auditory form or in a visual form, then it will be represented
in verbal system. However, if it is in auditory but not related to language;
then it will be represented in nonverbal/imagery representations.
The Representation-Processing (or Structure-Function) Distinction
The structural representations refer to relatively stable long-term
information corresponding to perceptually identifiable objects and activities,
both verbal and nonverbal. Processing refers to functional activities that
include activation of either by appropriate stimuli (encoding), activation
of one by the other (recoding), organization and the elaboration within
each, as well as transformation, manipulation, and retrieval of information
from either class.
Conceptual Structure of the Theory
Dual Coding Theory has a hierarchical conceptual structure. At the most
general level, the theory is about symbolic systems, that is, cognitive
systems that serve a symbolic or representational function. The general
level divides into verbal, nonverbal symbolic subsystems, which in turn
expand into sensorimotor (visual, auditory, haptic) subsystems at the next
level. The lowest level consists of the hypothetical representational units
of each system, called logogens and imagens.
The units of the representational systems, imagens and
be imagined as being similar to "chunks". They are unitary in the sense
but can function as integrated informational structures. The main assumptions
related to unitary structures are:
The representational units in each system are modality specific perceptual-motor
Units are hierarchically organized structures, vary in size.
Nonverbal units, imagens, is synchronously organized (permitting parallel/simultaneous
processing up to some informational limit), for example a face, part of
the face or a human body. Whereas, verbal components are sequentially organized
from small to larger units, for example, from syllabus, words, poems, to
stories (implying sequential constraints on intraunit processing).
The two systems are assumed to be structurally and functionally distinct.
Structurally, they differ in the nature of representational units and the
units organized into higher order structures. Functionally, they are independent
in the sense that either system can be active without the other or both
can be active in parallel. At the same time, they are functionally interconnected
so that activity in one system can initiate activity in the other. The
structural and functional distinctions combine to produce qualitative differences
in the kinds of processing.
The main assumptions related to system level are:
The functional independence but interconnectedness of the two systems,
The probabilistic nature of interunit relations between and within systems
Processing mechanism and different levels of processing of (representational,
referential, and associative) within and between system
Differential specialization for synchronous and sequential interunit processing
Processing operations: The functional properties of the representational
systems make it possible for representational information to be used in
cognitive tasks and the guidance of behavior.
The operative cognitive mechanisms are assumed to be at the unit level.
There is no top-down activating mechanism except in the sense that specific
input may activate complex representational structures from which component
information can be retrieved, or activate units that have a general control
function on subsequent processing, e. g. the instruction to image words
will activate the nonverbal system. Properties of processing are:
a) Activation of representations: The overall probability
of the activation and use of verbal and nonverbal representations is a
function of the combined effect of stimulus situation and relevant individual
difference variables. The stimulus variables include attributes of target
stimuli (such as items to be remembered, comprehended), and contextual
stimuli (such as instructions to arouse a set of task) in a given task.
b) Different levels of processing within and between systems:
Three types of processing are identified: (1) representational, the direct
activation of verbal or nonverbal representations, (2) referential, the
activation of the verbal system by the normal system or vice-versa, and
(3) associative processing, the activation of representations within the
same verbal or nonverbal systems. A given task may require any or all level
c) Organizational and transformational processing: Differential specialization
for synchronous (nonverbal) and sequential (verbal) interunit processing
d) Automatic and conscious level processing in both systems: Researches
on imageless thought and vividness of mental imagery supports the general
hypothesis that nonverbal symbolic systems can function at the unconscious
level as well as conscious level.
Basic Functional Categories Served by the Symbolic Systems
The functions and categories of symbolic systems implicates how all
the structural and functional properties of symbolic systems are utilized
in the case of complex task such as problem solving or remembering a set
of properties. Two main functions are mnemonic and evaluative.
Dual Coding Theory suggests that the memory trace is a modality-specific
encoded representation of verbal and nonverbal input information. Some
portion of the total encoded perceptual-motor reaction pattern constitutes
the episodic memory trace.
Representational encoding is automatic and sufficient for come level
of accuracy in recognition and recall test of memory (Pavio, 1986). Given
appropriate contextual cues and the instructions to image, referential
encoding may occur with high probability. The associative encoding occurs
in the same representational unit such as sentences to verbal items or
imaginal context to pictures.
Retrieval context as a production of the recoding activity is never
identical to the input context. The encoding-recoding differences are sharper
and more explicit in associative recall task. This implies that recoding
is subject to the same influences as memory for the target episode itself.
The independent functional structure of imaginal and verbal codes can
have an additive affects on recall. However, imaginal and verbal codes
are unequal in mnemonic value perhaps by a 2:1 ratio favoring image code (Pavio, 1986). The sequential structure of verbal representation makes
it more difficult to generate a functionally integrated verbal trace in
a single trial which may account for this unequal mnemonic value.
The term evaluation refers to the determination of quantitative values
of object and events: a) Scanning mental images to determine the relative
distances between locations on imaged maps, b) counting the corners of
an imagined blocked letter, c) reading out the information in an imagined
matrix, and symbolic comparisons on any perceptual dimensions such as size,
color are the examples from the research studies which represents evaluative
functions. These tasks utilize both symbolic systems most of the time.
Verbal cues can initiate and guide evaluative processing. The evaluative
functional processes are largely under conscious control of verbal mechanism.
The following section explains how the Dual Coding accounts for problem
Paivio (1986) relates Dual Coding Theory to the area of problem solving;
"performance is mediated by the joint activity of verbal and nonverbal
systems, with the relative contribution of each system depending on characteristics
of the task and cognitive abilities and habits of the performer. The more
concrete and nonverbal the task, the greater the contribution of the imagery
system; the more abstract and verbal the task, the greater the contribution
of the verbal system" (p. 201). In the same publication he noted that there
are differences in imaginal and verbal contributions to problem solving.
The imagery system is synchronously organized, meaning that portions of
images are retrieved in no particular order because an image is perceptually
available in its entirety (Paivio, 1991a). Therefore, retrieved information
is simultaneously available, and there is flexibility in processing content.
Paivio (1986) suggests that creative thinking may result from these imaginal
attributes. On the other hand, information from the verbal system
is sequentially retrieved. This suggests that the verbal system contributes
logical order to problem solving.
The interplay between the imagery and verbal systems is illustrated
Paivio’s explanation of creativity. Paivio (1971) notes that creative discoveries
have often originated with the use of concrete imagery. For example, Kekulé’s
invention of the benzene ring (an organic chemistry devise) emerged from
his visualization of a long row of atoms as a row of snakes, one of which
gripped its own tail. The later development of such theories likely relies
on the more logical verbal system.
Paivio’s theory of problem solving (1986) is illustrated through several
tasks: cognitive maps, syllogistic reasoning, and mental practice effects
on motor skills.
Cognitive map tasks involve spatial problem solving
The cognitive map research started with Tolman, who hypothesized that rats
used a cognitive map to navigate through mazes. This finding, and other
research, suggests that the use of cognitive maps involves synchronous
mental representations. The principle of equiavailability states that when
individuals learn a route map through sequential processing, they access
that information as a picture-like map in which locations and routes are
simultaneously available. Mapping may also facilitate basic recall. For
example, Amlund, Gaffney and Kulhavy (1985) found that when fifth and sixth
graders study a mimetic map (map features are drawings, not just labels)
while listening to a story related to the map content, they recall the
text better than students who do not use such maps. This study also supports
Dual Coding Theory’s contention that the interaction of spatial and verbal
information facilitates learning by increasing the number of relevant cues.
Syllogistic reasoning also illustrates problem solving from a Dual Coding
Theory perspective (Paivio, 1986). An example of a syllogistic reasoning
task is the following: "Bob is older than Jim. Kevin is younger than Jim.
Who is older?" Such tasks require deductive reasoning or forming a conclusion
from given information. Johnson-Laird (1972) suggests that learners may
alter strategies as they become more experienced with these tasks. For
example, learners may initially use an imagery strategy then switch to
a linguistic strategy. Also, individuals may use a linguistic or an imagery
strategy to solve this depending on their preferred strategy to use to
solve these problems.
Mental Practice Effects on Motor Skills
Research has show that mental practice affects motor skills. In a meta-analysis,
Feltx and Landers (1983) found that mental practice influences performance
more than no practice at all. In a specific context, Doheny (1993) found
that nursing students who used a guided imagery strategy performed better
at giving an intramuscular injection than students who did not use such
a strategy. Paivio (1986) explains such findings in light of motivational
and cognitive functions of imaginal and verbal processes. Motivational
effects includes physiological arousal, positive affect and goal directedness.
These motivational effects result in less tension, less negative verbalization
and images of failure, and more images of success. Mental practice cognitively
influences general behavioral strategies and specific motor responses.
Successful use of mental practice hinges on accurate and precise images
of successful performance (based on either previous experience or observation)
and how well the learner remembers this information during performance.
Representational and Referential Processing in Problem Solving
Mayer’s work also provides evidence for the problem solving implications
of Dual Coding Theory, in addition to support for the theory in general.
Mayer and Anderson (1991) note that the presentation of both verbal and
visual information promotes creative problem solving more than if verbal
and visual information is given separately. Using a two experiment approach,
Mayer and Anderson (1991) found that students who are given a verbal description
at the same time as animation significantly outperform other students (animation
without words, words without animation, and no training) on a measure of
creative problem solving. However, students did not differ on recall of
verbal information. Mayer and Sims (1994) found similar results in a study
using scientific content. Mayer and Anderson explain such findings in light
of Dual Coding’s principles of referential and representational connections.
Students in the verbal and animation condition have more opportunities
to build referential connections than the other students since they are
concurrently presented with verbal and visual information. Referential
connections facilitate problem solving. On the other hand, learners do
not differ in the extent of representational connections because all students
had an opportunity to form such connections.
Neuropsychological Evidence for
Dual Coding Theory
Paivio’s early research, discussed earlier, is confirmed by a large
number of recent studies that examine the independence hypothesis (Thompson
& Paivio, 1994), referential processing (Paivio, Clark, Digdon, & Bons, 1989), and recall and problem solving (Mayer & Sims, 1994). Furthermore,
there is neuropsychological evidence for Dual Coding Theory.
Dual Coding Theory is supported by neuropsychological evidence on the
functions of the hemispheres of the brain. EEG studies (Ley, 1983) support
the contention that the left hemisphere is primarily in control of verbal
processing, while the right and left hemispheres both contribute to performance
in nonverbal tasks. Paivio (1986) claims that such studies support the
independence assumption of Dual Coding Theory.
The neuropsychological evidence also supports representational and referential
processing. In Paivio and Ernest’s 1971 experiment, images were presented
via tachistoscope to either the left or right visual fields. The researchers
found that field had no effect on a recognition task. Their study suggests
that both hemispheres have the representations and processing skills for
Referential processing evidence comes mainly from brain damage research
(Paivio, 1991). People with damage to the left hemisphere have difficulty
naming objects but do not have any problems recognizing objects. This suggests
that referential processing has been impaired
but verbal representational and associative processes are unharmed.
The sequential and synchronous processing distinction is also addressed
in neuropsychological research (Paivio, 1986). The evidence implies that
the anterior regions of the left hemisphere are dominant in tasks that
call for sequential processing, such as speech. This evidence is consistent
with Dual Coding Theory’s assumption that verbal processing is sequential.
In contrast, synchronous processing seems to generate mostly from the posterior
regions of both hemispheres. Individuals with posterior brain
damage have difficulty with spatial organization, suggesting that
the posterior region of the brain is associated with imaginal processing.
Criticism of Dual Coding Theory
Discussion and criticism of Dual Coding Theory has been addressed by
Paivio as well as other researchers in the field. In 1982, Paivio reviewed
the studies related to the Dual Coding Theory, and identified sixty positive
findings that supported it. At the same time, he also categorized negative
evidence into three types: pseudo-negative findings, failure to find significant
differences, and differences that were inconsistent with Dual Coding Theory.
Pseudo-negative findings were those that were based on misinterpreatations
of the data, such as misunderstanding that Dual Coding Theory claims picture
naming is automatic, but Paivio said that he has never claimed automaticity
of picture naming (Paivio, 1982). Such studies would have shown consistency
with Dual Coding Theory if they had been interpreted correctly. Failure
to find significant differences were those such as the Marschark and Paivio’s
(1977) study regarding cue recall of concrete and abstract sentences.
The third type, such as asymmetrical transfer pattern between pictures
and words (Paivio and Marschark, 1980), is what Paivio declared were inconsistent
with the theoretical assumptions of Dual Coding Theory.
Based on the negative evidence that Paivio listed above, this section
intends to update the current status of Dual Coding Theory, and furthermore,
to present some criticism that was found in the literature.
Concrete and abstract verbal material fail to differ in tasks where
difference was expected
Paivio and his colleagues (e.g., Marschark and
Paivio, 1977) have repeatedly
failed to interpret integration of processing of memory for concrete and
abstract sentences, although overall memory scores were always better for
the concrete sentences than abstract sentences. The superiority of concreteness
is a common finding completely in accord with the Dual Coding Theory, but
the equivalent integration of what is remembered from concrete and abstract
sentences seems to require an additional coding system. Although their
recent experiments (Paivio, Walsh & Bons, 1994) reported that paired
concreteness and relatedness have independent and additive effects on free
recall and cued recall and further supporting the integration of memory,
the results still could not rule out the possible interaction on these
variables. However, for Paivio, it was not necessary to modify Dual Coding
Theory, even though he agreed that he should "continue to seek interpretations
consistent the basic assumption of Dual Coding Theory." (Paivio, 1982,
Picture–word studies that are inconsistent with Dual Coding Theory
The results of following studies showed inconsistency with Dual Coding
Theory. These results were, at the same time, inconsistent with common
The first study found faster symbolic comparisons with pictures than
with words and with concrete than with abstract words, even in the case
of such abstract attributes as pleasances, monetary value and animal ferocity (Paivio, 1982). Paivio argued that the attributes represented things and
were stored in the imagery system or at least in close association with
the core representations of such things, although abstract concepts that
are less consistently correlated with specific perceptual attribute than
is the case for more concrete concepts. Further, it might be patterned
after the prototypical shape or form of the perceptual object. Other associated
attributes, such as pleasantness, might involve stored proprioceptive or
interocepective information that can vary in a continuous fashion and constitute
the representational base for comparisons.
The second study regards picture word transfer effects in the comparison
task. Paivio & Marschark (1980) obtained asymmetrical transfer effects
with animal intelligence and pleasantness comparisons when subjects first
completed a block of trials with one kind of material and then switched
to comparisons of the same concept pairs with the other type of material.
Switching from pictures to words appeared to have a negative effect whereas
switching from words to pictures had a positive effect on subsequent reaction
time. This result is inconsistent with the Dual Coding Theory.
Another critical result occurred in Linde’s study (1982) regarding the
effect of picture–word manipulations in judgments of associative relatedness
of pairs of items. The original Dual Coding
position was that associative relations between concrete concepts are represented
in both verbal and image system. This implies that associative decisions
could be equally fast for picture pairs and word pairs, which is consistent
with the assumption. However, Linde found that the decisions were just
as fast with mixed picture-word pairs, which presumably would require the
additional step of recoding one concept into the format of the other if
associative relations are only represented within but not across systems.
Thinking about the storage capacity
Pylyshyn (1973) argued that visual imagery stored in picture form would
exceed the storage capacity of the brain and also would require a perceiver
in the brain to read the pictures. However, the visual encoding theorists
maintain that the stored codes are not pictures but rather are analog representations
or analog memories; that is the images are structurally related to the
real objects in the same sense that keys and locks are related. Physically,
keys and locks are quite different. However, only the matched key will
open a particular lock. Similarly, only particular objects will activate
the neural processes with which the object is represented. Then, the problem
will shifted to whether analog representations are the same as those of
verbal information or not. And if they are different from those of semantic
memory, the size of these analog data should be determined into reasonable
Imagen (non verbal encoding system) encoding channel
Zimler and Keenan (1983) conducted three experiments compared congenitally
blind and sighted adults and children on paired associate, free-recall,
and imaging tasks presumed to involve visual imagery in memory, in all
three, blind subjects’ performances were remarkably similar to the sighted.
Hunt and Ellis (1999) mentioned some similar study reporting that blind
subjects performed exactly as sighted subjects did on the memory test.
Instructions to imagine objects as spatially contiguous produced better
memory than did imagining objects as spatially separated for both blind
and sighted subjects. These results suggest that imagery effects upon memory
do not depend upon prior visual experience, nor does the remembered image
seem to be the same thing as a visual perception. And it does not explain
whether the "imagery storage" for blinds and sights are the same.
Conceptual critique: Do we need a Dual Coding Theory?
There are various models that describe human cognitive functions. Some
researchers (e.g. Pylyshyn, 1973) proposed that the verbal mediation alternative
to imagery was replaced by a computational analogue in which the language
of mind become the logical proposition. The approach encompassed language
as well, so that at the deep representational level, all perceptual and
behavioral knowledge was reduced conceptually to a common form. The alternative
was widely adopted in cognitive theories and explicitly contrasts with
Dual Coding Theory (Paivio, 1991a). As Kieras states, "There is no fundamental
difference in how perceptually based and verbal based information is represented
in memory." (Kieras, 1978, pp. 533-534)
In addition to common code theories and Dual Coding Theory, there are
still some other researchers who argue that more information should be
encoded independently. For instance, a triple coding system may exist that
includes imagery, language, and a prepositional interface. Another prospective
is the new computational theories, connectionism. It seems to be prominent
now in cognitive science and cognitive psychology.
The growth of Dual Coding Theory has reshaped the theory and has created
more solid assumptions and reasonable interpretations to human cognitive
functions. Although there are still many unresolved problems (e.g., the
critiques we mentioned above), Paivio and his colleagues work continually
to build a more powerful theoretical model. And along with their work done
so far, the Dual Coding Theory has been applied to a variety of fields.
The following will be some examples of the applications.
Application of the Theory
According to Dual Coding Theory we have more than one information process
system, which implies that it will be better if more than one communication
channel is used during learning. In particular, an increasing body of research
evidence supports the contention that student learning is affected positively
by presenting text and illustrations together (Bernard, 1990; Glenberg
& Langston, 1992; Guri-Rozenblit, 1988; Mayer and Sims, 1994; Purnell
& Solman, 1991; Reed & Beveridge, 1986; 1990; Waddill, McDaniel,
& Einstein, 1988). Furthermore, computer-generated animation offers
a potentially powerful medium for presenting visually based information
to learners (Rieber, 1990b, 1991; White, 1984)
In addition to the presentation of illustration with words, mental images
can be intentionally and systematically induced to aid in recall. Pressley
(1977) found that subjects (age seven and up) perform as well or better
when given instructions to generate images as they do with pictorial representations
of the interactive images; that is, self-generated images are sometimes
more effective than experimenter-imposed images.
Mnemonic Techniques in Primary and Secondary Language Learning
Imagery also can be used as mnemonic techniques in second or foreign
language learning (Paivio, 1980). Vocabulary learning had previously been
shown to increase using a rhyming mnemonic scheme that consists of a series
of ten or twenty peg words that rhyme with numerals which are then used
as stimuli for the recall of new words by imagining the referents of the
words in interaction with the rhyming pegs (Paivio, 1971). The same technique
was applied to foreign language acquisition. Atkinson and his students
(Atkinson, 1975; Atkinson and Raugh, 1975) developed an experimental version
called "the keyword technique". The process involves establishing an acoustic
and an imagery connection between an unfamiliar foreign word and its native
language equivalent. The acoustic link reminds the learner of the relevant
word whereas the imagery link provides a clue to the meaning of the word.
The links are provided by a native-language word that sounds like the foreign
word or part of it. Atkinson and his colleagues (1975) found that the keyword
technique was more effective than standard translation practice and other
non-imagery control conditions when used by American Anglophone students
to learn Russian and Spanish vocabularies.
With respect to the meaning of larger verbal units, such as sentences
and paragraphs, experimental research has also demonstrated a major role
for imagery processes. Educational research has confirmed the importance
of imagery and concreteness for the comprehension of sentences and larger
textual units (Clark and Paivio, 1991). Image generation and supplementary
pictures generally benefit text comprehension (Denis, 1984) and readability
including speed (Glenberg et al., 1987, Flesch, 1950). Moreover, the same
imagery manipulations that benefit memory for text should also benefit
memory for orally presented information as in classroom lessons. Levin
and Berry (1980), for example, asked fourth graders to recall information
from tape-recorded newspaper stories. Children who listened to the stories
while viewing relevant pictures recalled more than children who only heard
Mnemonic Techniques on Study Skills
Another kind of evidence for the important mnemonic role of imagery
in educational learning comes from research on study skills. Clark and
Paivio (1991) found a number of behaviors associated with quality study
skills to be linked to imagery processes as did Kulhavy and Kardash (1988).
Imagery processes are further emphasized in books and programs designed
to improve learning and study skills. An early text by McMurry (1909) included
a chapter on supplementing thought, which recommended the use of such imagery-related
methods as imagination, elaboration, making illustrations, experiencing
the material, and listing details. Contemporary study guides (e.g., Robinson,
1970) and study skills programs stress similar processes. Dansereau et
al. (1979), for example, included imagery and network construction methods
in their program. The network or cognitive mapping methods use spatial
imagery to represent verbal associative structures. Imaginal elaboration
is also a central component in the cognitive learning strategies program
of Weinstein and her colleagues (1979) and in Wittrock’s model of generative
learning (Wittrock and Alesandrini, 1990).
Dual Coding Theory as a Framework for Assessment
Besides providing a theory to promote more effective learning, Dual
Coding Theory provides a useful framework for thinking about cognitive
tests (Clark and Paivio, 1991). A large part of educational assessment
involves tests of intelligence, achievement, and related cognitive processes;
the distinct verbal and nonverbal cognitive systems of Dual Coding Theory
corresponds to analogous distinctions in most test and theories of intelligence.These
tests usually include subscales that measure nonverbal abilities, and factor
analyses of general test batteries identify a perceptual-spatial ability
factor that is distinct from verbal abilities (Anastasi, 1988). Nonverbal
or imaginal processes contribute to the Performance scales of Weschler’s
test (Weschler, 1974) and of Jackson’s (1984) Multidimensional Aptitude
Battery, and to the simultaneous processing dimension (Das et al., 1975; Paivio, 1975) that underlies the Kaufman Assessment Battery for Children
(Kaufman and Kaufman, 1983).
A second illustration of how Dual Coding Theory can model specific assessment
tasks is provided by picture vocabulary tests, which are used widely to
assess child language abilities and dysfunctions in early grades, in special
education and clinical settings, and in many research studies (Denckla
et al., 1981; Snowling et al., 1988; van der Wissel, 1988). Performance
on production tests that require active naming, such as the Expressive
One Word Picture Vocabulary Test (Gardner, 1979), varies with factors that
are associated with a Dual Coding Theory model of picture naming. For example,
pictures with a single label (e.g., apple, scissors) are easier to name
than pictures with multiple labels (e.g., cat, purse). This effect of response
uncertainty (i.e., number of different names) has been observed on the
probability of errors (Johnson and Clark, 1988) and on reaction times in
experimental research on naming (e.g., Lachman, 1973; Paivio et al., 1989).
Dual Coding Theory also provides analytical models for nonverbal tests
and for educational correlates of imagery ability (Clark and Paivio, 1991).
Nonverbal tasks involve the special properties of the imaginal system descried
earlier, such as the capacity to integrate and redintegrate information.
One property that has been particularly important in the individual difference
domain is the capacity of images for spatial transformations. Many spatial
tests, such as mental rotation tasks, involve dynamic transformations of
spatial stimuli. Understanding how imaginal transformations are performed
could therefore shed light on some controversial issues associated with
individual differences in imagery ability and mathematics education, including
controversial relations between gender, mathematics, and imagery ability
(Clark and Paivio, 1991)
In this paper, the positive contributions of Dual Coding Theory as a
general perspective on educational and psychological phenomena has been
described, but many of the topics involve controversial empirical and theoretical
issues that remain unresolved (Clark, and Paivio, 1991). For example, concrete
sentences do not always demonstrate more integration than abstract sentences
on memory tasks, and use of concrete examples in lectures does not always
correlate with learning. Consideration of item attributes, individual differences,
and instructions, and the joint contribution of imagery and verbal processes
to psychological phenomena, need to be explored further (Clark, and Paivio,
Sadoski and Paivio (1994) claim more investigations are needed in several
areas. First, the investigation of the measurement of imagery’s effect
is needed. Traditional multiple-choice tests do not really measure the
effect of imagery. Instead, Gambrell (1982) suggests open-ended prediction
questions to study the imagery of students. Second, we need to consider
and study the effect of individual differences in the spontaneous use of
imagery or in employing imagery as a conscious strategy. For example, according
to Pressley (1977), readers, particularly good readers, spontaneously and
regularly use mental images in reading, and it is therefore difficult to
determine to what extent experimental groups benefit from imagery instruction
or training. Individual differences in this area can exert a confounding
effect in studies, making mental imagery appear less valuable than it may
actually be. Paivio (1988) cited the complex nature and predictive uncertainty
of individual differences as one major problem facing imagery researchers.
Third, a major problem facing researchers is the manner in which imagery
serves to integrate information from text and to what extent such integration
effects are contributed by the verbal system as well (Paivio, 1988). Verbal
associative contexts, such as causal connections, logical sequences, cohesive
ties, and strong verbal associations among words and phrases, may serve
to integrate abstract texts to a degree as well. A central question is
to what extent both imaginal and verbal integration processes operate independently
or interactively (Sadoski and Paivio, 1994). Fourth, the role that Dual
Coding Theory could play in explaining students’ response is another fruitful
area for research. Few theories of reader response have linked themselves
to established theories of psychology. Fifth, the application of Dual Coding
Theory to text design needs to be studied. School textbooks are commonly
criticized as being inadequate to promote learning or subject appreciation.
The design needs to induce more comprehension, retention, and appreciation
and Dual Coding Theory may play a role. Finally, further research involving
the role of imagery in processing metaphor and figurative language and
in composing text, and the objective physiological correlates of the occurrence
of imagery is needed (Sadoski, 1992; Paivio, 1988).
Amlund, J. T., Gaffney, J., & Kulhavy, R. W. (1985). Map feature
content and text recall of good and poor readers. Journal of Reading
Behavior, 17, 317-330.
Anatasi, A. (1988). Psychological Testing (6th ed.),
New York: Macmillan.
Atkinson, R. C. (1975). Mnemotechnics in second-language learning. American
Psychologist, 30, 821-825.
Atkinson, R. C., & Raugh, M. R. (1975). An application of the mnemonic
keyword method to the acquisition of a Russian vocabulary, Journal of
Experimental Psychology: Human Learning and Memory, 104, 126-133.
Clark, J. M., & Paivio, A. (1991). Dual Coding Theory and education.
Educational Psychology Review, 3(3), 1991, 149-210.
Dansereau, D. F., Collins, K. W., McDonald, B. A., Holley, C. D., Garland,
J., Diekhoff, G., and Evans, S. H. (1979). Development and evaluation of
a learning strategy training program. Journal of Educational Psychology,
Das, J. P., Kirby, J., and Jarman, R. F. (1975). Simultaneous and successive
syntheses: An alternative model for cognitive abilities. Psychological
Bulletin, 82, 87-103.
Denckla, M. B., Rudel, R. G., and Broman, M. (1981). Tests that discriminate
between dyslexic and other learning-disabled boys. Brain and Language,
Denis, M. (1984). Imagery and prose: A critical review of research on
adults and children. Text, 4. 381-401.
Doheny, M. O. (1993). Effects of mental practice on performance of a
psychomotor skill. Journal of Mental Imagery, 17, 111-118.
Feltz, D. L. and Landers, D. M. (1983). The effects of mental practice
on motor skill learning and performance: A meta-analysis. Journal of
Sport Psychology, 5, 25-57.
Flesch, R. (1950). Measuring the level of abstraction. Journal of Applied
Psychology, 34, 384-390.
Gambrell, L.B. (1982). Induced mental imagery and the text prediction
performance of first and third graders. In J. A. Niles & L.A. Harris
(Eds.), New inquiries in reading research and instruction (31st
yearbook or the National Reading Conference, 131-135). Rochester, NY: National
Gardner, M. F. (1979). Expressive One-Word Picture Vocabulary Test,
Novato, CA: Academic Therapy Publications.
Glenberg, A. M., Meyer, M., and Lindem, K. (1987). Mental models contribute
to foregrounding during text comprehension. Journal of Memory and Language,
Hunt, R. R. & Ellis, H.C. (1999). Fundamentals of cognitive psychology.
6th ed. Boston, MA: McGraw-Hill.
Jackson, D. N. (1984). Multidimensional Aptitude Battery (MAB): Manual,
Port Huron, MI: Research Psychologists Press.
Johnson, C. J., and Clark, J. M. (1988). Children’s picture naming difficulty
and errors: Effects of age of acquisition, uncertainty, and name generality.
Applied Psycholinguistics, 9, 351-365.
Johnson-Laird, P. N. (1972). The three-term series problem. Cognition,
Kaufmann, A. S., and Kaufman, N. L. (1983). Kaufman Assessment Battery
for Children, Circle Pine, MN: American Guidance Service.
Kieras, D.(1978). Beyond picture and words: Alternative information
processing models for imagery effect in verbal memory. Psychological
Bulletin, 85, 532-554
Kulhavy, R. W., and Kardash, C. A. M. (1988). Studying study: An analysis
of instructional encoding behavior. Educational and Psychological Research,
Lambert, W.E. and Paivio, A. (1956). The influence of noun-adjective
order on learning. Canadian Journal of Psychology, 10, 9-12.
Ley, R. G. (1983). Cerebral laterality and imagery. In A. A. Sheikh
(Ed.), Imagery: Current theory, research, and application. New York:
Marschark, M., & Paivio, A. (1977) Integrative processing of concrete
and abstract sentences. Journal of Verbal Learning and Verbal Behavior,
Mayer, R. E. & Anderson, R. B. (1991). Animations need narrations:
An experimental test of a Dual Coding hypothesis. Journal of Educational
Psychology, 83, 484-490.
Mayer, R. E., & Sims, V. K. (1994). For whom is a picture worth
a thousand words? Extensions of a Dual Coding Theory of Multimedia Learning.
Journal of Educational Psychology, 86(3), 389-401.
McMurry, F. M. (1909). How to study and teaching how to study. New
York: Houghton Mifflin.
Mowrer, O.H. (1960). Learning Theory and the Symbolic Processes.
New York: Wiley.
Paivio, A. (1963). Learning of adjective-noun paired associates as a
function of adjective-noun word order and noun abstractness. Canadian
Journal of Psychology, 17 370-379.
Paivio, A. (1965). Abstractness, imagery, and meaningfulness in paired-associate
learning. Journal of Verbal Learning and Verbal Behavior, 4, 32-38.
Paivio, A. (1968). A factor-analytic study of word attributes and verbal
learning. Journal of Verbal Learning and Verbal Behavior, 7, 41-49.
Paivio, A. (1971). Imagery and Verbal Processes. New York: Holt,
Rinehart and Winston.
Paivio, A. (1975). Imagery and synchronic thinking. Canadian Psychological
Review, 16, 147-163.
Paivio, A.(1978) Mental comparisons involving abstract attributes. Memory
and Cognition, 6, 199-208.
Paivio, A. (1980). Imagery as a private audiovisual aid. Instructional
Science, 9, 295-309.
Paivio, A.(1982) The empirical case for dual coding. In John Yuille
(Ed.). Imagery, memory and cognition: Essays in honor of Allan Paivio.
Paivio, A. (1986). Mental Representations. New York: Oxford University
Paivio, A. (1991a). Images in Mind. New York: Harvester
Paivio, A. (1991b). Dual Coding Theory: Retrospect and current status.
Canadian Journal of Psychology, 45(3), 255-287.
Paivio, A., Clark, J. M., Digdon, N., and Bons, T. (1989). Referential
processing: Reciprocity and correlates of naming and imaging. Memory
and Cognition, 17, 163-174.
Paivio, A. and Csapo, K. (1973) Concrete-image and verbal memory codes.
Journal of Experimental Psychology, 80, 279-285.
Paivio, A. and Ernest, C. (1971). Imagery ability and visual perception
of verbal and nonverbal stimuli. Perception & Psychophysics, 10,
Paivio, A. & Marschark, M.(1980). Comparative judgments of animal
intelligence and pleasantness. Memory and Cognition, 20, 532-539.
Paivio, A. & Walsh, M. Bons, T.(1994) Concreteness effects on memory:
when and why? Journal of Experimental Psychology:Learning Memory and
Cognition, 20, 1196-204.
Paivio, A. and Yarmey, A.D. (1966). Pictures versus words as stimuli
and responses in paired-associate learning. Psychonomic Science,
Paivio, A., Yuille, J.C. and Madigan, S. (1969). Concreteness, imagery,
and meaningfulness values for 925 nouns. Journal of Experimental Psychology
Monograph Supplement, 78, (1, pt. 2).
Pressley, M. (1977). Imagery and children’s learning: putting the picture
in developmental perspective. Review of Educational Research, 47,
Pylyshyn , L.(1973) What the mind’s eye tells the mind’s brain: a critique
of mental imagery. Psychological Bulletin, 80, 1-24.
Robinson, F. P. (1970). Effective Study (4th ed.),
New York: Harper & Row.
Sadoski, M. (1992). Imagination, cognition, and persona. Rhetoric
Review, 10, 266-278.
Sadoski, M., & Paivio, A. (1994). A Dual Coding view of imagery
and verbal processes in reading comprehension. pp.582-601 in R.B. Ruddell, M.R.
Ruddell, H. Singer (eds). Theoretical Models and Processes of Reading
(4th ed.) Newark, DE: International Reading Association.
Snowling, M., Van Wagtendonk, B., and Stafford, C. (1988). Object naming
deficits in developmental dJyslexia. Journal of Research in Reading,
Staats, A.W. (1961). Verbal habit-families, concepts and the operant
conditioning of word classes. Psychological Review, 68, 190-204.
Te Linde, J (1982) Picture and words in semantic decision. In John Yuille
(Ed.). Imagery, memory and cognition: Essays in honor of Allan Paivio.
Thompson, V. A. and Paivio, A. (1994). Memory for pictures and sounds:
Independence of auditory and visual sounds. Canadian Journal of Experimental
Psychology, 48, 380-396.
Van der Wissel, A. (1988). Hampered production of words as characteristic
of school failure. Journal of Learning Disabilities, 21, 517-518.
Weinstein, C. E., Underwood, V. L., Wicker, F. W., and
E. (1979). Cognitive learning strategies: Verbal and imaginal elaboration.
In O’Neill, H. F., and Spielberger, C. D. (eds.), Cognitive and Affective
Learning Strategies, New York: Academic Press, 45-75.
Weschler, D. (1974). Manual for the Weschler Intelligence Scale for
Children-Revised, New York: The Psychological Corporation.
Wittrock, M. C., and Alesandrini, K. (1990). Generation of summaries
and analogies and analytic and holistic abilities. American Educational
Research Journal, 27, 489-502.
Zimler, J. Keenn, J. M (1983) Imagery in the congrenitally blind: how
visual are visual images? Journal of Experimental Psychology: Learning,
Memory, and Cognition, 9, 260-282.