You are on page 1of 25

IMAGINATION, COGNITION AND PERSONALITY, Vol.

30(4) 363-387, 2010-2011

IMAGE OF A CONCEPT: THE ABSTRACT VISUAL


IMAGE AND ITS RELATION TO THE GENERAL
AND THE SPECIFIC IMAGE

NAFTALI ROSENFELD
SHLOMO KANIEL
Bar Ilan University, Israel

ABSTRACT

Visual imagery has been examined over the past few years with focus on
general visual images in contrast to specific visual images. A general visual
image is the most primitive image, a prototype of an image that represents
characteristics of an object. In comparison, specific visual images are
attributed to one of the categories of the prototype and they are richer in
detail. In this article we propose a theoretical and experimental extension of
these two categories into a third category defined as: “abstract visual image.”
This is an image created when subjects form a visual image of symbolic
pictures of abstract concepts. The study analyzed the characteristics of these
three visual images and their efficiency of transfer of learning physics. The
findings show: 1) specific visual imagery was employed with a noticeable
low frequency in comparison to general and abstract imageries which were
engaged with a high frequency; 2) general and abstract imagery were jointly
employed by means of a mental process which can be characterized by four
styles of a combination of concrete and abstract images; and 3) intensive
employment of abstract imagery leads to high study achievements in contrast
to intensive employment of general visual imagery which involves more
concrete images. Nonetheless, if general visual imagery is combined with
intensive abstract imagery, the joint processing leads to high study achieve-
ments. The findings are explained by a theory that abstract imagery reduces
cognitive load and leads to more efficient learning.

363

Ó 2011, Baywood Publishing Co., Inc.


doi: 10.2190/IC.30.4.c
http://baywood.com
364 / ROSENFELD AND KANIEL

INTRODUCTION

Visual imagery is the ability to “see with the mind’s eyes” anything, even if
the object is not tangibly placed in front of us [1, 2]. “Working memory” creates
an internal picture by which the mind can observe it and survey each detail,
as well as alter it by rotating, enlarging, moving and discovering associations
between objects.
A “visual imagery” (defined also “mental imagery”) picture represents the
object but is not completely identical to it. The representation preserves com-
plete analogy of the spatial relations between parts of the object in reality but
details are always missing. This aspect of visual imagery has been investigated
[3, 4] based on the concept of a general visual image and a specific visual
image. A general visual image is the most primitive and most immediate image.
It is the prototype or scheme of the skeletal image which represents the
fixed characteristics of the image. In contrast, the specific visual image is
attributed to one of the object’s categories. It has all of the fixed character-
istics of the general object and it has, in addition, the details characterizing its
specificity and it is richer in detail. For instance, a general visual image is created
when an individual is asked to imagine the image of an “automobile.” The
subjects in this case will state that they pictured a general schematic image
of an automobile chassis with four wheels. A specific image will come to
mind if the subject is asked to envision “his neighbor’s automobile.” In this
case the image will include a picture richer in details that characterize the
specific automobile.
The distinction between concepts expressing a specific concrete object and
a more generally abstract one was experimentally supported by tests conducted
by Rosch [5, 6] about four decades ago. Arnheim [2, 7] takes the same approach
and suggests that thinking can deal with directly perceived objects which
often are handled physically. When no objects are present they are replaced
by some sort of imagery, concrete (pictures) and abstract (symbols, signs).
He suggests that these pictures, symbols, and signs can be scaled from low
abstraction (replicas) to medium abstraction (stylized objects) and complete
abstraction (non-mimetic).
Translating this approach we can find here that a “specific visual image” means
“replicas”—a more concrete object. “General visual image” means “stylized
objects” and “non-mimetic”—more abstract.
Most of Arnheim’s ideas are rooted in concepts of art. Rosch also did not
refer to visual images, but to general categories of concepts expressing objects.
Much evidence has accumulated in the field of visual imagery supporting
the distinction between general visual images and specific visual images. This
substantiation includes evidence such as the time needed to generate visual
images [8] and neurological evidence such as the mapping of response regions
in the brain by means of FMRI [9].
ABSTRACT VISUAL IMAGE / 365

ABSTRACT VISUAL IMAGERY

In this article we propose to extend the theoretical and experimental concept


of the general visual image and the specific visual image to a third category
defined as the “the visual imagery of the abstract.” This is the image created
when subjects generate a picture of a symbol expressing an abstract concept
in their visual imagination. For example, an individual may analyze the abstract
concept of “speed of a car” either by non-visual mental procedure or by visual
imagery. Using visual imagery he would create in his mind a vision which is
the image or symbol of this abstract concept. Possible images for “the speed
of a car” could be a fragmented line expressing movement, a picture of a vector
arrow symbolizing the magnitude of speed and its direction, a mathematical
equation like v = x/t, or a graph depicting movement on frames of reference.
In contrast to general or specific imagery of an object, these abstract images:

1. do not stem from an analogy to the figure of an object; instead, their meaning
is derived from a definition;
2. are the generalization or abstraction of a concept and not a simple meaning
of an object;
3. are not used in everyday language but are derived from a personal or
accepted definition.

With abstract images, an individual indeed envisions an “object” in his


“working memory,” but it is a mental symbol or model of an abstract concept
and it does not represent a concrete object. In the framework of this study
these symbols and images are defined as “pictures, symbols or images of abstract
visual imagery.”
The third category proposed (i.e., the imagery of an abstract symbol) extends
the model of three levels which Rosch [5, 6] developed for the categories of
objects. It also extends the model of three levels of Arnheim from art to visual
imagery, constituting a practical clear definition. Based on Rosch and Arnheim
models, we propose to define a sequence of three levels for visual imagery that
move on the concrete—abstract dimension:

1. The subordinate level—specific imagery of a certain concrete object with


many details unique to the specific object.
2. The basic level—general imagery of a concrete partially abstract object. The
vision, lacking details, is a prototype of the concrete object with general
minimal descriptive lines by which the subject still identifies the form of
the object.
3. The super-ordinate level—the abstract imagery of a symbol. Images at this
level do not at all represent an analogy of the figure of an object but rather
a symbol and a model with association to abstract concepts.
366 / ROSENFELD AND KANIEL

According to this proposal, the degree of abstractness of the visual image


is represented in Figure 1 as a sequence of three levels moving along the
concrete-abstract dimension.
A person is able to create images on a dimensional sequence expressing the
degree of concreteness or abstractness. Three main points can be distinguished
in this sequence: a) on the one end of the sequence is a specific concrete image of
an object which is the subordinate level; b) in the center is the general image
constituting the basic level; and c) on the other end is the abstract image—a picture
of a symbol, constituting the super-ordinate level. The closer the image is to the
specific concrete side of the sequence, the greater the literal meaning emerging
from the image. It stems from the exact analogy of the form of the object. In
contrast, the closer the image is to the abstract end of the sequence, the person
indeed visions an image in the form of an object on the visual board, but the image
represents and symbolizes something abstract which is not the simple meaning
of the object, and does not stem from an analogy of the form of the object.
Throughout this article, the basic level—the general imagery—will be studied
as a concrete level relative to the super-ordinate level which is completely
abstract visual imagery.

Visual Imagery in Teaching Sciences and Physics


In the field of visual imagery, the tools that have been developed usually
measure imagery generated by means of contents of situations of daily life.

The Concrete-Abstract Continuum of Visual Imagery

Specific imagery General imagery Abstract imagery


[Subordinate] Basic level [Super-ordinate]
Low level High level

Concrete imagery Partially abstract Completely abstract


imagery a prototype imagery

Exact analogy of the General analogy of the No analogy of the


form form form

Many details of the Few details of the No details of the


object object object

Figure 1. The concrete-abstract sequence of visual imagery.


ABSTRACT VISUAL IMAGE / 367

However, in science teaching, visual images are also examined by measuring


parameters of success in completing studying assignments. Exercising visual
imagery in the field of mathematics has been found to support all thinking
processes, to improve processes of solving all types of mathematical problems
[10, 11] and to be supportive in solving physics problems [12, 13].
The distinction between general imagery and specific imagery which has
been presented hitherto is similar to those proposed in the field of teaching,
with semantic differences. Kozhevnikov et al. [14] distinguish between pictorial
imagery, which is imagery that expresses the external form of an object, and
schematic imagery, which conveys the spatial relations between objects and is
more general. They discovered that different types of people use different types
of these imageries [11] and they found a positive correlation between the use
of schematic imagery and solving mathematical problems.
Presmeg [15-17] and Owens and Clements [10] categorize images created
when solving mathematical problems. They distinguish between concrete pic-
torial images—images representing analogies of an object, and pattern imagery—
images of a general format representing abstract figures lacking concrete details.
The suggested concept of “abstract visual imagery” in teaching sciences and
physics offers a practical foundation for Zimmerman and Cunningham’s [18]
approach in mathematics. They claim that when creating a visual image in
mathematics, a different visual imagery is used in contrast to the one used on a
daily basis. A subject uses visual imagery in mathematics to visualize a concept
and not an object. The person envisions an object in his mind’s eyes, but the
interpretation accompanying this picture is associated with the abstract concept.
In this work, we measured the characteristics of general imagery and images
of abstract imagery in the field of studying physics. The students’ images were
measured in the process of solving physics problems and were compared to
their achievements in acquisition and transfer of the studied material.

Acquisition and Transfer

Three levels of efficient learning can be distinguished. The first level is


acquisition, where the student performs the same task he learned and practices
retrieving it from his memory. The second and third levels are the transfer stages.
The student transfers the principles and the notions of the task studied (source
task) to new tasks which have not yet been studied (target tasks). Transfer com-
prises two dimensions [19]: a) near transfer which occurs when the student
applies what he learned to examples which stem directly from the material
studied; and b) far transfer which occurs when the student develops new ideas
and applies what he learned from the source tasks to target tasks that do not
directly stem from the material studied. The current study measures the acquisition
and transfer in studies while examining the cognitive load created when the
368 / ROSENFELD AND KANIEL

student utilizes his visual imagery during the process of solving study assign-
ments in physics.

Cognitive Load
Cognitive load is when the “working memory” attempts to simultaneously
process large amounts of knowledge beyond its capability and limited sources.
This situation results in slowdown of processing procedures. The load can relate
to the entire network of the working memory. This study focuses mainly on
the load stemming from visual imagery activities. Two situations cause loads
in studying [20, 21]: a) the amount of material studied—there is insufficient
resources to simultaneously process the large quantity of information details
in an efficient manner; b) the contents of the information studied. Three types
of content loads are eminent:
1. Intrinsic cognitive load—the difficulty of the complexity of the assign-
ment, stemming from the presence of distracting contents or contents with
complex interaction between them. The more the associations between
the ideas, the more the resources needed to simultaneously process the
ideas. In this case the assignment is considered more complex and diffi-
cult to process due to the emerging load.
2. Extraneous cognitive load—of an inefficient teaching method.
3. Germane load—activities that do not constitute a load since they directly
support the processing procedures [22].
Two main study principles prevent load [21, 23]:
1. Organization of the information—efficient organization of the information
studied in the form of several small schemes and assimilation of former
knowledge as one efficient scheme for retrieval. One implementation of
this principle is abstract visual imagery which represents a large amount
of information as one symbol, for example, in a mathematical formula.
2. Automation—large amount of practicing enables automated response
without any need of awareness or attention. This enables the working
memory to act without the need of resources, which can be directed to
more conscious processing. This principle was not examined in this study.

Research Hypothesis
The main hypothesis of the current study is that the more the physics student
uses visual imagery at a higher level, the more the working memory load will
be reduced, resulting in more efficient studying in both acquisition and transfer.
This hypothesis is based on the following arguments:
a. Abstract symbols prevent cognitive load and improve studies: the amount of
details decreases as the level of processing of the visual imagery increases
ABSTRACT VISUAL IMAGE / 369

from concrete to abstract [1, 4], thus the load is smaller. The abstract images
lack details and they are simple, immediate, and it is reasonable to assume
that the time it takes to create them is shorter than general imagery pic-
tures, just as generation of the latter takes less time than creation of specific
imagery pictures.
For example, in the first assignment [no. 1] in the current study, the
subject was asked to solve a physics problem of motion influenced by
gravity: “a person is standing on the roof of a house holding two balls at
the same level. One ball he throws at any speed horizontally to the ground,
and at the same time he lets the other ball fall freely downward without
any initial speed. Explain which ball will be the first to hit the ground”.
There are possibilities of representing the visual image during the process
of solving this problem, based on the three levels of the concrete-abstract
sequence: a) the lowest level which is specific imagery—the subject can
create an image of a specific concrete object. For instance, the subject could
depict the teacher standing on the lab roof throwing balls. The time needed
to create this vision is very long, due to the vast details characterizing
the specific example of the image of the teacher and the lab building, thus
causing high cognitive load during the solution process. b) The basic level
which is general imagery (i.e., general images of concrete objects) repre-
sented by a picture lacking many details but maintaining the general form
of the object. For example, a subject may create a general vision of any
person’s hand, not anyone in particular, throwing balls from the roof of
an unidentified general building. This picture is less concrete and includes
fewer details than the earlier example from the lower level of imagery. It
would take less time, thus causing less cognitive load during the solution
process. c) A high level which is abstract imagery—an abstract image of
a symbol whose meaning does not stem only from the representation
of a certain concrete object. For instance, the subject creates an abstract
image—a symbol of a “vector,” a picture of an arrow which represents
the speed and its direction. The arrow looks as though it is progressing
in accordance with the motion equations. The creation of this image is
more economical due to its simpler form, thus the cognitive load will be
smaller and the activity of the working memory will be more efficient.
Visual processing is generated together with non-visual processes in
the “working memory” (dual code theory). The load created during proc-
essing is the result of the load stems from all components of “working
memory.” Less load of the visual processing means less load on “work-
ing memory” resulting in more efficient working memory.
The study assignments in sciences and physics in the upper high school
classes is complex. The complexity of these assignments stem from contents
which include intricate interactions causing a load that affects processing
efficiency [20]. These assignments require saving resources. The efficient
370 / ROSENFELD AND KANIEL

solver regulates the use of visual imagery levels according to the load
created during the processing procedure, so that the entire load required
for solving the assignment will decrease. The more complex the assignment,
the higher the level of visual imagery the efficient solver will utilize. The
achievements of students utilizing general imagery will be higher than
those applying specific imagery and activation of abstract imagery will
result in the highest achievements.
b. Abstract symbols promote transfer: The study shows that in transfer any
strategy which reduces the cognitive load contributes to the transfer [19, 21,
22, 24-26]. Consequently the more the scholar utilizes the higher level of
visual imagery, the more efficient his studies will be. The achievements
of general imagery will be higher than specific imagery and the use of
abstract imagery will result in the highest achievements.

METHOD

Subjects
Fifty-eight 17-year-old students from 11th grade, 41 males and 17 females,
participated in the study. They majored in science on their matriculation exams.
The students voluntarily cooperated throughout all stages of the study and they
received no retribution for their participation. The study was integrated into
the regular matriculation exam study program for high level students. Each of the
subjects chose to study science of their own free will and their achievements in
the previous years in mathematics and natural sciences were above 80%.

Tools
Three tools were used:
1. an intervention program;
2. a tool to categorize the imagery visions into the different categories based
on the level of imagination; and
3. a tool to measure study achievements.

Intervention Program: Studying a Chapter in Physics


in the Classroom Setting

The students studied the subject “motion of objects influenced by the earth’s
gravity.” This chapter is part of the regular curriculum of 11th graders in the
high level of science matriculation study program. The topic is studied for a
period of 6 weeks about 6 hours a week, totaling 36 hours. Following their
studies, the subjects were intensively interviewed by means of research tools
on the contents of the chapter. There was no change in the form of the method of
ABSTRACT VISUAL IMAGE / 371

teaching from previous years. As every year the teaching program included
a variety of methods such as activities in the classroom and at home, lab work,
and computer activities in school and at home, etc.

Tool to Categorize Images into Different Categories


Based on the Imagery Level

This tool was designated to categorize the images during the process of solving
the assignment in physics based on the three levels of the concrete-abstract
sequence. It was built based on the dimensions, which are well established in
the literature on the field of visual images:
a. Vividness of the image—This dimension measures the amount of infor-
mation of the image, i.e., how similar the image is to the actual object.
Cornoldi et al. [27, 28] identified several variables in the vividness concept:
richness of details, rich association, eminent characteristics, contour and
form, color, general imagery versus specific imagery.
b. Visual preference—This dimension measures the extent the subject spon-
taneously uses visual imagery in solving the assignment in which the
solution allows use of verbal imagery as well as visual imagery. An
example would be the tendency to envision a visual image when reading
the contents of the physics assignment. This dimension was not examined
in this work. Nonetheless, most of the physics contents studied at high
schools cause rich spontaneous use of visual imagery [29]. All the subjects
in this study reported the spontaneous creation of rich images which
enabled clear measurement of the processes.
c. Controllability—This dimension measures the extent the subject can move
the pictorial images and rotate them according to the accuracy required.
Controllability is also a comprehensive name for variables and different
types of visual use: a) moving a picture in contrast to a fixed picture [30];
b) complexity in comparison to simplicity, i.e. a simple rotation in contrast
to a complex one; and c) manipulation of familiar objects in contrast to
unfamiliar objects [31, 32].
The tool was built according to the think aloud method and managed according
to the protocol analysis [33-35]. The subject solved the problems presented to
him while verbalizing his thinking process. After solving the assignment the
subject was asked to reconstruct the imagery visions created during the thinking
process. The tool categorized the images produced by the subjects, as reported
in the protocols in four stages:
1. Initial categorization based on the variable proposed—the imagery levels,
concrete versus abstract: a) low level—specific imagery; b) basic level—
general imagery; c) high level—abstract imagery.
2. Secondary categorization based on the vividness of the image variable;
372 / ROSENFELD AND KANIEL

3. Secondary categorization based on the controllability of the image variable;


4. Categorization of the use of the levels based on the four “styles” proposed
in this study. The details and definitions of the “styles” are presented below.

The stages and the criteria for categorization are presented according to the
three variables by means of an analysis of example no. 1 discussed above: “a
person is standing on the roof of a building holding two balls at the same level.
He throws one ball at any speed horizontally to the earth and at the same time
he lets the other ball fall freely downward without any initial speed. Which
ball will hit the ground first? Explain.”

Stage 1: Initial categorization based on the imagery level—“concrete vs.


abstract variable”—In the first stage of processing the results, the imagery
pictures were categorized into three levels based on the central variable the tool was
measuring, which was defined as the “imagery level–concrete versus abstract.”
This categorization answers the question whether the imagery is specific concrete,
general, or abstract imagery. The principles of the categorization of the tool into
levels are depicted by means of three responses of the subjects to the assignments:
• The specific concrete [low] level—The criteria for categorizing the low level
is the presence of imagery of a specific image, like the image reported by
Abraham: “I saw the teacher, standing on the roof of the lab and throwing the
balls, similar to the experiment I remembered he did in class”. He imagined a
specific image and event experienced in the past and retrieved it from his
memory in order to solve the assignment. The picture is analogical in its form
to a specific object (i.e., the clear image of the teacher and the roof of the lab
building). The picture contains many details unique to the object and does
not include abstract details.
• The general [basic] level—The general image represents images with partial
abstract aspects. The subjects accurately identify objects of general form
with fuzzy or missing details, a prototype that represents the entire group
of objects analogically similar to it in the general form. For example, Hana
reported a category matching the basic level in her account: “I saw a general
picture of a person’s hand made of something like sticks, throwing the balls
from the roof that looked like a square of an unidentifiable square building.
The balls flew one at an angle and one downward.”
• The abstract [high] level—The abstract imagery represents images of
abstract symbols. The basic level, called “general imagery,” is more abstract
than the low specific imagery level. However, relative to the high level
of imagery, the basic level represents more concrete objects. Indeed, the
representation of the basic level is partially abstract due to the missing
details, but the representation of the high level is completely abstract and
there is no analogy between the image and the object. At this high level the
subjects combined a variety of symbols in the image: figureless points or lines
ABSTRACT VISUAL IMAGE / 373

representing the objects, vectors, equations and different symbols that the
subjects invent. Nir, in his account, stated: “I saw a point moving along the
parabolic line, vector y in motion on the path to the peak of the height and
then equal to zero; vector x did not change.” The points represent the moving
object in an abstract manner, while the vector represents an abstract meaning
of motion of the stone. Consequently, this student’s account is categorized
as high level imagery.
Each of the levels was measured on a scale of 1 to 5. The lower end of the
continuum—1 represents the “level which was not used at all in the imagery
picture,” the higher end of the continuum, 5 represents the “level intensively
utilized in the imagery picture.” The scoring of the assignment relative to the
median conclusively determines the level used in the assignment.
After determining the visual imagery level used, the next two stages involved
measuring the magnitude of their activity. The “magnitude activity” was defined
as a measure summing up the intensity of the imagery activities based on two
visual imagery dimensions: vividness and controllability.
Stage 2: Secondary Categorization According to Variable “Image Vivid-
ness”—With the current tool this dimension was measured by variables of
“prominent characteristics, richness of details,” which measure the variation in
the richness of all the objects and abstract symbols described during the solution
of the problem. In contrast, the other variables in the image vividness dimension
[27], “color, contour, and form,” are not present in these solutions due to the
scarcity of images which appear as schemes. The “image vividness” variable was
measured on a continuum scale of 1-5. The lower scale of the continuum,
1 represents “no vividness in the visual imagery,” while the higher end of the
scale, 5 represents “very much vividness in the visual imagery.”
Stage 3: Secondary categorization according to the variable “image con-
trollability”—The “controllability” variable was measured according to the
degree of motion of the objects described in the image and was presented on a
scale of 1-5, with the lowest level, 1 representing “no controllability of the
image” and 5 representing “extensive controllability of the image.”
The sum of the results of the last two stages of variables (i.e., vividness and
controllability), represent the total “magnitude activity” of the visual imagery for
each task. This measure, together with the results of the first stage, which
determined the level of imagery, enables us to determine the final result of the
measuring tool:
1. the level of activity when solving the task; and
2. the magnitude of activity applied.
An analysis of the findings (as detailed below) demonstrated that a low level
of imagery was applied at a very low frequency and the basic and high levels
374 / ROSENFELD AND KANIEL

of imagery were applied at very high frequencies. The low level was applied
at a frequency of 3.3% in contrast to 83.5% of the basic level and 78.2% of
the high level.
This finding supports the theory and is explained by the character of the tasks
of the experiment and their complexity, which dictated the use of the basic and
high levels of imagery. These two levels create a lower load than the load pro-
duced with the low level imagery. Their application alone results in more efficient
working memory activity. Based on these findings, a further analysis of the
results was conducted according to two instead of three levels: the basic level—
general imagery, and the high level—abstract imagery, without the low level.
Another finding shows that solving the task in physics is a long mental process
in which the subject generates several images, combining the basic level, which
is relatively concrete in comparison to the high level, with the high level that is
completely abstract. Thus, this study proposes to organize the coactivity of the
two levels in four “imagery activity styles” to provide insight into the combina-
tion of the concrete and abstract levels.

Definition of “four imagery activity styles” of the two imagery levels: Since
the possible combinations of the two levels are four, we propose to define the
four styles as “imagery activity style” which will clarify the imagery activity
from the combination of the two levels. Furthermore, proposing the term “style”
contributed to a higher sensitivity level of the methodology for this tool. Instead
of discussing two points of measurement on the concrete-abstract continuum,
the basic and high levels, each level was divided into two sublevels. This created
four points on the continuum, enabling measurement of higher sensitivity than
only two points. An in-depth examination of the subjects’ answers depict that
these four points correspond to four styles that move along the continuum and
upon which the level of imagery is applied (see the summary presented in Table 1).
When solving the physics problem, the subject generates the basic level and
the high level imagery jointly. The subject can generate each of the levels at a

Table 1. Division of the Two Imagery Levels into Four Possible Styles,
from Rich to Poor for Each Level

High level activity Basic level activity Activity Style


magnitude magnitude definition
[Abstractness] [Concreteness]

Poor abstractness Poor concreteness Poor style


Poor abstractness Rich concreteness Basic style
Rich abstractness Rich concreteness Combined style
Rich abstractness Poor concreteness High style
ABSTRACT VISUAL IMAGE / 375

high magnitude (defined below also as “rich value”) or a weak magnitude


(defined below also as “poor value”). In this manner, each level was divided
into two sub-levels of the same level: a group that was given a “poor value” and a
group that was given a “rich value” of the same level. These proposed four
sub-levels defined the “four imagery activity styles”: High style, Combined
style, Basic style, Poor style.
Alpha coefficient: The alpha coefficient was measured for each of the imagery
levels, over all the tasks included in the same level. The basic level was 0.85
and the high level was 0.81.
Reliability of the judges: Three judges, physic teachers who have been pre-
paring students for their matriculation exams for the past 5 years and who
specialize in solving tasks of the type included in this study, examined the
elements of the tools. They separately analyzed the protocols of 12 subjects
according to the criteria and the scoring systems presented and expressed their
consensus with the analysis results of 93.1% of all the tasks of all the subjects.

Tool to Measure Study Achievements

Study achievements were measured by the tasks the students solved in physics
which included six tasks solved with pen and paper and five by verbal protocol
analysis. The achievements measured included efficiency of learning based on
the following details and examples.
Acquisition
Example: “A stone is thrown from the roof of a building, vertically upward,
beginning with a speed of 40 meters per second. What distance will the stone be
from the point it was thrown, after 9 seconds?”
The rationale for categorizing the acquisition example: The task can be imme-
diately solved through spontaneous reconstruction of memory of an identical
example done in class and in homework.
Near transfer
Example: “Two stones are dropped from the top of the roof vertically down-
ward at the same time. One stone is let go downward at a speed of 0. At the same
time the second stone is thrown downward at an initial speed of 30 meters per
second. Find the equation of motion for the two stones (let V be a function of T).”
The rationale for categorizing the example as a near transfer: The subject
reaches the solution after synthesizing with a similar topic practiced well—
equations of vertical motion. The identical exercise was not practiced, thus the
task is not considered to be known through acquisition but rather through near
376 / ROSENFELD AND KANIEL

transfer. Though the student did not solve this exact task, he should solve it
spontaneously due to practicing similar skills.

Far transfer

Example: “A shot was fired at an initial speed of 200 meters per second and
at an angle of 30° above the horizon. The shot hit the top of the mountain at a
height of 400 meters. If you are asked to sketch a graph of the speed of the
bullet as a function of the time (let V be a function of T), argue mathematically
whether the graph will be a linear line or a nonlinear graph.

The rationale for categorizing the example as a far transfer: The subject must
reach new generalizations from the source, extract them from the original context
and transfer them to a task he has not yet practiced. The mental process of
solving this task is not immediate and is not part of the skills the subject has
yet acquired. Furthermore, the task is presented as an open question that does
not need to lead to a numerical solution and it is not the routine method of study
to which the subject has become accustomed.

Additional measuring characteristics of the tool

Scoring: The solutions of the subject for each task were measured according
to a set of criteria customary for matriculation exams in physics with the total
possible score for near transfer ranging between 1 and 26, processed into per-
centages between 1 and 100, with the highest value indicating better learning
and transfer abilities.

Alpha coefficient: For acquisition the alpha coefficient was 0.68, for near
transfer it was 0.73 and for far transfer it was 0.54.

The reliability of the judges: The same three judges separately analyzed the
answers of 12 subjects and expressed their consensus with the analysis results
of 93.1% of all the tasks of all the subjects.

Procedures

The study was conducted according to the following stages:

1. Intervention program: studying the topic of “the influence of gravity on


motion” in the framework of the regular curriculum in school.
2. Achievement test with pen and paper that checks study achievements.
3. Interview for thought protocol: solution of tasks by means of “thinking
aloud” and management of a protocol which measures: a) visual imagery
categorization; b) additional study achievements.
ABSTRACT VISUAL IMAGE / 377

RESULTS

Finding 1: The Frequency of the Low Level


in Contrast to the Basic and High Levels

The analysis of the protocols shows three levels, but the specific image
appeared significantly less frequently than the general and abstract imagery.

Method of Analysis

For each question the subject could apply three levels of imagery, and each
level was generated independently of the other. The calculation of the frequency
of each level was done by counting the number of times the subject generated
each of the levels, when performing each of the tasks of the experiment. The
calculation did not relate to a specific subject, and each question was measured
as a separate test. The maximum number of possible answers for each level was
522 (9*58 = 522, 9 tasks times the number of subjects, total 522 different tasks
in one basket were performed). The calculated sum of activations of a level
represents the frequency of that level. The relative frequency in percentage for
each level is the frequency multiplied by 100 divided by 522, since each level
was generated independent of the other. The results are presented in Table 2.
The results show that the subjects generate the basic and high levels more
than 78% of the time when they were asked to solve a task.
The chi square test was performed to distinguish between the three groups.
In the chi square test the observed value is presumed as equal to activating each
level with a theoretical probability of 50%, because there are two theoretical
possibilities for each level (i.e., generated or not generated). The chi square test
resulted in 82.0, p < 0.000. Figure 2 depicts the results in graphic form.
The findings show that the low level (specific image) was generated signifi-
cantly less in comparison to the basic level (general image) and the high level
(abstract image) which were generated more frequently.

Table 2. Frequency of Activation of the Imagery Level

Frequency Maximum frequency Activation frequency


of level possible in percentage

Low level 17 522 3.26


Basic level 436 522 83.52
High level 408 522 78.16
378 / ROSENFELD AND KANIEL

Figure 2. Imagery level generation frequency in percent.

Finding 2: Style Achievements for all Measures


Table 3 presents the mean and standard deviations used for MANOVA
(Multivariate Analysis of Variance) with repeated measures in the form of 3 × 4
(three measures of learning efficiency: acquisition, near and far transfer, and four
different styles).
Significant differences were found between the three levels of study achieve-
ments (F(2, 108) = 125.55, p < 0.000, eta^2 = 0.70). The style factor was also
found to be significant [F(3, 54) = 22.68, p < 0.000, eta^2 = 0.56]. Slightly
significant interaction was found between study achievements and the style
(F(6, 108) = 2.15, p < 0.053, eta^2 = 0.11). Scheffe post hoc analysis indicates
significant differences between the “high and combined” group of styles whose
achievements are better than the “basic and poor” group styles.
These findings demonstrate that the high style and the combined style
achievements are higher than the basic and poor style achievements in reference
to all three achievement measures: acquisition, near transfer, and far transfer.
Slightly significant interaction was found between study achievements and
the style. There are no differences between high style and combined style
ABSTRACT VISUAL IMAGE / 379

Table 3. Mean Percent and Standard Deviation (in Brackets)


of Study Achievements in the Four Different Styles
of Imagery Activation

Style

Poor Basic Combined High


Study achievements N = 10 N = 18 N = 13 N = 17

Acquisition 75.4 76.0 98.0 99.4


[20.9] [17.1] [3.8] [1.5]

Near transfer 67.8 54.5 87.6 93.3


[22.1] [22.3] [14.8] [10.9]

Far transfer 43.1 29.9 63.5 71.1


[16.3] [15.4] [24.7] [19.5]

achievements and there are also no differences between basic style and poor
style achievements.
Figure 3 shows a diagram of achievements of four styles According to acqui-
sition and transfer.

DISCUSSION

The findings of the study show:

1. When activating visual imagery to solve tasks in physics three imagery


categories are revealed, specific imagery (low level), general imagery (basic
level), and abstract imagery (high level). The low level was generated
at a very low frequency while the basic and high levels were generated at
high frequencies.
2. Long and complex mental activity performed by subjects such as solving
physics tasks results in the generation of two imagery levels—basic
and high and the generation of combined imagery characterized by
concept of style.
3. The high style and combined style achievements are better than the
basic and poor styles in reference to all three measures: acquisition
and near and far transfer. Slightly significant interaction was found
between study achievements and the style. There is no difference between
the high and combined style achievements or between the poor and
basic styles.
380 / ROSENFELD AND KANIEL

Figure 3. Graphic illustration of the acquisition and transfer


achievements according to the different styles.

Transition from Three Levels to Four Styles

The results of this study show that the efficient solver controls the use of
visual imagery levels so that the total load during the process is small and he
uses two processing strategies:

1. Reduce the internal load by processing through the basic and high levels
of visual imagery. The low level utilizes a greater amount of memory
resources [3], thus the solver will avoid activating the low level for complex
tasks such as physics, as indicated in the study findings.
2. Organize the knowledge for separate stages without interaction between the
knowledge elements and process each stage separately [21], which dictates
integration of the basic and high levels—the concrete and the abstract.

Based on the two strategies above, we proposed to reorganize the two active
levels, the high and basic levels, into four styles that thoroughly explain inte-
gration of the concrete and abstract levels. Each level was divided into two
sub-levels, based on the criteria of rich activity in contrast to poor activity of
each of the levels. The division created four groups termed “imagery activity
ABSTRACT VISUAL IMAGE / 381

styles” which move along a concrete-abstract continuum. Each individual com-


bines the two levels—the high and basic levels—when solving the task, but there
are differences in their activation and integration. The differences stem from
a different magnitude of activity for each level—rich in contrast to poor—and
the possible combinations of the integration between them. In this manner,
the four styles characterize the visual imagery activities with more clarity than
characterization based on only two levels.
In order to understand the term “style,” a demonstration of the different styles
is again presented by means of response to another task (no. 3) given to the
subjects in this study: “a child throws a tomato at a certain angle toward the
horizon between 0 to 90 degrees. What type of path of movement will the tomato
follow? Explain and provide arguments.” The four styles detailed below by the
images generated by the subjects when solving this problem.

1. Poor Style: A subject who generated the abstract (high level) in a poor
manner combined with the concrete (the basic level) in a poor manner was
defined as having a poor imagery activity style. For instance, a subject
reported: “I saw the tomato moving along a line.” It was categorized as poor
style because he saw only one concrete object of the data (i.e., the tomato
(poor basic level)). The line he drew in his imagination is an abstract symbol
that represents the movement. In the process of solving this problem, a
wide use of abstract symbols is possible like vectors, graphs. The subject
used only one poor symbol which indicates limited information of the
movement (poor high level).
2. Basic Style: A subject who generated the abstract (high level) imagery in
a poor manner combined with the concrete (basic level) imagery in a rich
manner, was defined as having a basic imagery activity style. For instance,
subject stated “I saw a child standing on the ground, holding a tomato,
throw the tomato, it moved along a line.” It was categorized as basic style
because the subject imaged all the given objects in his mind and added the
ground to it, a detail which was not given (rich basic level). Like example 1,
the subject saw poor abstract symbol of the movement (poor high level).
3. Combined Style: A subject who generated the abstract (high level) in a rich
manner combined with the concrete (basic level) in a rich manner was
defined as having a combined imagery activity style. For example: “I saw
a child standing on the ground, holding a tomato, threw the tomato, it
flew throughout the entire path. Vector y is large at the beginning of the
movement and then it gets smaller until the highest point where it becomes
0. From this point the vector inverts and grows, vector x remains the
same.” It was categorized as combined style because the subject imagined
all the given objects and added the ground, a detail that was not given
(rich basic level). He saw rich activity of vector symbols throughout the
entire path (rich high level).
382 / ROSENFELD AND KANIEL

4. High Style: A subject who generated abstract (high level) imagery in a rich
manner combined with concrete (basic level) imagery in a poor manner
was defined as having a high imagery activity style. For example a subject
reported “I saw a point move along a path on a line, and vector y is large
at the beginning of the movement, becomes smaller until the peak height
where it becomes 0. From that point the vector inverts and grows and
vector x remains the same.” The explanation for categorization in this
case is that the subject did not see any object from the given data (poor
basic level), but used symbols to present the entire task. He saw a point
as a symbol of tomatoes and the symbols of the vector along the entire
movement time (rich high level).

New Scope Over the Concrete-Abstract Continuum


One of the main contributions of this research is that it provides a new
and sensitive scale which is valid and reliable for measuring the levels of
abstraction of visual images used by individuals to solve physics problems.
Five levels of visual imagery on the concrete-abstraction continuum can now
be presented:
1. specific imagery, concrete analogy—a vivid level, rarely used by participants;
2. poor style, low level of abstraction and concreteness;
3. basic style, mixed level with greater emphasis on prototype concreteness;
4. combined style, mixed level with equal emphasis on abstract and con-
creteness; and
5. high style, greater emphasis on symbolic-abstract level.
Figure 4 depicts this new scale on the concrete-abstract continuum presented
in the introduction.

Achievements Based on Styles and the


Principles Behind Them
The findings show that the achievements of high and combined styles are
higher than the achievements of basic and poor styles. The findings are explained
by the concept of the visual imagery presented in this study and the principles
behind them:
• Intensive generation of abstract images (high style) improves processing
and study achievements. The abstract symbols represent a large amount of
knowledge and their generation as one visual scheme saves resources and
improves the processing both in acquisition and transfer.
• Intensive generation of concrete images (basic style) affects achievements.
It creates a load in the working memory due to the numerous details appear-
ing in the images. However, if the concrete images are accompanied by
ABSTRACT VISUAL IMAGE / 383

Imagery Activity on
The Concrete-Abstract Continuum of Visual Imagery

Specific General Imagery Abstract Imagery


Imagery Basic Level High Level
Low Level
[generated at a
low frequency]

Poor Style Basic Style Combined Style High Style


poor basic level rich basic level rich high level rich high level
poor concreteness rich concreteness rich abstraction rich abstraction
poor high level poor high level rich basic level poor basic level
poor abstraction poor abstraction rich concreteness poor concreteness

Poor abstract activity Rich abstract activity

Figure 4. Imagery activity on the concrete-abstract continuum.

intensive generation of abstract imagery (combined style), the joint


processing yields high study achievements. The combination with abstract
imagery compensates for the load caused by concrete imagery according
to the principle presented.
These principles explain the findings for each style below.

The High Style

The subject intensively generates high level abstract symbols which represent a
large amount of data as one scheme. Generation of one scheme saves resources
and thus makes the processing more efficient both in reference to acquisition
[21] and in reference to transfer [19, 21, 22, 24-26].

The Basic Style

The basic style yields reduced achievements due to intensive generation of


concrete imagery pictures of the basic level combined with poor abstract symbols
of the high level. This creates a load resulting from concrete images which
affects the acquisition and transfer.
384 / ROSENFELD AND KANIEL

The Combined Style

The combined style yields achievements like the high style. The achievements
stem from the combination of the generation of basic level concrete imagery with
high level abstract symbols. Generation of concrete imagery does not harm the
processing efficiency if it is generated with a high amount of abstraction.
The affect of the abstraction is stronger than the affect of the load of concrete
imagery. This result was not observed at the beginning of the study. As stated
in the introduction, the simple assumption presumed that only generation of the
highest level of abstract imagery would contribute to high achievements. The
conclusions from this study show that generation of concrete imagery does not
harm the efficient process, if it is generated with a high amount of abstraction.
This finding supports results of studies from similar areas and strengthens our
understanding in this field.
For example, the differences in study achievements between the poor and
basic styles in contrast to the combined and high styles add insight to the findings
of Sloutsky et al. [25, 36]. Their studies have shown that use of abstract models
contribute more to studies than concrete models.
The concepts presented here can also contribute to the understanding of studies
in art, giving an exact and sensitive measure to Arnheim’s scale of imagery from
low abstraction to entirely abstract image [2, 7].
It can also contribute to the understanding of studies on “spatial forms”
of the visual image [37]. Some people experience “spatial forms” when they
think about a sequential set of items such as numbers, days, months. This
experience is involuntarily accompanied by spatial sequential forms created by
visual imagery. The nature of these forms is not analog to concrete visual imagery
but to abstract visual imagery. The concept of abstract imagery presented here
can help in understanding it.
In conclusion, the findings of this study propose a comprehensive theoretical
explanation in the framework of the processing model for abstract, general and
specific visual imagery. The explanation includes:
1. creation of a theoretical and practical scale from concrete to abstract, which
describes the visual imagery activity with high sensitivity;
2. construction of a tool that describes categories of this scale, categorizes
the two levels, basic-concrete and the high-abstract, into four styles; and
3. understanding the combined activities of two levels of imagery when proc-
essing studying task.

REFERENCES
1. S. M. Kosslyn, Image and Brain: The Resolution of the Imagery Debate, MIT Press,
Cambridge, Massachusetts, 1994.
ABSTRACT VISUAL IMAGE / 385

2. R. Arnheim, Visual Thinking, University of California Press, Berkeley, California,


1969.
3. P. Palladino and R. De Beni, When Mental Images are Very Detailed: Image Gener-
ation and Memory Performance as a Function of Age, Acta Psychologica, 113,
pp. 297-314, 2003.
4. S. Gardini, R. De Beni, and C. Cornoldi, Can We Have an Image of a Concept?
The Generation Process of General and Specific Mental Images, Imagination,
Cognition and Personality, 23:2&3, pp. 193-200, 2003-2004.
5. E. Rosch, C. Mervis, W. Gray, D. Johnson, and P. Boyes-Braem, Basic Objects in
Natural Categories, Cognitive Psychology, 8, pp. 382-439, 1976.
6. E. Rosch and B. Lloyd (Eds.), Cognition and Categorization, Erlbaum, Hillsdale,
New Jersey, 1990.
7. E. Winner, Visual Thinking in Arts Education: Homage to Rudolf Arnheim,
Psychology of Aesthetics, Creativity, and the Arts, 1:1, pp. 25-31, 2007.
8. R. De Beni, F. Pazzaglia, and S. Gardini, The Generation and Maintenance of Visual
Mental Images: Evidence from Image Type and Aging, Brain and Cognition, 63:3,
pp. 271-278, April 2007.
9. S. Gardini, R. De Beni, C. Cornoldi, A. Bromileyb, and A. Venneri, Different Neuronal
Pathways Support the Generation of General and Specific Mental Images,
NeuroImage, 27:3, pp. 544-552, September 2005.
10. K. D. Owens and M. A. Clements, Representations in Spatial Problem Solving in
the Classroom, The Journal of Mathematical Behavior, 17:2, pp. 197-218, 1998.
doi: 10.1016/S0364-0213(99)80059-7.
11. D. Van Garderen, Spatial Visualization, Visual Imagery, and Mathematical Problem
Solving of Students With Varying Abilities, Journal of Learning Disabilities, 39:6,
p. 496, 2006. doi: 10.1177/00222194060390060201
12. M. Hegarty and V. K. Sims, Individual Differences in Mental Animation during
Mechanical Reasoning, Memory and Cognition, 22, pp. 411-430, 1994.
13. V. K. Sims and M. Hegarty, Mental Animation in the Visual-Spatial Sketchpad:
Evidence from Dual-Task Studies, Memory and Cognition, 25, pp. 321-332, 1997.
14. M. Kozhevnikov, M. Hegarty, and R. Mayer, Students’ Use of Imagery in Solving
Qualitative Problems in Kinematics, ERIC ed 433239, 1999.
15. N. C. Presmeg, Visualisation in High School Mathematics, For the Learning of
Mathematics, 6:3, pp. 42-46, 1986.
16. N. C. Presmeg, Reasoning with Metaphors and Metonymies in Mathematics
Learning, in L. D. English (Ed.), Mathematical Reasoning: Analogies Metaphors,
and Images (pp. 267-279). Lawrence Erlbaum Associates, Mahwah, New Jersey,
1997.
17. N. C. Presmeg and P. E. Balderas-Cañas, Visualization and Affect in Non-
routine Problem Solving, Mathematical Thinking and Learning, 3:4, pp. 289-313,
2001.
18. W. Zimmerman and S. Cunningham, Visualization in Teaching and Learning
Mathematics, Mathematical Association of America, Washington, DC, 1991.
19. D. N. Perkins and G. Salomon, Transfer of Learning, International Encyclopedia
of Education (2nd Edition). Harvard Graduate School of Education, Pergamon
Press, Oxford, England, 1992.
386 / ROSENFELD AND KANIEL

20. F. Paas, A. Renkl, and J. Sweller, Cognitive Load Theory: Instructional Implications
of the Interaction between Information Structures and Cognitive Architecture,
Instructional Science, 32, pp. 1-8, 2004.
21. J. Sweller, The Worked Example Effect and Human Cognition, Learning and Instruc-
tion, 16:2, pp. 165-169, 2006.
22. J. G. Van Merrienboer, L. Kester, and F. Paas, Teaching Complex Rather than
Simple Tasks: Balancing Intrinsic and Germane Load to Enhance Transfer of
Learning, Applied Cognitive Psychology, 20, pp. 343-352, 2006.
23. D. Pawley, P. Ayres, M. Cooper, and J. Sweller, Translating Words into Equations:
A Cognitive Load Theory Approach, Educational Psychology, 25:1, pp. 75-97, 2005.
24. J. D. Bransford, A. L. Brown, and R. R. Cocking (Eds.), How People Learn:
Brain, Mind, Experience, and School, Committee on Developments in the Science
of Learning, National Research Council, National Academy Press, Washington,
DC, 1999.
25. V. M. Sloutsky, J. A. Kaminski, and A. F. Heckler, The Advantage of Simple
Symbols for Learning and Transfer, Psychonomic Bulletin & Review, 12:3, pp. 508-
514, 2005.
26. F. Paas and T. Van-Gog, Optimising Worked Example Instruction: Different Ways
to Increase Germane Cognitive Load, Learning and Instruction, 16:2, pp. 87-91,
2006.
27. C. Cornoldi, R. De Beni, A. Cavedon, G. Mazzoni, F. Giusberti, and F. Marucci, How
Can a Vivid Image be Described? Characteristics Influencing Vividness Judgments
and the Relationship between Vividness and Memory, Journal of Mental Imagery,
16, pp. 89-108, 1992.
28. C. Cornoldi, R. H. Logie, M. A. Brandimonte, G. Kaufmann, and D. Reisberg,
Stretching the Imagination: Representation and Transformation in Mental Imagery,
Oxford University Press, Oxford, 1996.
29. N. Rosenfeld, Abstract Visual Imagery as an Aid in the Study of Physics, doctoral
dissertation, Bar Ilan University, Israel, 2007.
30. D. J. Livesey, Age Differences in the Relationship between Visual Movement
Imagery and Performance on Kinesthetic Acuity Tests, Developmental Psychology,
38:2, pp. 279-287, 2002.
31. G. M. Dean and P. E. Morris, The Relationship between Self—Reports of Imagery
and Spatial Ability, British Journal of Psychology, 94, pp. 245-273, 2003.
32. W. Smith and I. E. Dror, The Role of Meaning and Familiarity in Mental Trans-
formations, Psychonomic Bulletin and Review, 8:4, pp. 732-741, 2001.
33. K. Ericsson and H. Simon, Protocol Analysis: Verbal Reports as Data, MIT Press,
Cambridge, Massachusetts, 1993.
34. K. A. Ericsson, Toward a Procedure for Eliciting Verbal Expression of Nonverbal
Experience without Reactivity: Interpreting the Verbal Overshadowing Effect within
the Theoretical Framework for Protocol Analysis, Applied Cognitive Psychology, 16,
pp. 981-987, 2002.
35. A. K. Ericsson, Valid and Non-Reactive Verbalization of Thoughts during Per-
formance of Tasks, Journal of Consciousness Studies, 10:9-10, pp. 1-18, 2003.
36. J. A. Kaminski, V. M. Sloutsky, and A. F. Heckler, Concrete Instantiations of
Mathematics: A Double-Edged Sword, Journal for Research in Mathematics
Education, 40:2, pp. 90-93, 2009.
ABSTRACT VISUAL IMAGE / 387

37. C. M. Price, What Kind of Mental Images are Spatial Forms? Cognitive Processing,
10(Suppl. 2), pp. 276-278, September 2009. doi: 10.1007/s10339-009-0322-7

Direct reprint requests to:


Naftali Rosenfeld
9 Giza St.
Jerusalem 97279
Israel
e-mail: naftaliros@gmail.com

You might also like