ABSTRACT
A DEVELOPMENTAL STUDY OF VISUAL AND
HAPTIC PERCEPTION OF FORM
By

Gary Clinton Johnson

The intent of this research was to test certain hypotheses im—
plied by theory and past research concerning the deve10pment of cross—
modal (inter-system) perception of form in vision and touch. One
hypothesis (Hebb, 1941; Birch & Lefford, 1963; Rudel and Teuber, 1964)
was that developmental change in visual—haptic inter-system perception
is due to change in visual or haptic perception alone. A second hy-
pothesis (Piaget and Inhelder, 1956; Pick, 1964; Hebb, 1949; Gibson,
1966; Gibson, 1969) was that deve10pmental change in haptic perception
can be attributed to change in children's exploratory behavior. A
third hypothesis Munsinger & Kessen, 1966) was that developmental
change in form perception and exploratory behavior is related to the
complexity of the forms used.

To test these hypotheses 144 kindergarten, second- and fifth-grade
children were given targets successively in pairs to explore visually
or haptically, and asked to tell whether each comparison target was the
same as or different from the preceding standard. Twelve children at
each grade level were assigned to each of four conditions corresponding
to the perceptual systems to which the targets were presented: visual-
visual, visual-haptic, hapticdvisual, and haptic—haptic.

The visual targets were projected slides of irregular, black,

planometric forms differing in perimeter and ranging from 5 to 48

 

 

‘ A-

\

 

angles in contour. Haptic targets, identical to the visual ones, were
cut out of k inch masonite and mounted on plywood fields. On haptic
trials on opaque shield was used to occlude the targets from view. Ex—
ploration time for the targets was not limited; the inter—target interval
was 7 seconds. Sixteen same pairs and sixteen different pairs of targets
were presented to each subject.

The children's identifications of targets as the same or differ-
ent and the time they spent exploring them was recorded. Haptic ex-
ploratory behavior was recorded on video-tape.

The children's correct responses were analyzed as a function of
Grade (age), presentation condition, type of pair (same or different),
and target complexity, by analysis of variance. In general, children
at all grade levels in all conditions performed better than chance.
Correct responses were made more often when standard targets were ex—
plored visually than when they were explored haptically and when com—
parisons were explored with the same perceptual system as the standard
than when they were not.

In general, correct identifications increased with age in all
conditions although more in the haptic-haptic and visual-haptic than
in the visual-visual and hapticdvisual conditions. The rates of change
with age in the inter-system conditions were dissimilar and change in
neither condition can be attributed to change in visual or haptic per—
ception alone. Complexity effects on correct responses were negligible:
high complexity targets were correctly identified only slightly less
often than low complexity targets and this effect was unrelated to age.

Several aspects of the children's exploratory behavior, especially

as they could be identified in the video-tape records of their haptic

exploration, were quantified and analyzed in an attempt to relate
variation in exploration to variation in correct identification of
targets. Age—related differences in exploration on correct, as compared
to incorrect, response trials were also analyzed. In general, it was
found that older children explore more thoroughly than younger children,
especially on forms of higher complexity. Exploration on correct trials
was found to be different from exploration on incorrect trials but

only for Older and not for younter children. Since exploration differ-
ences were related to response accuracy for fifth—graders but not kinder-
garteners it appears that variation in some factor other than explora-

tion influences the accuracy with which kindergarteners identify targets.

REFERENCES

Birch, H. G. & Lefford, A. Intersensory development in children.
Monographs of the Society for Research in Child Development,
1963, 28, (Serial No. 89).

Gibson, E. Perceptual learnipg;and development. New York: Appleton-
Century-Crofts, 1969.

 

Gibson, J. J. The senses considered as perceptual systems. Boston:
Houghton—Mifflin, 1966.

 

Hebb, D. O. The organization of behavior. New York: Wiley, 1949.

 

Munsinger, H. L. & Kessen, W. Stimulus variability and cognitive
change. Psychological Review, 1966, 12, 164—178.

 

Piaget, J., & Inhelder, B. The child's conception of space. New York:
Humanities Press, 1956.

Pick, H. L., Jr. Perception in Soviet psychology. Psychological
Bulletin, 1964, 62, 31-35.

 

Rudel, R. G. & Teuber, H. L. Cross—modal transfer of shape discrim—
ination by children. Neuropsychologia, 1964, g, 1-8.

 

A DEVELOPMENTAL STUDY OF VISUAL AND

HAPTIC PERCEPTION OF FORM

By

Gary Clinton Johnson

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Psychology

1973

DEDICATION

This thesis is dedicated to the memory of my grandfather

Duncan A. McRae .

ii

ACKNOWLEDGEMENTS

There are many people who helped me with this research and
who deserve thanks. I would like to eXpress my appreciation to John
Paul McKinney, Ellen Strommen, Ralph Levine, and Lester Hyman for
their advice and cooperation at many phases of this research and to
Merrill Mitler for his help with some of the computer analyses.

The friendly cooperation of Mrs. Jachalke, principal Of Dimondale
Elementary School, Dimondale Michigan, and her staff were also appre-
ciated.

I would especially like to thank my advisor Lauren Harris for
his constant support and good advice throughout this entire project.

This research was supported by a National Institute of Health

Predoctoral Fellowship number to the author.

11;

TABLE OF CONTENTS

INTRODU CT ION I I I I I I I I I I I I I I I I I I I I I I I I
THE SENSATION-BASED THEORY OF INTER-MDDAL PERCEPTION OF FORM
THE GIBSONIAN THEORY OF TRANS-SYSTEM PERCEPTION OF FORM . .

RESEARCH RELATING TO A SENSATION-BASED THEORY OF INTER-
SYSTEM PERCEPTION I I I I I I I I I I I I I I I I I I I I

A. Research Using a Transfer—of—Training Method . . . .

B. Research Using a Simultaneous Match-to-Sample
HethOd I I I I I I I I I I I I I I I I I I I I I I

FORM AS AN INDEPENDENT VARIABLE IN PERCEPTION RESEARCH . . .
EXPLORATORY BEHAVIOR AND PERCEPTUAL—COGNITIVE DEVELOPMENT .
METHOD . . . . . . . . . . . . . . . . . . . . . . . . . . .
A. Subjects . . . . . . . . . . . . . . . . . . . . . .

B. Targets and Their Presentation . . . . . . . . . . .

C. Apparatus . . . . . . . . . . . . . . . . . . . . .

D. Design and Procedure . . . . . . . . . . . . . . . .
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . .
A. Analysis of Children's Same-Different Judgements . .

1. Effects of Grade and Condition . . . . . . . .

2. Effects of MOde of Presentation of Standard
F0 ms I I I I I I I I I I I I I I I I I I I

3. Effects of Mede of Presentation of Comparison
Forms I I I I I I I I I I I I I I I I I I I I

iv

Page

12

17

17

21
32
34
4O
4O
4O
44
45
48
48

51

56

57

4.

5.

6.

Results Pertaining to Inter—System and Intra—
System Presentation of Standard and Com—
parison Forms . . . . . . . . . . . . . . .

Effects of Form Complexity . . . . . . . . .

Effects of Type of Form Pair . . . . . . . .

B. Analyses of Children's Exploratory Behavior . . . .

1.

DISCUSSION

APPENDIX A

APPENDIX B

REFERENCES

Analysis of Duration of Visual and Haptic
Exploration and the Relation Between Dur—
ation and the Number of Correct Judgements

Relation Between Duration of Exploration of
Standards and Correct Judgements . . . . .

Relation Between Ihration of Exploration of
Comparisons and Correct Judgements . . . .

Within-Subject Variation in Standard and
Comparison Exploration Times and Their
Relation to Correct Responses . . . . . . .

Analyses of Video—Tape Records of Children's
Haptic Exploratory Behavior and the Re-
lation Between Exploration and Correct
Judgements . . . . . . . . . . . . . . . .

Page

58
59
61

7O

7O

7O

86

96

97
129
140
142

144

10.

11.

12.

13.

14.

LIST OF TABLES

Partial Summary of Analysis of Variance of Correct
Responses . . . . . . . . . . . . . . . . . . . . . .

Partial Summary of Analysis of Variance of Explo—
ration—time for Standard Forms . . . . . . . . . . .

Partial Summary of Analysis of Variance of Explo—
ration—time for Comparison Forms . . . . . . . . .

Summary of Analysis of Variance of Percent of Explo—
ration—time on the Perimeter of Standard Forms . . .

Summary of Analysis of Variance of Percent of
Perimeter Explored of Standard Forms . . . . . . . .

Summary of Analysis of Variance of Percent of Explo—
ration-time on the Perimeter of Comparison Forms . .

Summary of Analysis of Variance of Percent of
Perimeter Explored of Comparison Forms . . . . . . .

Summary of Analysis of Variance of Difference-Scores
for Time Exploring the Perimeter of Standard . . . .

Summary of Analysis of Variance of Difference-Scores
for the Amount of Perimeter Explored of Standards . .

Summary of Analysis of Variance of Difference—Scores
for Time Exploring the Edge of Comparisons . . . . .

Summary of Analysis of Variance of Difference—Scores
for the Amount of Perimeter Explored of Comparisons .

Summary of Analysis of Variance of "Overlap" of
Perimeter Exploration . . . . . . . . . . . . . . . .

Summary of Analysis of Variance of Overlap of
Perimeter Exploration Associated with Correct
Respons as I I I I I I I I I I I I I I I I I I I I I I

Summary of Analysis of Variance of the Percent of
Irrelevant Perimeter Exploration . . . . . . . . . .

vi

Page

50

71

'72

100

101

102

103

114

115

117

118

122

124

127

LIST OF TABLES——continued

Table Page
15. Summary of Analysis of Variance of the Percent

of Irrelevant Perimeter Explored Associated
with Correct Responses . . . . . . . . . . . . . . . 128

vii

LIST OF FIGURES

Figure Page

1. Correct responses as a function of Grade and
Condition I I I I I I I I I I I I I I I I I I I I I I 52

2. Correct responses as a function of form Complexity . . 6O

3. Correct responses as a function of Session and
Complexity. . . . . . . . . . . . . . . . . . . . . . 62

4. Correct responses as a function of Complexity and
type Of form Pair I I I I I I I I I I I I I I I I I I 64

5. Correct responses as a function of Grade and type
Of Pair I I I I I I I I I I I I I I I I I I I I I I I 66

6. Correct responses as a function of type of Pair
and test Session . . . . . . . . . . . . . . . . . . 67

7. Correct responses as a function of type of Pair
and condition I I I I I I I I I , I I I I I I I I I I I 68

8. Standard exploration time as a function of form
compleXity I I I I I I I I I I I I I I I I I I I I I 7 5

9. Standard exploration as a function of form Com—
PleXity and Grade I I I I I I I I I I I I I I I I I 7 7

10. Standard exploration time as a function of form
Complexity and Condition . . . . . . . . . . . . . . 79

11. Standard exploration time as a function of form
Complexity, Condition, and Grade . . . . . . . . . . 8O

12. Standard exploration time as a function of test
Session and form Complexity . . . . . . . . . . . . . 85

13. Comparison form exploration time as a function of
Grade and Complexity . . . . . . . . . . . . . . . . 89

14. Comparison form exploration time as a function of
Complexity and Condition . . . . . . . . . . . . . . 90

viii

LIST OF FIGURES-—continued

Figure Page

15. Comparison form exploration time as a function of
Complexity, Condition, and Grade level . . . . . . . 92

16. Comparison form exploration time as a function of
Complexity and type of Pair . . . . . . . . . . . . 95

17. Exploration time on the edge of Standard forms as
a function of Grade and Complexity . . . . . . . . . 104

18. Exploration of perimeter of Standard forms as a
function of Grade and Complexity . . . . . . . . . 107

19. Exploration time on the edge of Comparison forms
as a function of Grade and Complexity level . . . . 110

Research in the psychology of perception has long been directed
toward the description of the capacities and an understanding of the
development of capacities of the various perceptual systems in man and
other animals. Within recent years there has been a burgeoning of
interest in the development of perceptual skills that require use of
more than just a single perceptual system. The terms used to describe
this kind of perception have varied.

Most commonly, perception in which.more than one perceptual system
is involved has been called "crossdmodal", "inter-modal", or "inter—
sensory" perception where the perceptual systems are thought of as sense
modalities. Adherents of this view conceive of perception as grounded
in the sense data deriving from the sensory modalities that the target
is stimulating.

Less commonly, perception involving more than one perceptual sys—
tem has been called "amodal" perception (Gibson, 1969) where the per—
ceptual systems are thought to be modes of external attention to in-
formation in stimulation. James Gibson (1966) conceives of perception
as the detection of information specifying targets in the organism's
environment and, when this information can be detected by more than a
single perceptual system, amodal perception is possible.

In the pages to follow, these two views of perception, as they
pertain to perception in more than one perceptual system, and the re-

search relating to each, will be discussed.

5

THE SENSATION-BASED THEORY OF INTER-MODAL PERCEPTION OF FORM

The sensation-based theory of perception has seen variation and
elaboration since its origins in the writings of the British Empiricists.
Current major proponents of the theory include Hebb (1949), Vernon
(1954), and Bruner (1957). Since their views vary slightly from one
another it is hard to attribute the general theory described here to
any single person.

Although vision has been the focus of most of the interest in
perception, theories of perception and perceptual learning have evolved
to explain perception in other modalities as well. Inter-modal percep—
tion is merely a special case to which these general theories have been
applied.

Psychologists have long recognized that any target is complex in
terms of the dimensions along which.variation can take place over time.
The "same" target can assume many different forms, colors, and textures,
and the relations among these elements can vary widely. Furthermore,
the orientation of the target relative to the receptor (the retina or_
the hand) can change over time. Clearly, the pattern of energy at the
receptor (the impingement) aroused by the target at any moment in time
is a unique event that, in all liklihood, will never be duplicated.

The problem facing the psychologist is to explain the fact that many
unique impingements can lead to a "common" response; this seeming
paradox has been called the "constancy" problem in the research liter-
ature.

The traditional sensation-based theory of perceptual learning

holds that the organism must somehow take all of these unique sensory

6
experiences and combine them, or "categorize" them, as Bruner (1957)
has put it, so as to be able to attach the "same" response to the
group. Thus, a single class of stimuli can be associated with some
response (class), and learning can be observed. This kind of stimulus
classification includes what has been called "stimulus generalization"
(both primary and secondary) and "acquired equivalence of cues", and
both phenomena have been demonstrated in numerous studies.

The sensation-based theory of perception as specifically applied
to inter-modal perception dates back to Sherrington (1906), who sug-
gested that, while man does not have a greater number of sensory
channels and while his senses are not necessarily more sensitive than
other animals', man does possess more adequate "linkages" between his
sensory channels. The significance of this difference lies in the
adaptive advantage given an organism having a greater capacity to per-
ceive a threat to his well—being when the threat can be perceived by
more than one rather than just a single modality.

Inter-modal perception is possible, of course, only with targets
that can be sensed by more than one modality. For example, color can
be "sensed" only visually; form can be sensed either visually or
tactually and therefore is a target dimension that has been studied in
visual and tactual interdmodal perception. Before discussing research
pertaining to visual and tactual inter—modal perception of form the
theoretical basis for this research will be discussed.

Although the sensation-based theory of perceptual learning has
been applied explicitly only to intra-modal perceptual learning, its
applicability to inter-modal learning is obvious. In intra-modal per-

ceptual learning a person comes to realize that an object having a

concentrically radiating texture or brightness gradient and a round
appearance is three dimensional and is called a sphere. Thus, over
time and experience with spheres he comes to associate those visual
cues with the three-dimensional quality of sphericity. Inter-modal
learning takes place when he looks at and feels a sphere at the same
time. Under such circumstances of contiguous visual and haptic sen-
sations, these sensations become associated with each other. In the
future the feel of an object will arouse sensations similar to past
haptic sensations which in turn had been associated with particular
visual sensations aroused by a particular object that had been looked
at. Feeling an object thus can give some "idea" of the visual appear-
ance of the object: one should be able to identify an object in a
visual array that is the.same as another object that is simultaneously
being felt but not seen.

Hebb's (1949) view of perceptual learning is in many ways similar
to this traditional sensation—based theory, but because of its impor-
tance among theories of perceptual learning it deserves special mention.
According to Hebb, when an organism first encounters a target the
energy at the receptor field (impingement) is conveyed to the brain
where that particular impingement is somehow recorded. Subsequent ex-
periences with the target will produce impingements in some ways
identical to, and in other ways different from, previous impingements,
and the "record" will reflect this partial overlap. Brain cells that
are aroused consistently whenever a target is presented becomes "linked"
in what Hebb calls a "cell assembly". Once an assembly has been "laid
down" it subsequently can be aroused in its entirety when only part of

the assembly is stimulated. The development of such a mechanism would

8
facilitate recognition of targets because the organism need detect
only a small part of the total information (impingement configuration)
inherent in a target to respond apprOpriately. Errors in recognition
also become possible because different targets can have elements in
common so that a new target may not be distinguished from an old one
for which an assembly has already been established. This model best
describes perception with the eye, but it also can be seen to describe
perception with the hand and, to a lesser extent, perception with other
modalities.

Perceptual learning, then, consists of laying down of cell
assemblies, and, extending this notion, inter~moda1 perceptual learning
would consist of the laying down of assemblies that are aroused simul-
taneously by impingements deriving from different modalities. Such an
extension of Hebb's model to inter~moda1 perceptual learning would re—
quire the assumption that cell assemblies, though deriving from dif-
ferent modalities, ggp_have cells in common given certain circumstances
of arousal. Once links between assemblies relating to different modal-
ities have been established, stimulation deriving from only a single
modality, touch, for example, would cause the entire assembly, in-
cluding its visually laid-down parts, to be aroused.

Given a sensation-based theory such as Hebb's, it is clear that
inter-modal and intra-modal perception should require different exper-
iences for their development: the contiguity of sensation from vision
and touch is essential for the formation of an interdmodal assembly;
interdmodal perception should be impossible so long as the perception
of a target in the different modalities takes place successivly.

Evidence that interdmodal and intraamodal processes are distinct might

be obtained in either of two ways: In one, normal development might

be 'simulated in a training experiment; in the other, children of dif-
ferent ages would be tested in visual and haptic intra-modal and inter-
modal tasks and inferences drawn from performance differences among

the different tasks to the process or processes underlying those dif-
ferences.

In a training experiment subjects would be trained to discrim-
inate among novel (unfamiliar) forms which they can only touch, only
look at, or simultaneously look at and touch. Subjects in this latter
condition in this way presumably receive practice associating visual
and tactual cues deriving from the same form. Subsequent transfer
tests could then be given in which visual and tactual input, though
simultaneous, are kept separate. (Subjects would not be able to see
the form they are feeling but, instead, could see a form that is likg
the form they are feeling.) Only those subjects receiving training
in which bimodal input was associated shOuld be able to transfer this
training to the test condition.

The most important consideration in this experiment is that the
forms be unlike anything the subject has experienced before. If such
forms could be devised, no age differences relating to the forms would
be expected, as they should be equally unfamiliar to subjects of dif-
ferent ages. In practice, the criterion of complete novelty is im-
possible to meet, since any form will be similar in sogg_way to a form
with which a subject has had prior experience. Therefore, the relative
novelty of a form might be expected to decrease as the subject's age,
and therefore his experience, increases.

Transfer between vision and touch, without explicit training in

10
the context of the experiment, would be expected to vary as a function
of the novelty of the objects used relative to other objects with
which the subject has had prior eXperience associating input to the
two modalities. Age differences should be found here too just as in
conditions of intra—modal presentation.

Despite the inevitable confounding of novelty effects with
training effects in any age differences observed, a bimodal training
study would be valuable insofar as it could be presumed to model the
natural conditions for inter—modal perceptual learning. Such a study
has never been conducted. If the study were developmental, it would
have the further value of identifying the lower limits of training
effectiveness. Birch and Lefford (1963) hypothesize that the develop-
ment of interdmodal functioning found in their research (to be discussed
later) reflects neurological change in the "linkages" between the
senses. Only a training study that does not succeed with youngest
subjects can lend support to neurological and not to experiential in—
terpretations. Obviously failure to train young children can be due to
many factors including the one hypothesized by Birch and Lefford, so
any such failures would not constitute very strong support for the
hypothesis.

In the absence of evidence to the contrary, it is as reasonable
to postulate an experiential basis for behavior change as it is
reasonable to postulate a neurological basis, and it should be pointed
out that Birch and Lefford's evidence is entirely behavioral and not
histological.

Finally, negative results in and of themselves are of question-

able value, but their value increases significantly in certain contexts.

11

The value of training studies would be enhanced by supportive findings
from "controlled observation" of the natural development of inter-modal
processes.

All past research on inter—modal perception has been designed
to measure the natural development of the capacity. In terms of a
traditional view of perceptual learning one would expect a positive
relation between intrarmodal competence in perception (recognition or
identification) of common objects and the subjects' age because of the
increased experience with those objects that is presumed to take place
over time. For the same reason, intermodal perception ought to improve
with age. While intervmodal and intrarmodal perception of common ob-
jects require distinct sorts of experiences for their development,
strictly speaking it is impossible to tell to what extent these ex-
periences are distinct as they naturally occur. Perhaps exclusively
intradmodal experience with most forms occurs only rarely and, in the
usual course of events, any target is apprehended with more than one
modality. If such is the case, interdmodal and intrarmodal perception,
though relying on distinct experiences, may show overlapping and hence
indistinguishable courses of development. 0n the other hand, if inter-
modal perceptual development were not found to parallel perceptual
development of either modality taken separately, one could justifiably
conclude that distinct processes are involved in the two types of per-
ceptual learning.

Research on inter~moda1 perception has been oriented implicitly,
if not explicitly, toward this traditional view of perceptual learning.
But before considering that research, Gibson's view of perceptual learn-

ing will be described, although it has stimulated but a single study,

12
because this view is so different from a sensation-based theory and

is of such potential importance for understanding interdmodal phenomena.

THE GIBSONIAN THEORY OF TRANS-SYSTEM PERCEPTION OF FORM

Though Gibson (1966) acknowledges that no two impingements are
ever exactly alike, he does not accept the view that one's perceptual-
cognitive apparatus somehow puts impingements together, fitting them
into categories. Instead, he suggests that there are invariant pro-
perties (critical features) of objects which organisms learn to detect.
Objects are identified by an organism despite variation in impingements
because the information defining the form for the organism is present
in each impingement. "Constancy" for Gibson has nothing to do with
impingements somehow being put together; rather it means that there is
constancy of information over successive impingements.

Similarly, perception of an event by means of more than one

perceptual system1 does not involve "associating" or "putting together"

 

1James Gibson prOposes that "perceptual system" and not modalities
are more reasonable units and he seems completely correct in this.
Under the traditional view, the visual modality refers to retinal and
other processes neurologically related to those receptor cells. Al-
though the muscles that control the eye are very much a part of visual
perception and have been thought to play an important role in illusions
and depth perception, for example, they are not included in what is
properly called the visual modality. Similarly the tactual modality
has not included prOprioceptive and kinesthetic cues although all three
occur, inextricably linked together, in most tasks in tactual percep-
tion. Gibson has recognized this confusion over what is meant when one
refers to a visual or a tactual perceptual task and has proposed that
the term 'system' be used to include all the different kinds of infor-
mation from the various receptors that are used in a visual or haptic
perceptual task.

The term "modality" rather than "system" has been used thus far
in the description of the sensation—based theory because that is the
term used by these theorists. Probably they intended to mean what
Gibson means with "system". Because Gibson's term is so much more
reasonable it will be used in the remainder of this paper.

13
the information deriving from each. Rather, the organism detects in-
formation with both systems, with each "kind" of information specify-
ing the same thing. In other words, information does not have to be
associated because the target does not vary; the only change is in the
perceptual system by which the information is detected. As was men-
tioned earlier, Eleanor Gibson (1969) has called information that is
not specific to a perceptual system "amodal". This term will not be
used here.2 Instead, information relevant for more than one system
will be called "trans—system" information. Texture density and the
relations among‘the sides of an irregular shape are two aspects of
targets that might be distinctive critical features and relevant for
both touch and vision and hence could be called trans—system infor-
mation.

Trans-system information constitutes a subset of all the infor-
mation that can be detected with any perceptual system. The psycholo—
gists' job, so far as Gibson's theory is concerned, is to determine
what kinds of perceptual invariants are detectable exclusively by a
given system and what kinds are trans-systemic. Perceptual develop-
ment, according to Gibson's view, amounts to learning to detect those
features of events or things that distinguish them from.other events
or things. Children, therefore, should become better in perceptual
tasks as they grow older because they should become increasingly able

to detect these distinctive features.

 

2The term "amodal" is misleading in two respects: first, amodal
suggests the use of modalities and not systems in perceptual tasks when
James Gibson has argued for the greater reasonableness of systems as
information detecting units. Second, calling information 'amodal' sug-
gests that it has no effect on a perceptual system. James Gibson has
argued against the possibility of perception without sensation, which
is what the term 'amodal' implies.

14

There should be similar improvement with age in trans—system as
in single system tasks although performance in trans-system tasks may
be lower. Performance should be lower in a trans-system task because
there is less information relevant to more than one system than there
is information relevant to any single system, and a direct relation
ought to exist between the amount of relevant information and perfor-
mance in a perceptual task.

Training studies would be relevant to Gibson's theory of per-
ceptual development insofar as the training experience could be pre-
sumed to model "natural" experience. Eleanor Gibson (1969) suggests
that a training study involving trans-system invariants need only con-
sist of training in one perceptual system; learning to detect distinc-
tive features in one system can subsequently facilitate detection in
another system, since the features are common to both. Such a train-
ing study is an appropriate test of a prediction from this theory, but
it seems to be an unnecessarily stringent test because it assumes
equivalent capacity for detecting information in the different systems.
So far as the theory is concerned, it is only necessary that the subject
not be given an opportunity to "associate" information detected by the
two systems. Thus, he might be given training in each of two systems
separately and then asked to perform in a trans-system task. Such
training should facilitate performance because his ability to detect
information with each system, including trans—system information,
should improve.

Unfamiliar objects must be used so that the child will have
something to learn and also to minimize the role of past experience in

which discriminating the objects could have been accomplished with both

15
systems Operating concurrently. The stimulus objects, of course, also
must be detectable by either perceptual system.

Pick, Pick, and Thomas (1966) conducted a training study that
bears on Gibson's theory, in which first grade children were taught to
discriminate visually and haptically among certain types of letter-like
forms and their transformations. In previous research (Gibson, Gibson,
Pick, & Osser, 1962; Pick, 1965) the distinguishing features of these
forms had been found to be detectable in both the visual and haptic
perceptual systems. In the Pick, Pick, and Thomas study, initial
training took place either in the visual or haptic system followed by
re—training in the other system. The effects of training were measured
by comparing performance in a system with and without prior training
in the other system.

This was not a developmental study; its objective was to compare
the effects of different characteristics of forms on transfer of train-
ing. There were several types of forms and several variations of each
type. The different variations followed certain principles which were
applied to each type so that the same variation was found across the
several types of forms.

All children learned to discriminate among a particular set of
letter-like forms containing certain types and certain transformations
of each type. The forms used in the transfer task differed in three
ways from those used in the initial discrimination task, and a differ-
'ent group of subjects was used with each. The first group was re-
trained with the same types of forms but with different variations of
each type of form. Transfer would be expected for children in this

group if, in the initial learning trials, they had learned something

16
about the basic type of form with which the different variations
could be compared in a kind of match—mismatch fashion. A second group
of children was retrained with different types of forms but forms that
had been transformed in a similar way as the initial training forms.
Transfer would be expected in this condition if the children had
learned something about the critical features (dimensions of differ-
ence) among the forms. A third group of children was retrained with
forms of a different type and transformed in a different way from the
training forms. No transfer would be expected of this group except,
perhaps, that of a general "learning-to-learn" variety not specific
to the forms used.

The results showed transfer only in the first two groups though
inconsistent transfer in the first. The Gibsonian predictions are
'most consistently supported by this study (the second group described
above), but the associationist (Hebbian) position also receives some
support (the first group described above); it is not known what factors
produced the uncertain results in the latter case. This study is in-
teresting nonetheless because it marks a first attempt to use forms
that differ in a systematic way from each other and for which the re-
levance of the dimensions of difference for both perceptual systems
has been demonstrated. Whether or not the transfer shown can be
attributed to the trans—system nature of these dimensions of difference
cannot be determined. Since this was not a developmental study and
since intra-system comparison groups drawn from prior research had
slightly different test conditions, it could be that transfer was
possible because prior bi—system experience was relevant for these

forms. In another experiment bi-system training might also be carried

17
out to model what might be normal experience in perceptual development.
Further research with forms like those designed by the Gibsons looks
particularly promising.

Numerous studies have been conceived in the spirit of a sensation-
based theory of perceptual development and in general have attempted
to show that: 1) human subjects are capable of inter-system perception,
2) the capacity for inter-system perception improves with age and, 3)
inter-system and intra-system perceptual development are distinct pro-
cesses. Few of these studies have attempted to consider all three of
these basic questions within a single design. Furthermore, because
different methods have been used to study the same process, it is
likely that the picture or characterization which each study gives us
is colored to some degree by the method employed.

Each of these methods will be considered separately and the re-
search using each presented and evaluated to determine to what extent
these three questions have been answered. Finally, we will examine
how each.method treats the development of the perceptual-cognitive

organization of visual and haptic information about form.
RESEARCH RELATING TO A SENSATION—BASED THEORY OF INTER-SYSTEM PERCEPTION

A. Research using a transfer-of-trainingpp§thod. Inter-system per-
ception of form has been studied using a transfer-of-training method.
With this method subjects are divided into two groups. Several forms
are presented to one group for visual exploration and to the other for

haptic exploration.3 The forms are presented repeatedly until they

 

3To be strictly accurate, it is necessary to describe the Opera—
tion of presenting the forms and not the subject's response (looking

/

18
can be discriminated to some criterion level of performance. The same
forms are then presented to each group in the other perceptual system
until the criterion level performance is again reached. The number of
trials or errors to criterion.might be used to show the rate of learn—
ing in each system. The effect of pre-test exposure to forms in the
other system can be shown by comparing the results from both groups.
For example, the visual performance of one group that has had prior
experience in the haptic system can be compared with the visual per-
formance of the other group that has had no prior experience with
the forms.

An important drawback to this method is that learning experience
2££“22.13 confounded with transfer effects. Two additional control
groups, shown different forms in each system, are needed to distinguish
general learning experience from experience that is specific to those
forms presented previously in the other perceptual system.

Finally, initial training in visual and haptic discrimination
must be studied as a function of age in order to tell whether or not
changes with age in inter-system discrimination occur independently of
improvement in intra-system discriminations. Change in inter-system
discriminations may be merely an artifact of improvement in intra-
system discriminations.

Gaydos (1956) and Eastman (1967) used a transfer of training

‘method to investigate inter-system perception of form with adult

 

at them, say) because his response does not always correspond per-
fectly to the operation. Although it is important to distinguish
between the activities of the experimenter and the subject it will be
awkward to maintain this distinction throughout the discussion to
follow. Therefore the distinction has been deliberately blurred in
the text.

19
subjects. Gaydos paired verbal labels with amorphous shapes which she
allowed one group of subjects to look.at and another to feel. When
the labels associated with the forms had been learned in one system
the forms were presented in the other until the proper labels were
again learned. Gaydos compared the relearning performance of one
group with the initial learning performance of the other for both touch
and vision. Transfer was demonstrated in both systems with greater
transfer from touch to vision than from vision to touch.

Eastman used irregular randomly generated forms of 4, 8, and 16
sides for discriminanda but in other respects treated his subjects in
the same way as did Gaydos. His transfer test, however, consisted of
only one trial in the other perceptual system and his measure was the
number of forms labeled correctly. Like Gaydos, Eastman found more
transfer in the T-V than in the V-T condition. There was no form com-
plexity effect: the number of correct judgements the subjects made
did not vary as a function of the variation in the discriminanda.

Two similar studies have been conducted with children. Blank,
Altman, and Bridger (1968) taught four—year-olds to discriminate
between two block-like forms. The forms were presented simultaneously
for visual or haptic exploration (the children could use both hands
in the haptic condition) until the criterion level for discrimination
was met. The two forms then were presented in the other perceptual
system until the subjects were able to discriminate them. The re-
sults show transfer in the V-T but not in the T—V condition.

Pick, Pick, and Thomas (1966) presented several letter—like
forms to first grade children (six—year—olds) for visual or haptic ex-

ploration. This study, which was discussed earlier in detail, was

20

designed to answer a different question from the three main questions
posed above and, as a consequence, the training methods used in the
two transfer conditions were different. Their results show transfer
of discrimination training in both the V-T and the T—V conditions in
six-year-olds. The differences in training under the two conditions
preclude evaluation of the relative efficiency of the transfer con—
ditions.

To summarize, these studies show transfer of experience from
vision to touch in subjects ranging from four—year-olds to college
students. Transfer from touch to vision has also been found in sub-
jects between six years and college age although not in four-year-olds.
Finally, there is some suggestion that transfer is more efficient among
college students from touch to vision than from vision to touch, which
suggests that, although later to develop, transfer from touch to vision
is greater.

These differences in the relative efficiency of transfer across
subjects of different ages may or may not reflect important develop-
mental change. Any conclusion.must be tentative because these studies
not only used subjects of different ages but they also differed in the
forms used and in procedural details so that what appear to be age dif—
ferences in transfer may only be differences in procedure. It is also
unclear whether the observed age differences in transfer reflect
inter-system or intra-system perceptual development and finally,
whether these differences might merely reflect improvement with age in
"learning to learn".

Only the study by Pick, Pick, and Thomas used control groups to

ensure that what was transferred was information specific to the forms

21
and not merely general task experience. Although their results show
no non-specific transfer effects, it would be premature to assume that
such effects may not Operate in other settings with subjects of dif-
ferent ages and with different forms to be discriminated.

Any conclusion about the develOpment of inter—system transfer
with age from research.using this method must await a study in which
the same forms and procedures are applied to subjects of different
ages in both an inter-system and an intra-system transfer task.

B. Research usipg:a simultaneous match-to—sagple method. A simultan-
eous match-to—sample method (often called the "conflict" method) has
been used to study inter-system equivalence of form perception. With
_this method forms are presented to the two perceptual systems simulta-
neously: a single form (the standard) is presented in one system,

and several forms (the comparisons) are presented in the other. The
subject is asked to pick from the comparisons the form that is the same
as the standard. A correct choice presumably reflects equivalence
between the two perceptual systems.

Two important points should be made about this method. First,
the relatively greater efficiency of visual relative to haptic ex-
ploration makes haptic and visual identification of a standard among
several comparisons very different tasks. Results, therefore, are
likely to depend on the system of presentation, so counterbalancing
is necessary. Counterbalancing is required only so long as several
comparison forms are used; the problem is avoided when only one form
is presented to each perceptual system.

Second, this method, though commonly used, does not really

permit conclusions about the development of inter—system equivalence.

22
Improved performance with age may mean improvement in equivalence
o£_improvement within a perceptual system. Two intra-system groups
(V-V and T-T) are also needed (the haptic condition would require
that the subject use both hands) so that inter-system and intra-system
changes in performance can be distinguished from each other.

The simultaneous match-to—sample.method has not been used to
study inter—system perception in adults, though there have been
several studies with children. Rudel and Teuber (1964) presented
three-dimensional geometric shapes to two groups (V~V and T-V) of
three—year-old children. Children in the V-V condition made more cor—
rect judgements than children in the T-V condition (3.3 vs. 2.3).
These findings are suggestive at best: no statistical analyses were
reported, and the means for the two groups were based on the perfor-
'mance of three children per group each of whom.made only five judge—
ments.

Lavrent'eva and Ruzskaia (1960) presented Gaydos' forms to
three—, four—, five—, and six-year-old children. Two forms were pre-
sented in one perceptual system while a third was presented in the
other. The subjects were asked to indicate which one of the two com-
parison forms was the same as the standard. Both a V—T and a T-V con-
dition were used.

The data were presented as percent errors but, with no further
analysis and no mention of the number of subjects involved, the
reliability of differences between conditions at different age levels
is hard to assess. In general, the data suggest that there was no
improvement in discrimination from a chance level in either condition

between the ages of three and four. Subjects in the T-V condition

23
showed a marked improvement from 502 (chance) to 25% errors between
four and five years of age. There was no further improvement between
five and six. However, subjects in the V—T condition performed no
better than chance through five years of age. Then, between five and
six, their performance improved markedly from chance to near—perfect
level. The results suggest earlier improvement in the T-V condition,
and later improvement and ultimately superior performance in the V-T
condition.

Birch and Lefford (1963) studied inter-system equivalence between
vision and touch in five- to eleven-year-old children. They used forms
from the Sequin form board test; the children looked at a single form
while feeling several others in succession. Nine comparison forms
corresponded to each standard form with two of the nine being the same
as the standard. The children were asked whether the comparison and
the standard were the same or different. Consistent with Lavrent'eva
and Ruzskaia, the older children made fewer errors than did younger
children.

Birch and Lefford's study is flawed in several important ways,
however; first, they used only a V-T condition and did not counter-
balance with respect to the perceptual system in which comparison
forms were presented (a particularly significant oversight, since it
was their aim to show the development of transfer between different
pairs of perceptual systems). Thus, they cannot tell whether develop-
mental differences between conditions are due to differences between
the pairs of systems or to differences with respect to the particular

system in which the comparisons or standards were presented.

24

Second, the error rate was found to vary as a function of the
judgement required: ‘more errors were made when two forms were the
same than when they were different and a minimum error rate was reached
by eight years of age for "same" pairs while errors for different
pairs continued to drop after eight years. These findings might mean
that different perceptual processes are involved in same and different
judgements. Birch and Lefford did not present an equal number of "same"
and "different" pairs of forms, so the children's responses of "same"
and "different" were not equally likely events and as a consequence,

a response bias may have been induced. Therefore, the different rates
of improvement for "same" and "different" pairs may reflect a per-
ceptual-process difference or merely a methodological bias; a proper
interpretation cannot be made.

Finally, Birch and Lefford presented the same standard and com-
parison forms in the same order to all children in a visual-kinesthetic
condition before presenting them again in the visual-haptic condition.
While the first condition is not of interest here, it is likely that
this prior experience influenced the results in the visual-haptic con-
dition, particularly in view of the partial identity of information
deriving from haptic and kinesthetic exploration. Without counter-
balancing the order of the two conditions it is impossible to tell what
effects, if any, prior kinesthetic experience with the forms had on
the subsequent haptic condition. Thus, experience effects as a function
of age were confounded with changes in inter-system.equivalence as a
function of age. In view of the significant methodological ambiguities
in this study, no conclusion appears warranted regarding the develOp-

ment of transfer between.vision and touch.

 

I."

 

25

A study by Blank and Bridger (1964) has also used the simulta-
neous match—to—sample method. Three-, four—, and five—year—old children
were given pairs of common forms (cylinder, triangle) to look at while
feeling one behind their backs. The children were asked to point to
the one form they were shown that was the same as the one they were
simultaneously feeling. Blank and Bridger did not counterbalance for
the system in which comparisons were presented. Consistent with
Lavrent'eva and Ruzskaia, four— and five-year-olds but not three—year—
olds were able to identify the form.

It is clear from LaVrent'eva and Ruzskaia's findings that
equivalence is in part a function of the perceptual system in which
the comparison forms are presented. Thus Blank and Bridger have only
partly described the changes in inter—system equivalence between the
ages of three and five. Finally, in view of the absence of intra—
system comparison groups in these studies using the simultaneous
match-to—sample method, research with this method has not convincingly
shown that inter-system and not intra-system processes are involved in
the developmental changes observed.

C. Research using a successive match-to-sample method. This method

is the same as the one just described except that a memory component

is involved, since forms are presented to the two perceptual systems
successively rather than simultaneously. Typically, a single form

(the standard) is presented in one perceptual system followed by
several forms (the comparisons) in the other. As with the simultaneous
method, groups that are counterbalanced with respect to system of
presentation (V-T and T—V) are needed. Intra—system comparison groups

(T-T and V-V)) are also required with this method.

26

This method has been used to study inter—system perception in
two experiments with adults. Vaught (1968) used Gaydos' forms in a
haptic exploration, visual comparison (T-V) task. (The number of forms
and their manner of presentation are not described in his report).

The subjects identified the form they had felt among the several they
looked at significantly better than chance indicating that there was

transfer from touch to vision. The appropriate controls were lacking
however: there was no V-T group, nor were there intra-system groups.

Cashdan (1968) also used Gaydos' forms but had his adult subjects
explore the forms through a hole in an opaque board thereby forcing
exploration of successive parts of the form. In this way visual and
haptic exploration were made similar. Following explorations of each
standard, five comparison forms were presented in succession and the
subjects were instructed to identify the standard among the comparisons.
Four conditions were used, two inter—system (V-T, T—V) and two intra-
system (V-V, T—T). Each subject had two trials in each of the four
conditions.

This study is unique in that subjects' exploration was con-
trolled. Unfortunately, the order of treatment conditions apparently
was not varied across subjects, and the inter-system treatment con-
ditions were confounded with forms (all subjects in a treatment con-
dition saw the same forms and different forms were used in each treat-
‘ment condition). Apparently realizing that factors were confounded,
Cashdan replicated the first study with new forms assigned to each
treatment (and still confounded). Since the assignment of forms made

a difference, the only consistent findings were that the intra-system

27
conditions (V—V and T-T) were easiest and most difficult respectively.
The inter-system groups were of intermediate difficulty.

The successive presentation method also has been used to study
inter-system perception in children. Lobb (1965), using random
shapes, presented five standards each followed by an array of four
comparisons consisting of the standard and three other forms. The
subjects, eighth-grade children, were assigned to two inter-system
and two intra-system groups. Each child received five trials with
each standard. In general, presenting the standards visually led to
better performance than presenting them haptically. Differences
between the groups were evident only on the first and not on later
trials: especially on early trials, performance was depressed by
having the subjects switch to the haptic system to identify comparisons
when standards had been presented visually. Finally, there was some
evidence on early trials that the V-T group performed better than did
the T-V group.

Rudel and Teuber (1963) presented three-dimensional forms to
four- and five-year—old children. Five comparisons were presented
in succession following each of five standards. In general, children
in the V-V condition made the most correct judgements, and children
in the T-T condition made the least, with children in the inter-
system (V-T and T-V) conditions performing in between; five-year-olds
tended to make more correct judgements than four-year-olds. These re-
sults are only suggestive, however. The data were not analyzed; only
group means were presented; each mean was based on the performance
of only three children each of whom got but five trials, and the dif-

ferences between means were small.

28

Hermelin and O'Connor (1961) presented 10 pairs of letter—like
forms, members of each pair shown successively, to four groups (two
inter-system and two intra—system) of five-year-old children. The
children were asked whether the two forms of a pair were the same or
different. The analysis showed no differences between the groups,
but the authors do not report whether performance was better than chance
although it might be assumed that it was.

Researchers in the Soviet Union have studied inter-system per—
ception of form among children between the ages of three and seven.
Lavrent'eva and Ruzskaia (1960) tested two groups of children (the
number per group was not mentioned) at each age level, one age group
in each of the two inter-system conditions. Each child was given
Gaydos' forms one at a time to either look at or feel. Following
exploration of each standard the children were asked to identify it
in an array of three comparison forms presented in the other system.
The results indicated greater T—V transfer than V-T transfer among
three-, four-, and five-year-olds. By six years of age, however,
there was no difference between T-V and V-T groups.

The report makes no mention of intra-system control groups. How-
ever, a reference to this experiment in Pick's (1964) review of Soviet
research in perception includes a table giving summary scores for two
intra-system groups of subjects as well as for the two inter-system
groups named above. The results for these two intra-system conditions
show fewest errors in the visual system and most errors in the tactual
system, relative both to each other and to the inter-system conditions.

Improvement was marked in the visual system as a function of age:

29
errors drOpped from 50% at age three to zero at age five. Errors
also dropped in the haptic system, from 70% at age three to 40% at
age six.

The reliability of these results is hard to assess. Typically,
the Soviet researchers report only percentages, give no indication of
the variability of subjects within groups, and report no statistical
tests. Pick (1964) regards the results as suggestive at best.

In general, it does not appear that any of the research thus
far conducted with any of the methods has been both properly designed
and properly executed so as to permit inter-system and intra-system
perceptual development to be distinguished from each other. Bryant
(1968), in a review of a subset of studies reviewed here reaches this
same conclusion. The greatest need in all of this research is for a
study of inter-system perception of form in children between the ages
of three and eleven having appropriate intra—system controls.

Even.with adequate control groups and data analysis, the extant
studies would remain inconclusive because the manner of use of the
three basic methods of analysis often has allowed the developing pro-
cess under study to be obscured by other psychological processes that
are developing concurrently. One particularly important instance of
this confounding is the extent to which these studies have relied on
the subject's memory. Both the successive match-to-sample method and
the transfer—of-training method require the subject to remember some—
thing about the form or forms to which he was initially exposed. While
memory for form within or between perceptual systems has not been
studied developmentally, it seems likely that changes in this capacity

would take place between the ages of three and seven. Therefore, it

7!.

30
seems unwise to study perceptual development in such a way that memory
factors are inextricable from the perceptual process of interest.
Support for this methodological point can be found in the study by
Rudel and Teuber (1964), who report that their three- and four-year-old
subjects were unable to perform on either the within- or between- per-
ceptual systems—tasks when forms were presented successively. These
same children could perform when the forms were presented simultane-
ously. Since an important difference between successive and simulta—
neous discrimination tasks is the memory required of a subject, the
observed performance differences between children in the two tasks
may be attributable to an insufficient memory capacity. It must be
mentioned, though, that Rudel and Teuber not only presented standard
and comparison forms successively; they also presented multiple com-
parisons among which the standard was to be identified.

The presentation of multiple comparisons, especially in succes-
sion, may confuse children of different ages in different ways. For
example, Rudel and Teuber (1964) asked the children to identify "the
one" that was the same as the standard. If a child understood those
instructions and decided that the first or the second comparison form
was "it", what must he have thought when the experimenter said nothing,
proceeded to the next comparison form, and asked whether that one was
"it"?

Age-related confusion when many comparisons are successively
presented might arise in another way: while the child tries to remem-
ber the standard form as each comparison is presented, he must also try
to £22 confuse memories of incorrect comparisons with his memory of

the standard as other comparisons are subsequently presented. In

31
other words, when a subject views the fifth comparison in a series,
it should be more difficult for him.to remember the standard that pre-
ceded the series because of interference from memories of the first
comparisons. There is no reason to assume that such confusion will
not arise in different degrees with children of different ages. Again,
these features of the procedure make it difficult if not impossible
to separate change in the process under study from change in sub-
jects' responses to some aspect of the measurement of that process.
All but one of the studies reported above, using a successive match-
to—sampleemethod, used multiple comparisons following a single prior
presentation of the standard.

Some of the studies also imposed time limits on exploration,
thus preventing free exploration of the forms. The use of time limits
could be appropriate in a study of the efficiency of exploration, but
would be clearly inapprOpriate when the aim.is to assess the child's
capability to perform. The use of time limits also would suppress
individual differences between subjects of the same age and could con-
ceal differences in the efficiency of exploratory behavior. Since
the effects of a time limit are likely to be related to the subjects'
age, the results would tend to show age change in behavior as
dichotomous when it might be continuous.

While we probably can conclude that inter-system perception is
possible in children as young as four or five years of age, the extant
studies have not satisfactorily plotted the growth of the capacity
for inter-system perception, and they have not shown that inter-system

and intra-system perceptual learning are distinct processes. Finally,

32
while we know that children can make inter—system judgements at age
four, we do not know whether their inability to make such judgements
before that time reflects lack of experience or some state of neurolog—

ical immaturity.'
FORM AS AN INDEPENDENT VARIABLE IN PERCEPTION RESEARCH

Virtually all research on inter-system perception has been con-
cerned with the £§g£_of inter-system perception rather than with the
nature of the discriminanda involved. In other words, most investi-
gators have attempted to assess the information-organizing capacity
of cognitive—perceptual systems while neglecting to ask what is being
organized. Only Eastman (1967) and Pick, Pick, and Thomas (1966)
(reviewed earlier) have attempted to measure the relation between
specific properties of forms and inter-system perception and only the
Pick et a1. study showed any relation between form properties and per-
formance. This lack of interest in the role of form parameters in
inter—system perception probably reflects the lack of an adequate
metric for form for either vision or touch. Indeed, as recently as
1968, Brown and Owen suggested that attempts to define a metric for
form lead to "methodological dyspepsia". From their and others' work
it seems to be impossible to design forms that differ on only a few
dimensions (that can be simultaneously controlled) while not differing
unsystematically on other dimensions at the same time.

Even though a form cannot be completely specified, numerous
studies of visual perception of random shapes have been carried out,

and some interesting relations have been demonstrated between certain

33
form variables and certain psychological variables. In particular,
Munsinger, Kessen, and Kessen (1964) have shown that stated prefer—
ence for random shapes varies as a function of the "complexity" of
the shapes (although now it seems that the critical variable was in
fact the perimeter of the forms). Looking time has also been found
to vary as a function of form complexity or perimeter (Johnson, 1968).
Since the information contained in a form is a direct function of its
complexity it seemed reasonable to these researchers, and to others as
well, that more "processing" would be required for a subject to learn
to recognize a complex shape than a simple one. Mere complex shapes
should also be harder to remember because of the greater information
load they impose on.memory. Munsinger and Kessen (1964) and Munsinger,
Kessen, and Kessen (1964) hypothesized that stated preference for
random shapes would accord with the information processing ability of
subjects (which should vary as a function of age) and the available
information in each form to be processed. They found an age—invariant
preference for forms of about 10 sides and differences between five—
year-Olds, ten-year-olds, and college students in their preferences
for complex shapes. In general, the older the subjects the lg§§_
preference shown for the most complex shapes. Munsinger and Kessen
suggest that this relation reflects changing information—processing
ability with a concomitant change in exploratory behavior. They suggest
that young children explore only parts of forms, consequently process
about the same amount of information from both simple and complex
forms, and therefore prefer them about equally. Adults, however, ex-
plore the forms in their totality and, finding the more complex forms

harder to process, prefer them less than simpler forms. This work

34
is pertinent to the analysis of inter—system perception because it
suggests that a subject's ability to process information about form
should vary as a function of his age and/or experience with random
shapes and the information content, or complexity, of the forms to
be processed in the task. It also implies a direct relation between
the subject's ability to process information and his exploratory be—

havior. This relation deserved further discussion.

EXPLORATORY BEHAVIOR AND PERCEPTUAL-COGNITIVE DEVELOPMENT

Two views of perceptual-cognitive development emphasize the re-
lation between exploratory behavior and knowledge of the object ex-
plored. On the one band, Piaget and Inhelder (1956) suggest that
the child's perceptual activity is essential to his cognitive develOp-
ment:

...the evolution of spatial relations proceeds at

two different levels. It is a process which takes

place at the perceptual level and at the level of

thought or imagination. (p. 3)
Perceptual activity results in a child's develOping a "scheme" which
is a kind of cognitive representation of an object-as-explored. In
other words, the representation of an object, a child's knowledge of
it, is a function of the child's perceptual activity. It reflects
the information detected by the perceiving organism and is not a 22px,
of the object as it exists independent of the organism.

Representation of an object thus has a figurative aspect.

Piaget and Inhelder (1956) emphasize that it also has an operative

aspect in that the representation (scheme) is an organizing structure

35
for external acts: "... during the development of representational
space, representational activity is, in a manner of speaking, re—
flected or projected back on to perceptual activity." (p. 4).

0n the other hand, Leont'eva's theory of perception (Pick, 1964)
holds that kinesthetic feedback from motor acts (reafference) con-
stitutes the basis for perception. Exploratory behavior thus pro-
duces a cOpy of the physical object. Thus, for both Piaget and In-
helder and Leont'eva there is some correspondence between perceptual
activity and what the organism knows about an object although the
closeness of that correspondence differs for the two views.

Piaget's description of perceptual—cognitive development also
can be seen to parallel Hebb's. Piaget's prototypical child, through
systematic and thorough exploration, "puts the form together" in his
representational system; Hebb's child forms cell assemblies. The
similarity does not end here: Piaget has found that young children,
while they explore erratically and make mistakes identifying geometric
shapes, are very good at identifying familiar objects in spite of in—
complete exploration of them. Older children also explore only parts
of familiar forms and are also good at identifying them. It is as if
the form has already been "put together" in their representational
systems such that detecting any part of the composite can in a sense
elicit or activate knowledge of the whole. Hebb has ascribed the same
sort of economy to the arousal of cell assemblies once they have been
established. It seems, therefore, that complete exploration of a
form is required for the establishment of its representation, but,
once it has been represented, only partial exploration is necessary

for its identification.

36

Although James and Eleanor Gibson (1966, 1969, respectively) do
not consider representation to play a part in perception, they do re-
cognize the importance of perceptual activity in perception. Accord—
ing to their view, organisms actively seek information, invariant over
time, that specifies objects (distinctive features) in their environ-
ment. Perceptual development involves improvement in the ability to
extract information, or, to paraphrase E. Gibson, increased economy of
information pickup.

Facts about visual and haptic exploratory behavior are scarce
but suggest that age changes in exploratory behavior parallel changes
in other measures of psychological development. Munsinger and Kessen
(1964) suggested that young children look at parts of forms while
older children and adults, though they may look at parts, put them
together to form wholes. They subsequently attempted to show that if
children look just at parts and not wholes, they will make more mis-
takes in a form identification-discrimination task than will adults
who can grasp wholes (Munsinger & Kessen, 1966). In general, this
was what they found though other interpretations are possible.

More recently, vurpillot (1968) has shown, by directly recording
eye movements, that children under five years of age tend to explore
complex figures only partially while older children explore them in
detail before making judgements about them. She found a close re-
lation between the completeness of a child's exploratory behavior and
the accuracy with which he could tell whether two complex pictures
were the same or different. Vurpillot's work is clearly the best to
date on the relation between visual exploration and visual discrimin-

ation.

37

Zinchenko and Ruzskaia (1960a, b) have also linked children's
visual exploratory behavior with accuracy of form discrimination.

They used Gaydos' amorphous shapes and recorded eye movements with a
camera positioned at the center of the form. They found that three-
and four-year-olds tend to look at the center area of the forms much
of the time while children five years and older tend to follow the
edge of the form with their eyes. Older children could discriminate
better than younger ones presumably because their exploratory behavior
was more thorough. Since information about the shape of a form is
located at its boundaries, older children should be getting more
information about the shape than younger ones. This research would be
more convincing, however, had the camera not been placed at the center
for the forms; the sample record of eye movements that Zinchenko and
Ruzskaia present shows that the younger children who looked at the
center of the form.looked mostly at the lens of the camera. Therefore
the age related differences in exploration and discrimination accuracy
may have merely reflected the younger children's greater interest in
the camera.

Soviet research on haptic perception indicates similar age-
related changes in exploratory behavior and accuracy of discrimination,
though haptic discrimination develops later. Piaget and Inhelder
(1956) also related haptic discrimination to exploratory behavior in
a way similar to the Soviets. In experiments with two- to seven—year-
old children they report that, until five years of age, children ex-
plore unfamiliar geometric shapes haphazardly, centering on only some

part of each shape before making a judgement. Older children explore

38
in much greater detail, and, presumably, their identifications of
haptically explored forms consequently are more accurate.

Piaget's geometric shapes and the amorphous shapes used by the
Soviets were unfamiliar to the subjects. Older children explored them
in detail as would be expected when the forms have not yet been
schematized, but the younger children explored them only in part which
should prevent them from adequately representing the forms. Yet
Piaget's finding that young children can identify familiar things in-
dicates that such children are capable of representing forms. It
appears that they can "put the forms together" though it might take
much longer to do so. Presumably if they had had several opportunities
to explore the geometric shapes they would have been able to identify
them too though still exploring them only partly. The suggestion here
is that, while both four- and seven-year-old children are capable of
representation, they form the representation in greatly different
ways.

One interpretation of these research findings is that children
of different ages differ in the strategies they use to discover dis-
tinctive features of forms. Older children might explore all features
of unfamiliar forms so as to detect all their potentially distinctive
features. They need explore only parts of familiar forms because the
features that distinguish those forms for the child are only a subset
of all features that comprise them.

Young children, perhaps, explore only some of the features of
unfamiliar forms thereby detecting fewer distinguishing features. They
can identify familiar forms so long as they can find their distinguish-

ing features.

39

Another interpretation of these research findings is that
children of all ages will try to explore all aspects of a form to be
able to identify it subsequently but that how much they explore will
vary as a function of their ability to organize the information they
are obtaining and the amount of information to be gotten from the form.
The forms used in research thus far have been very simple but appar-
ently contain too much information for young children to organize.
Consequently, young children explore only part of them. Older
children are able to organize the information in simple forms so they
explore them completely. The implication, however, is that more com-
plex forms might exceed older children's organizing capacity, so
exploration of such forms would be incomplete.

It therefore would be particularly interesting to present very
simple and very complex forms to children of different ages to see
whether older children explore the simple forms completely, the complex
forms only partly.

The current research represents an attempt to relate develOp-
mental change in exploratory behavior to developmental change in
inter-system and intra-system perception of form in vision and touch.
Clearly past research has not satisfactorily resolved questions con-
cerning inter-system as distinct from intra-system development. It
has also regarded inter-system perception as a strictly perceptual
and not a perceptual-cognitive capacity relating in some way to the
information aspects of the forms themselves. Finally, it has not con-
sidered exploratory behavior as a function of the subject's ability
to organize information and the information in the forms to be or-

ganized. Consequently, exploration of form has not been studied beyond

40
six years of age in spite of other evidence that cognitive develop—
ment continues well beyond that time.

The current study is a developmental analysis of intra—system
and inter—system perception of form in vision and touch in children
between the ages of five and eleven. Four groups of children at each
age level were tested to permit a distinction between inter—system
and intra-system processes. Forms of various complexities were used
to permit assessment of the relation between information complexity,
correct judgements, and exploratory behavior. Haptic exploratory be-
havior was recorded and related to children's correct judgements,
form complexity, and whether forms, taken in pairs, were the same or

different.

METHOD

A. Subjects. The subjects were 144 kindergarten, second grade, and
fifth grade children, 48 at each level, attending Dimondale Elementary
School, Dimondale, Michigan. The subject pOpulation could be described
as lower middle class and rural. The children were tested in a vacant
classroom in their school.

B. Targets and their presentation. The visual targets were projected
slides of irregular black planometric forms on a white background.

The haptic targets were irregular planometric forms cut from k inch
(.64 cm) masonite, each.mounted on a 14 inch (35.56 cm) plywood field.
The forms ranged from approximately 8 inch (20.32 cm) to approximately

13 inch (33.02 cm) across their greatest extent.

41

As yet there is no metric for form that specifies certain
dimensions along which variation would be appropriate for both touch
and vision. There is some evidence from research on visual behavior
(McCall & Kagan, 1967; Karme14) that suggests that the amount of edge
(perimeter) constituting a form is a relatively good predictor of
subjects' stated preferences and looking times. Although research
has not been extended to the haptic system, it seemed reasonable
to vary perimeter because spatial extent as well as shape is relevant
for both the visual and haptic perceptual systems.

To simplify generation of these forms, they were permitted to
vary simultaneously both in perimeter and in number of angles as well.
Number of angles correlates highly with perimeter but it has been
found to be a slightly poorer predictor of visual behavior. While
these factors need not be thus confounded, it is impossible to control
for all factors. For example, number of angles could be kept constant
and perimeter alone varied. However, this would necessitate vast dif-
ferences in the size of the forms. Inevitably, controlling some factors
leads to the confounding of others. At this point there is no end to
the problem and one can only control for what past research suggests
are the major factors, and leave others to vary as they may.

The forms were randomly generated and had 5, 8, 10, 14, 20 28,
36, and 48 angles in contour. (Details of the method used to generate
the forms are presented in Appendix A.) There were six different forms

at each level of variability, altered so as to all have the same

 

4Personal communication, 1966.

42
perimeter. Perimeter varied between successive levels by a factor of
1.255.

The forms were presented in pairs with eight pairs in each of
four blocks. Each level of variability was presented once in each
block. Different counterbalanced orders of the levels of variability
were designed for each block (Appendix B). The four blocks were pre—
sented in two different orders.

The first member of a form pair will be called the standard and
the second the comparison. Standard and comparison forms were the
same for half the pairs at each level of variability and different for
the other half. These pairs within a level were distributed across
blocks such that an equal number of 'same' and 'different' judgements
were correct in each block with the restriction that no more than
three of the same type of judgement would be correct in succession.

Standard and comparison forms were presented successively.
Successive rather than simultaneous presentation was used for two
reasons. First, simultaneous presentation of these forms, given
their size, would have necessitated presenting them in different planes
and/or in different positions relative to the children in the inter-
system condition. This was thought to have certain disadvantages: In
particular it would make the four conditions less nearly alike in all
respects except the system to which targets were presented.

Second, this study set out to determine the role of the develop-

ment of the information-organizing capacity in inter-system perception,

 

5An exception was made for the intervals between the 8 and 10
.and 10 and 14 turn forms. The interval between 8 and 14 turn forms
was a factor of 1.25 with the lO-turn forms lying midway between.

43
and memory is an.important aspect of this capacity. Successive pre—
sentation would seem to require that the child process the standard
form in its entirety so as to be able to compare the second with it;
a simultaneous presentation would not impose such a requirement. If
children have to process all the information in the standard, it is
more likely that age differences and differences within age as a
function of complexity can be observed.

Standard and comparison forms were presented for as long as each
child wished to explore them. The inter-target interval was 6 seconds
and the inter-trial interval was approximately 8 seconds. A six second
inter-target interval was used rather than a shorter time period to
allow for fading of the "icon" which results when a visual target is
presented and which can facilitate identification of a subsequently
presented visual target (Neisser, 1966). The projection secreen was
dark during the inter—trial and the inter—target intervals.

The inter-target interval was electronically controlled and the
duration of presentation of standard and comparison targets was
electronically timed and automatically recorded. The experimenter
began a trial by advancing the carousel slide projector. The counter
began cumulating elapsed time, in half seconds, simultaneously with
the presentation of the target on the screen. When the experimenter
terminated presentation of the standard the total elapsed time was
printed out, the counter automatically reset, and the timed 6 sec.
inter-target interval commenced. At the end of this interval the
comparison target was automatically presented and, simultaneous with
its appearance, the counter began to coumlate elapsed time during pre-

sentation of the comparison form. At a signal from the experimenter

44
target presentation was terminated, total elapsed time was printed
out, and the counter reset to zero completing the trial. Thus, the
experimenter controlled the onset and offset of the standard target
and the offset of the comparison target. The projector was run even
during haptic exploration so that the experimenter could pace the
presentation of the haptic targets. It was, of course, impossible
for the children to see the target on the screen while they were ex-
ploring a target haptically.
C. Apparatus. The apparatus consisted of a 30 inch (76.2 cm) x 48
inch (121.92 cm) opaque vertical panel with a 14 inch (35.56 cm)
square translucent glass screen 5 inch (12.7 cm) from the bottom of
the panel and centered left to right. A carousel projector presented
slides of the forms for the subjects' visual exploration from behind
the panel. The panel was placed on a table in front of and 40 inches
(101.6 cm) away from the subject.

The forms to be explored haptically were placed in a stand and
moved to a vertical position in front of and approximately 12 inches
(30.48 cm) away from the subject. Thus, both visual and haptic ex-
ploration took place with the forms in the same plane and position
relative to the subject. It was impossible to present visual and
haptic targets at the same distance from the subject because presenting
visual forms as close to §_as the tactual forms h§g_to be placed
created an 'unnatural' situation in which pilot subjects reported
feeling uncomfortable.

A chin rest was attached to the edge of the table and kept the
subject's head in a nearly constant position with respect to the loci

of presentation of the targets. An opaque visor that could be raised

.__ __,_,..-.—

 

 

 

 

45
and lowered was attached to the chin rest and was used to occlude
the subject's vision during haptic exploration trials.

A 1 inch (2.54 cm) diameter hole was cut in the panel 2 inches
(5.08 cm) above the top edge of the projection screen. Through it a
closed-circuit—television camera (CCTV) monitored the subjects'
visual exploratory behavior. A second camera was placed to the side
and slightly behind the subject to monitor the subjects' hand move-
ments. A video—tape record was made of the children's exploratory
behavior by means of these cameras.

D. Design and_procedure. Forty—eight children at each grade level
were randomly assigned to one of four treatment conditions giving
each condition 12 subjects per grade level. The conditions differed
with respect to the ways in which standard and comparison forms were
presented for exploration. There were two intra-system conditions,
one in which both forms were presented for visual exploration (V—V)
and another in which they were presented for haptic exploration (T-T).
In one of two inter-system conditions standards were presented for
visual exploration and comparisons for haptic exploration (V—T). In
the other inter—system condition the systems in which standards and
comparisons were presented were reversed (T—V).

Children were brought to the experimental room one at a time
and pre—tested to ensure that they understood what 'same' and 'dif-
ferent' meant. The pre—test consisted of 12 pairs of 2 x 2 white cards
each bearing a black quadrangular form. The forms were randomly
generated and equated for perimeter so as to be similar to, though
smaller than, the test forms. For six pairs of the forms both members

of a pair were the same, for the other six pairs, different. The

46
pairs were presented side by side on a table in front of the child,
who was then asked whether the two were the same or different. Any
child making two or more errors in twelve trials was dropped from the
experiment. Only one child was rejected on the basis of the pre-test.
Upon completion of the pre-test the children were told:

I have been showing you pictures two at a time. Now
I'm going to show you some more things two at a time
only instead of showing you two things at the same
time, I'm going to show you first one thing and then
another. (Successive presentation was demonstrated
with the pretest cards). And, "I'm going to show you
these things over here" (a motion was made toward

the apparatus.)

Children, were then seated at the table facing the apparatus
and made comfortable in position at the chin rest. ThOse in the V-V
condition were told:

Now Ifm going to show you some more funny looking
pictures two at a time, first one and then another
one, only you will see them through this window.
0.K. Here is the first one. (A practice target
was shown) Now you can look at this for as long
as you want...until you think you can remember
what it looks like. Then I'll show you the next
one and you tell me if the next one is the same,
or not the same, as this one. This is kind of a
remembering game, so look at that picture until
you think you can remember what it looks like. Do
you think you can remember it? When you think you
can remember it say '0.K.' and I'll show you the
next one.

If the child continued to look for a while longer he was asked again
whether he had seen it enough and reminded to say 'O.R.' when he had.
After he responded, the second (comparison) target was shown and he
was asked whether it was the same as or different from the first one.
He then was told that he would be shown some more pictures two at a

time, and the instructions were partly repeated.

47
Children in the T—T condition were told:

Now I'm going to show you some more things two at

a time, only I want you to feel them rather than
look at them. Do you see this funny looking thing
here? (The child was shown the practice form for
haptic exploration). Here, go ahead and feel it.
This is going to be a feeling game and not a look-
ing game so let me lower this visor here so you
can't see this thing you're feeling. (The visor
was lowered). O.K. Now here's what I want you

to do. I'm going to give you things to feel two at
a time, first one and then another one, and I want
you to tell me if the two things are the same or
different. Feel all around this first one until
you have a good idea as to what it's like. As soon
as you have felt it enough and you think you know
what it's like, say 'O.K.' and I'll show you the
next one.

The rest of the instructions were identical except for such changes as
required by the change in condition.

Children in the inter-system conditions were told that they
would be shown things two at a time, first something to look at (feel)
and then something to feel (look at). The instructions to these sub-
jects were essentially the same as for the intra-system conditions.
Children in all conditions were not told whether or not they were
correct but were told in a general way that they were performing well
in the task.6

Children in all conditions were tested on 16 of the 32 trials
without interruption. Because of the length of time taken by children

in the inter-system conditions and the haptic intra-system condition

(some children in the latter condition took well over 30 minutes to

 

6Children‘were not told whether they were correct on each trial

because: 1) it seemed desirable to avoid differential reinforcement
of children, as this would be confounded with age differences in task
performance, and 2) this research is concerned with the exploration
strategies children use, not the strategies they can learn to use;
reinforcement might shape exploratory behavior differentially as a
function of age.

48
complete the first 16 trials) all were returned to their classroom
for some period of time before completing the last 16 trials. The
children were reminded of the instructions upon their return to the
test room.

Most of the children remained in the classroom while another
child was tested although, for some, other interruptions intervened
between periods in the test room. No more than 18 hours was allowed
between testing periods (as occurred when the end of the school day
prevented completion of the task). Three children were eliminated
from the study because they were not again present in school until
more than 18 hours had elapsed. Other children eliminated from the
sample were: a kindergarten girl who refused to participate beyond
the pre-test, one kindergartener and one second-grader who insisted
on peeking at the forms they were feeling, another second-grader who
was on tranquilizer drugs, and a fifth-grader for whom two slides

stuck in the projector, upsetting the inter—stimulus interval.

RESULTS

Two kinds of data were collected in the task. The first was the
children's same-different judgements of the pairs of forms. The
second pertained to the quality and duration of the children's ex-
ploration of the forms prior to making the same—different judgement.
These data will be presented in this order followed by a discussion of
the relation between the judgements and exploratory behavior.

A. Analyses of children's Same-Different judgements. Each child's

same—different judgement for each pair of forms was scored 1.0 if

49
correct and 0.0 if incorrect. Average scores for each subject and for
each group of subjects ranged from approximately 0.50 (chance) to 1.0
(all judgements correct).

Two separate questions must be asked of these data: first,
under what conditions did the children make more correct judgements
than would be expected on the basis of chance?; second, how did the
independent variables affect performance?

The effects of the independent variables were determined by a

seven—factor analysis-ofdvariance, summarized in Table l.

 

Insert Table 1 about here

 

The four between-subject variables were the child's Grade in school,

the Condition in which the forms were presented, the Assignment of

 

particular forms within a complexity level to an order of presentation,
and the Order in which the eight levels of form complexity were pre-

sented. The three within—subject factors were level of form Complexity,

 

type of Pgig_to be identified (same or different), and test Session.
The level of significance used for all tests was p_‘<.05.

Analysis of sub-effects (i.e., reduction of significant main
effects and interactions) was carried out by Scheffe's method (1954).
The level of significance used was p_‘<.10. Since this test is very
conservative, it has been suggested that a higher p value he used
(Ferguson, 1971; Scheffe, 1959).

Finally, to check the extent of the influence of the independent
variables on the children's judgements, confidence intervals were

established for the proportion of correct responses expected by chance

Partial Summary of Analysis of Variance of Correct Responses

Source

Between

Grade (C)

Condition (C)
G X C

Error

Within
Complexity (L)
G X L
Pair (P)
G X P
C X P
G X C X P
L X P
C X L X P
Sessions (S)
L X S
P X S
Error

Total Error

72

14

2232

4607

50

TABLE 1

l

l

95

MS

1.399
4.351
0.648

0.266

0.930
0.139
0.521
1.454
2.657
1.967
1.122
0.356
0.705
0.415
3.391
0.174

1.583

2’.

42.810
53.987

2.434

5.330
0.798
2.987
1.454
15.233
11.275
6.434
2.042
4.042
2.381

19.440

51
with each significant effect according to a method described by Hays
(1963). These confidence intervals show the lowest value a mean can
have and still be considered to be above chance, with a probability
less than .05 (one-tailed) that a sample mean above that value is from
a population whose mean is 0.50.
1. Effects of Grade and Condition. It was expected that children's
correct responses would vary as a function of the test condition.
Specifically, V-V performance was expected to be best; inter-system
performance (T-V and V—T) relative to T-T performance could not be
predicted. Finally, it was expected that performance in the inter-
system conditions would be identical.

The results of the Condition effect (p_ < .0005) support only
the first prediction: V-V children.made more correct judgements than
did children in any other condition (p_< .01). There was no signifi-
cant difference between V—T and T-T children but bgghpmade more
correct responses than did T-V children (p_< .01), who performed poor—
est of all.

The relative performance in the four conditions varied as a
function of the ages of the children (Egsdg_x Condition interaction,

p < .03) as can be seen in Figure 1. For kingergarteners, V—V

 

Insert Figure 1 about here

 

children did better than T-T or T-V children (p_< .05, p_< .01,
respectively who did not differ from each other. Children in the

V-T condition made more correct judgements than T—T and T—V children

Percent Correct

52

 

100q

90"

80"

70"

60"

 

50‘

~17

 

 

 

Q . l I l
’ E ' 2 5
Grade

_.'- ..__... , y _ A 7 __ I, - - o , ...4..._._ __

Fig. 1. Correct responses as a function of Grade and
condition.

0- an

Percent Correct

~4 BI

52

 

100'

90"

80"

70"

60-

 

 

 

0"» - )chance p (.05

 

50“

 

 

9 I I I I

K ' 2 5

Grade

Rig. 1. Correct responses as a function of Grade and
condition.

1*“.

53
and fewer than V—V children, but these differences were not signifi—
cant. Finally, except for T—V children, all kindergarteners did
better than chance.

For second-graders, too, V—V children did better than T—T and
T-V children (p.< .01) who did not differ from each other. Again, the
number of correct judgements by the V—T children was between that made
by the T-T and T-V children in but not significantly different from
either group. Second-graders in all conditions did better than
chance.

Finally, for the fifth-graders, V-V children did better than
T-V children (p_< .01). The performance of children in the T-T and
V-T conditions was between that of children in the other two conditions
but not significantly different from either, nor did they differ
significantly from each other. All fifth—graders did better than
chance.7

To summarize these findings, V-V children did best and T-V
children did poorest at all grade levels, with V-T and T-T children
performing in between.

It was also expected that children's performance would improve
with age; the results are consistent with this hypothesis (Egsdg_
effect, p_< .0005). Fifth-graders made more correct judgements than
second—graders (p_< .01) who in turn made more correct judgements than
kindergarteners (p_< .01). All made more correct judgements than would

be expected by chance.

 

7In the results presented in the following pages, there were so

few circumstances in which children failed to perform better than
chance, it should be understood that all performance is beyond chance
unless otherwise noted.

 

54

This improvement with age was not uniform, however, across the
four conditions (Eggdg_x Condition interaction). As can be seen in
Figure l, the number of correct judgements appears to increase
steadily with age in all conditions although grade—level differences
were significant only in the V—T and T-T conditions. In both these
conditions the number of correct judgements increased significantly
between kindergarten and fifth grade (p_< .01), but most of the in-
crease within the T—T condition occurred between second and fifth
grade (p.< .075). Additional cell comparisons show that the rate of
improvement was different in the T-T and V-T conditions (p.< .01).8
Finally, improvement with age in the V-V and T-V conditions was not
significant;9 other differences between grade levels within these
four conditions were also not significant.

Since it was expected that the levels of performance ought to
be identical in the inter—system conditions (V-T and T—V) it was
necessarily expected that change in performance as a function of age
ought also to be identical in the two groups. Contrary to expecta-
tion, there is same suggestion that the rates are different: V-T
performance improved significantly with age while T—V performance did
not. On the other hand, further cell comparisons (described in foot-
note 8) were not significant (p_< .10). Despite the absence of

strong evidence that performance changes differently with age in the

 

8The curves were compared by doubling the value for the 2nd
graders and subtracting the values for the other two age groups in each
condition and testing for the significance of the difference between
the resultant values for the two conditions.
, 9The magnitude of improvement in these conditions may have been
diminished by floor (T—V) and ceiling (V—V) effects.

55
V-T and T-V conditions, goy_evidence that this was the case militates
against the conclusion that the curves and the processes they repre-
sent are identical.

In summary, children in the V-V and T-V conditions differed from

 

each other at all grade levels, and older children did slightly but
not significantly better than younger children. Children in the V-T
and T-T conditions performed at a level between that of children in
the other two conditions, and children in both conditions improved
significantly, though at different rates, with age.

These findings for the interaction of Eggdg_and test Condition
are most important, for they bear on certain fundamental questions to
which this research was directed. These questions concern: 1)
whether children included in this age range are capable of inter-
system perception; 2) whether inter-system perceptual development
can be distinguished from visual or haptic intra-system development;
and 3) whether inter—system perception improves over the age range
studied here. These questions, deriving from an associationistic
theory of perception, pre—suppose the conclusion that inter-system
perception is some kind of process or thing that is invoked whenever
a perceptual task requires the joint functioning of two or more percep—
tual systems. The results suggest that this presupposition is incor—
rect: Differences between the V-T and T-V conditions both in the level
of performance and, especially, in the developmental change found in
each suggests that these tasks are not tapping some common capacity.
Each of the four form—presentation conditions appears to be unique.
Since the supposition on which the above questions were based is in—

correct, the questions are somewhat inappropriate for these data and

 

56

must be revised. It seems apprOpriate now to ask: 1) what are the
factors that determine the relative level of performance in the

four conditions and 2) what are the factors that underlie the develop—
mental change found in the T—T and V—T conditions.

Although each of the four conditions was unique, certain features
were shared with other conditions: First, in the V-V and V—T condi—
tions, the children could only look at the standard forms, while in
the T—T and T-V conditions, the standards could only be touched.
Second, comparison forms could only be looked at (V—V and T-V) or
touched (T—T and V-T). Third, T-V and V-T conditions required shift-
ing between visual and haptic perceptual systems, while T—T and V-V
conditions, whether or not a shift between perceptual systems was re-
quired, or some or all of these factors, could have influenced the
relative levels of performance among the conditions and the develop-
mental change found in them.

2. Effects of mode of presentation of standard forms. Since vision
seems to be much.more efficient than touch for the perception of form,
it is reasonable that children who looked at the standards should have
obtained more or better information and consequently made more correct
judgements than children who only touched them. The effects of the
method of presentation of the standard can be assessed by comparing
the scores of children in the V—V and V—T conditions with the combined
scores of children in the T—V and T-T conditions.

When the standard forms were presented visually, performance
was better than when they were presented haptically (p.< .01). This
was true at all grade levels, though significant only for the second—

graders (p.< .05). .leferences between groups within the visual

57
(V-V and V-T) and haptic (T-T and T-V) standard—presentation conditions
were not significant at any grade level. Therefore it seems that the
superior performance of children in the visual relative to the haptic
standard—presentation conditions is a joint effect of the combined
treatment conditions rather than the result of any one.

Although performance was better with visual than with haptic
presentation of the standards, performance seems to have improved with
age only in the haptic conditions. In the haptic conditions, the num-
ber of correct judgements increased between kindergarten and fifth-
grade (p_< .075). Both the T—T and T—V groups appear to contribute
about equally to this change. In the visual conditions, improvement
with age was small and not significant.
3. Effects of mode of‘presentation of comparison forms. Was the
method of presenting the comparison forms related to incidence of
correct judgements? To answer this question the number of correct
judgements for groups V-V and T—V were combined and compared with the
combined performance of groups V—T and T—T.11 Differences between
visual and haptic presentation of comparisons proved to be non-signifi-
cant for the grade levels combined as well as for any individual
grade level.

Improvement with age was significant among children in the haptic

conditions, between kindergarten and fifth grade (p_< .05); improvement

in the visual conditions was smaller and not significant. While it

 

v

”Although children in the v—v condition differ from children

in the T-V condition at each grade level the rate of development
across grade in the two conditions is almost exactly the same so
there is some basis for combining these data.

58
seems correct to conclude that groups T-T and V-T are both contributing
to the improvement in the haptic comparison—presentation it is clear
that improvement is not identical in the two groups: improvement
occurs at different times in the two conditions and a comparison of
the curves (described in footnote 8) shows them to be significantly
different (p_< .01).
4. Results pertainiog;to inter—system and intra-system presentation

of standard and comparison forms. A third possible basis for dis-

 

tinguishing among the four experimental conditions is that conditions
T—T and V-V require intra—system perception of standard and comparison
forms while conditions T—V and V—T require inter-system perception.
Performance in the intra-system tasks should be better than in the
inter-system tasks because not all features of form that can be seen
can be felt and vicedversa. Correct judgements can be made only when
a feature distinguishing a standard form can be identified on a com-
parison form. Children using both vision and touch should make more
errors than children using vision or touch alone because they
occasionally should detect a distinctive feature of a standard with
one perceptual system that they cannot detect with another. Younger
children especially should make more inter-system than intra-system
errors.

The results generally bore out these predictions. Children in
the intra-system.conditions made significantly more correct judgements
than did children in the inter-system.conditions (p_< .01), although
this difference was significant only for all ages combined. There
was also improvement with age among children in the intra-system con—

ditions between kindergarten and fifth grade (p_< .05) although this

59
is probably due primarily to improvement in the T—T condition: Dif—
ferences between the T-T and V—V conditions were significant at
kindergarten and second grade though not at fifth grade; there was
significant improvement in performance with age in the T-T but not in
the V—V condition; comparing the curves (footnote 8) shows them to
be different (p_< .05). There was no improvement with age in the com-
bined scores of children in the inter—system conditions.

In summary, these analyses show that the relative level of per-

 

formance in the four conditions is determined by l) the perceptual
system to which the standards were presented and 2) whether or not

the standards and comparisons were presented to the same perceptual
system. These analyses also show that haptic perception, especially

- of the standards, and less clearly, of the comparisons, underlies imr
provement with age in these conditions. Inconsistencies in age change
in the conditions involving touch, however, show that it is likely to
be the way haptic perception was used in each task and not haptic per-
ception in general that is important for the developmental change ob-
served.

5. Effects of form Complexity. It was expected that highly complex

 

forms would be correctly identified less often than forms of low com-
plexity because highly complex forms, being comprised of more features,
would make a greater demand on.memory. This expectation was borne

out. As can be seen in Figure 2, the number of correct judgements

 

Insert Figure 2 about here

 

r v' v—

generally decreased across the eight levels as form complexity increased

Percent Correct

100

90

80

70

60

SO

60

 

 

 

 

1

1

1 )chsnce p (.05

‘f l l r’ l l l 1 r,
5 8 10 1h 20 28 36 as

-Fig. 2.

Complexity

Correct responses as a function of form Complexity.

 

 

60

 

100 .

901

Percent Correct

70 .

60 -

r—-———-—--—--*--fl--

50 -1 )chance 2 (.05,

 

 

l I l l l l | I

s 8 10 1h 20 28 36 he
Complexity

.Fig. 2. Correct responses as a function of form Complexity.

61
(p_< .0005). Most of the complexity effect can be attributed to
differences between the 5-, 8—, and 14-sided vs. the 36- and 48—
sided forms (p_< .01-.10).
It also was expected that young children would have more dif—
ficulty with more complex forms than would older children. The non-

significant Grade x Complexity interaction (p_< .10) indicates that

 

this prediction.was not borne out.

Finally, it was expected that correct judgements would increase
with practice, especially for the high complexity forms. That is
essentially what happened. Children made somewhat fewer errors in
Session 11 than in Session I (Session effect, p_< .04), and across
sessions the number of correct identifications of the most complex
forms increased relatively more than correct identifications of less

complex forms as can be seen in Figure 3 (Session x Complexity effect

 

 

Insert Figure 3 about here

 

p_< .02). However, differences between sessions at each complexity
level and within sessions between complexity levels were small and

not significant.

6. Effects of type of form pair. Past research suggests that whether
the forms to be compared are the same or different is an important
determinant of success. A sufficient condition for judging forms
"different" is finding any difference between them, while a necessary
condition for judging forms the same is finding them identical in all
respects. Therefore, same pairs would require much more careful and

lengthy exploration than different pairs. The difference between same

Percent Correct

100

90

80

70

60

50

62

 

>cha;c: p (.05 I

 

 

 

 

r l l l I l l l

S 8 10 '1h 20 28 36 h8
Complexity

4% ‘7‘... - —-— . H. _. *

.Fig. 3. Correct responses as a function of Session and
Complexity.

__ ._.___-.

63

and different pairs in required exploration would be much greater for
high than for low complexity forms. Therefore, same pairs, especially
the highly complex ones, should be more difficult to identify than
different pairs. Highly complex same pairs should also be more diffi-
cult to identify for younger than older children since the need for
lengthy exploration and the information processing associated with it
should be greatest for those forms and more difficult for younger
than for older children to accomplish.

The results are in some way ipconsistent with these hypotheses.
In general, same pairs were more often correctly identified than dif-
ferent pairs, although this difference was in fact very small (.72 pg
.70, p_< .08). The difference between same and different pairs was
not related to complexity level in any simple or consishanzway as can
be seen in the Complexity X _P;a_:_LL interaction (p < .0005) shown in

Figure 4.

Insert Figure 4 about here

 

Other findings were as expected. Identification of different pairs
was unrelated to complexity while same pairs were somewhat easier to
identify when they were less complex: the difference between 8- and
48-sided same pairs was significant (p_< .10).

The hypothesized relation between type of pair and the age of
the children also received some support as can be seen in the §£§d§_x
Pgip_interaction (p.< .0005) presented in Figure 5. Although differ—

ences between same and different pairs were not significant at any

Percent Correct

64

 

 

 

 

100 -
9o 1
80 -
70 "
60 "
)chence p<.05

5°?

0 I l l 1 l l r . k'

5 8 10 11, 20 28 36 he
Complexity
Fig. 4. Correct responses as a function of Complexity and

type of form Pair.

 

65

 

Insert Figure 5 about here

 

wi— wvi

any grade level the rate of improvement in performance with age was
greater for same than for different pairs. In general, performance
across age on same pairs started lower and ended higher than per-
formance on different pairs; the difference between kindergarteners
and fifth-graders was significantly larger for same than for differ—
ent pairs (p_< .05). For same pairs, kindergarteners, second-graders,
"and fifth-graders all differed from each other (p_< .10-.Ol), while
for different pairs, kindergarteners differed from fifth-graders
(p < .05).

Consistent with the effects of age, practice led to improved
performance for same but not for different pairs (Psi; X Session in—

teraction, p_< .0005) as can be seen in Figure 6. Across sessions,

——v #7 —

Insert Figure 6 about here

 

—v— ‘—

performance on.same pairs improved (p_< .01) while performance on
different pairs did not change. In session I same and different pairs
were identified about equally well while in session II same pairs were
identified correctly more often than were different pairs (p_< .05).
Performance on same and different pairs also varied with the
treatment condition (Pgi£_x Condition interaction, p_< .0005), shown

in Figure 7. Except in condition V—T, same pairs were easier to judge

Insert Figure 7 about here

Percent Correct

66

 

100 -

80-

70

j

 

) chance p <. 05

 

50%

 

 

0 I I
x 2 5
In. _ _grode

Fig. 5. Correct responses as a function of Grade and
type of Pair.

67

 

r—
100 ‘
90"

p 80‘

O

I

f:

O

U

43

5 7o-

0

Ft

0

04
(>0q

‘ r

 

so?
.0

 

> chance pg: 05

 

I ~ I

I II
Session

Fig. 6. Correct responses as a function of type of Pair

and test Session.

'.__—-,,__.

Percent Correct

68

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100-
90'
7?
fir—
80- §
g E
J .
7° / § 7
a; e /
/>< /
60- / é ;
/
/ / /
/ /g /x
5°" § § / '2'???
I o S D . S D 75 D s D
V-V V-T T-V T-T
Condition . .a,. 3‘

Fig. 7. Correct responses as a function of type of Pair
and Condition.

69
than different pairs. The only significant pair differences were in
the V-T and T—V conditions, however. Same pairs were judged correctly
about as often under both conditions, but different pairs were judged
correctly, relative to same pairs, more often in the V—T condition
Qp'< .05) and less Often in the T—V condition (p_< .075).

Children in Group V-V made more correct judgements of same pairs
than V-T, T—V and T—T children (p_< .01), who did not differ from
each other. Children in Groups V—V and V—T made more correct judge—
ments of different pairs thatn Group T-T children (p.< .05—.10), who
in turn made more correct judgements than Group T-V children (p_< .05).
No explanation for these differences can be offered.

Pair and Condition also interacted with Complexity(p_< .004)

 

but scrutiny of differences between cell means in this interaction
sheds no light on the role of complexity in the pattern of results
described above. Differences between types of pairs were not signifi-
cant at any complexity level in any of the conditions, and differences
between complexity levels were not significant for either same or dif-
ferent pairs in any of the conditions.

Performance improved with age in the T-T and V—T conditions and
the significance of the interaction of Epgdg and Condition with type
of Pgip_(p_<.0005) suggests that improvement in those conditions
might have reflected improvement with one type of form pair. Only
the improvement in the T-T condition was clearly related to the type
of pair: identification of same pairs improved between kindergarten
and fifth—grade (p_< .05). In the other conditions (V-T and also V-V
and T-V) the improvement in identification of same pairs with age was

small and not significant. In pg_condition did the identification of

70
different pairs change with age. Other differences, as between pairs
in any age in any condition, were not significant.

B. Analyses of children's exploratory behavior. Exploratory behavior

 

was recorded to permit analysis of the relation between exploration
and correct identification of form pairs. Two kinds of data were
collected. The first consisted of how long the children either looked
at or felt the forms; the second consisted of characterizations of
specific aspects of their actual hand movements.

1. Analysis of duration of visual and haptic exploration and the

 

relation between duration and the number of correct judggments.

 

Durations of exploration were recorded for all children in all con-
ditions on all trials, so it was possible to analyze the variation in
these scores by means of the same model analysis-ofdvariance used in
the analysis of the number of correct judgements. Two seven—factor
analyses of variance were performed, one for standard and one for com-
parison exploration. For analyses of sub-effects, Scheffe's method
again was used. The results of these analyses are summarized in

tables 2 and 3.

 

Insert Tables 2 and 3 about here

 

2. Relation between duration of exploration of standards and correct

judgements. The accuracy with which the children identified form

 

pairs ought to be related to the adequacy of their exploratory behav-
ior such that the same variables ought to be associated with both

correct responses and exploration time. A first step toward showing

71

TABLE 2
Partial Summary of Analysis of Variance of Exploration—time

for Standard Forms

Source if; MS _F_‘_ p

Between

Grade (C) 2 15941.420 5.172 .008

Condition (C) 3 99555.994 32.302 .0005
G X C 6 4581.799 1.487 .10
Error 72 3081.986

3211.12.

Complexity (L) 7 8190.649 139.740 .0005
G X L 14 493.111 8.413 .0005
C X L 21 1204.619 20.552 .0005
G X C X L 42 105.163 1.794 .001

Pair (P) 1 50.278 0.858 .10
L X P 7 598.962 10.219 .0005
C X L X P 21 144.906 2.472 .0005

Sessions (S) 1 19952.766 340.413 .0005
L X S 7 409.086 6.979 .0005
L X P X S 7 938.493 16.012 .0005
Error 2232 58.613
Total Error 4607 ll37295.268

72

TABLE 3

Partial Summary of Analysis of Variance of Exploration-time

Source

Between

Grade (G)

Condition (C)
G X C

Error

Within
Complexity (L)

G X L

C X L

G X C X L
Pair (P)

L X P

G X C X P
Sessions (S)

C X S

L X S

G X C X S

L X P X S

Error

Total

for Comparison Forms

df

72

14
21

42

2232

4607

MS

3368.767
64449.884
2477.357

1219.433

3025.508
95.827
736.352
68.284
217.303
163.023
34.850
3939.506
616.317
45.213
112.230
108.273
45.630

629238.904

I“!

2.644
50.600

1.945

66.306
2.100
16.138
1.496
4.762
3.357
0.764
86.337
13.507
0.991
2.460

2.373

.078
.0005

.085

.0005
.010
.0005
.021
.028
.001
1.0

.0005
.0005
.10
.023

.020

73
a link between exploration and correct responses consists of looking
for concommitant variation between them.

a. Effects of Grade and Condition. A glance at the exploration time

 

scores for standard forms reveals wide variation in exploratory be—
havior. Analysis of variance indicated many significant effects and
interactions and, of these, the effects of Egpdg_(p_< .0008), 9227
dition (p.< .0005), and their interaction are of primary interest.
The length of exploration time increased from kindergarten to second
grade to fifth grade, with only the kindergarteners and fifth graders
different from one another (Eggdg_effect, p_< .OIL

At this point, the results suggest some relation between ex-
ploration time.and response accuracy, since fifth-graders also made
more correct identifications than kindergarteners, but the particular—
ities of this relation will be explored later.

As would be expected, the main contribution to the Condition
effect was the difference between the two means of exploring the
standards: children who looked at the standards (V-V and V-T con-
ditions) looked at them for a significantly shorter time (p_< .01)
than did children who felt the standards (T—V and T—T conditions).

The children were told how the forms would be presented, and
every pair of forms, including a practice pair, was presented in the
same way. Therefore it is reasonable to ask whether the children
varied their exploration times for standard forms depending on whether
they would subsequently explore the comparisons with the same or with
a different perceptual system.

Children who looked at the standards tended to look longer when

they would subsequently feel than when they would subsequently look at

74

the comparisons Omean looking times: 5.93 seconds for V—V subjects,
10.38 for V—T subjects). Children who felt the standards, however,
showed an even greater difference in their standard-exploration times
depending on whether they subsequently looked at or felt the compar-
isons: standards were explored longer in the T-V than in the T-T con-
dition (means: 26.62 and 19.74, respectively, p_< .05).

The V—T condition was more difficult than the V-V condition,
and the T—V condition was more difficult than the T-T condition, as
shown by the number of correct judgements made by children in those
conditions. Therefore, it appears that children adjusted their ex-
ploratory behavior according to the difficulty of the task. Although
older children might be expected to be more sensitive to and make
more of an adjustment to a difficult task than younger children, such
was not the case: older and younger children's exploration times were

not different in any of the conditions (Grade X Condition interaction,

 

E < .10).

b. Effects of form Complexiyy. It was expected that children's ex-

 

ploration times would increase directly as form complexity increased
and that older children would be more influenced by form complexity
than younger children. These expectations were supported by the re-

sults. The significant Complexity effect (p_< .0005), presented in

 

Figure 8, shows that more complex forms were explored longer than less

 

Insert Figure 8 about here

.-
fir—

complex forms; the differences between successive complexity levels

were approximately equal. In other words, the constant rate of

Time in Seconds

75

 

50-

4%

30-

20..

1014

 

 

 

I
5‘ 8 10 1h 20 28 36 ha
Complexity

Fig. 8. Standard exploration time as a function of form
Complexity.

""“""'"“'3 3n -- -

76
increase in perimeter of the forms across complexity levels was re-
lated to a constant rate of increase in children's standard-exploratix:
time. This effect depended also on the age of the child (Complexity

X Grade interaction, p_< .0005) as can be seen in Figure 9. In

 

,— ‘1

Insert Figure 9 about here

 

general, there were no differences between grade levels in explora-
tion time for forms of low complexity but, for the more complex forms,
the older children increased their exploration time relatively more
than did the younger children.

Specifically, children at all grade levels explored 5-, 8-,
and lO-sided forms about equally. When given 36— and 48-sided forms,
however, fifth—graders increased their exploration times significantly
more than did kindergarteners (p_< .01). Second—graders showed an
intermediate rate of increase. Fifth—graders explored 20-, 28-, 36-,
and 48-sided forms longer than kindergarteners (p_< .05—.01) while
second-graders' exploration times were in between. Finally, kinder—
garteners explored the two most complex forms more than the two least
complex forms (p_< .05).

These findings evidence no simple relation between exploration
time and correct responses. Fifth-graders increased their exploration
time more for high complexity forms than did the kindergarteners. If
it can be assumed that the fifth-graders took just as much time to ex-
plore as they needed, one would expect that younger children took con-
siderably less time than they really needed especially to explore the

high complexity forms. Consequently, they would have made more errors

Time in Seconds

77

 

 

 

 

50-

40‘

30"

20‘4

10“

0 I_ l 1 I I I I I
5 8 . 10 1h 20 28 36 t8

Complexity

 

Fig. 9. Standard exploration as a function of form
Complexity and Grade.

 

78
with those forms relative to the older children. Contrary to this
prediction, there was no joint effect of age of the children and form
complexity on correct responses.
The effects of form complexity also depended on how the forms
were presented (Complexity X Condition interaction, p_< .005) as shown
in Figure 10. Children in conditions T—V and T—T explored the more

Insert Figure 10 about here

complex standards longer than the less complex ones, but children
in the V-V and V—T conditions did not. Specifically, children in the
visual conditions looked at the standard forms for a shorter time than
children in the haptic conditions felt them, and this difference held
at all levels of form complexity (p'< .01). Also, the difference
between visual and haptic exploration increased with increasing com-
plexity; the difference between the visual and haptic conditions was
significantly greater for the average of the four high complexity
levels than the average of the four low complexity levels (p_< .01).
This difference between the visual and haptic standard—exploration
times, especially for the more complex forms, was greater for older
than for younger children (Complexity X Condition X £5235 interaction,

p.< .001). This interaction is presented in Figure 11. The extent

 

Insert Figure 11 about here

 

f F'—

to which children's feeling times exceeded their looking-times was

about the same at all grade levels for the low complexity forms. For

Time in Seconds

79

 

504

40‘

3° ‘

20 ‘

10“

 

 

 

 

I l l I j
s- 8 . 10 14 20 28 36 he
Complexity _

.. ._. _ 4. -—_. __. -e.__.__.._.___—.

Fig. 10. Standard exploration time as a function of form
Complexity and Condition.

79A

Fig. 11. Standard exploration time as a function of form Complexity,
Condition, and Grade

..a

80

.1?! I if I.

 

m coupe N coupe M ou-hu
mm on .. 8 cm i: S o m 3 on mm cm i S m m m: om mm om l: S m m
. . _ p . . . . . . . . _ . . . _ — p p . _ . . . . 0
g
Ql101\\o.llo.\..o.

 
 

 

 

 

 

 

 

spuooos u: omrm

81
high complexity forms, the extent to which feeling—times exceeded
looking—times increased with the children's grade level although not
significantly. However, this differential rate of increase as a func-
tion of age is reflected in the greater number of significant differ-
ences between looking- and feeling—times for more complex forms among
older relative to younger children. For fifth-graders, the differ-
ences between the looking and feeling conditions were significant for
each of the four most complex forms (p_< .01); for second-graders, the
differences were significant for three of the four most complex forms
(2.< .05-.01); for kindergarteners, none of these contrasts was signi-
ficant. Visual and haptic exploration times for the three least com-
plex forms were not significantly different at any grade level.
c. Differences between visual and haptic exploration. The differences
between visual and haptic exploration as a function of form complexity
and grade level (Figure 11) may have resulted from changes in visual
or haptic exploration, or both. Consequently, changes with age and
complexity in visual and haptic exploration times must be looked at
separately. In the visual standard—exploration conditions (V—V and
V-T), complex forms were looked at only slightly longer than simple
(low complexity) forms, and the differences were not significant across
age or at any age alone. In addition, children at all ages looked at
the forms at each complexity level for approximately the same length
of time. Finally, there were no significant differences within or
between the V-V and V—T conditions associated with form complexity or
subjects‘ grade level. In other words, the children, young and old
valike, who looked at the standard forms looked at them for approximately

an equivalent length of time whether the forms were simple or complex.

 

 

82

This finding is particularly interesting because past research
with similar forms suggests that visual exploratory behavior varies
as a function of the complexity (perimeter) of the forms. But in that
research children either were asked to verbally express their prefer—
ences for forms or were allowed to look at forms for as long as they
liked in free—looking tasks. The present results show very clearly
that the complexity (perimeter) of random shapes is not a factor in
visual exploration time when the subject's task is to identify the
form.

0n the basis of these data for the visual standard—exploration
conditions, it appears that the age and complexity differences between
visual and haptic exploration are due to variation among subjects in
the haptic conditions (T-T and T—V) alone. Comparisons of means with-
in the haptic conditions support this conclusion. Children for whom
the standards were presented haptically felt the four most complex
forms significantly longer than they felt the four least complex forms
(p_< .01). Furthermore, group T-V children showed a greater increase
in exploration time as complexity increased than did T-T children.
Differences between the two conditions were significant for the five
highest but not the three lowest complexity levels (p_< .050-.Ol).

Group T-V children felt the four most complex forms longer than

the four least complex forms (p_< .01, Condition X Complexity inter-

 

action), especially the fifth—graders (Condition X Complexity X
§53g3_interaction). For kindergarteners, there were no differences

in feeling time across complexity levels, while for second— and fifth-
graders at least one high complexity form was explored longer than a

low complexity form (p_< .01). The difference in feeling time between

83
a low (S-sided) and a high complexity (48—sided) form was signifi—
cantly greater for fifth—graders than for kindergarteners (p_< .075).
The second-graders were in between and not significantly different
from either age group.
For the Group T-T children the four most complex forms were felt

longer than the four least complex forms (p_< .01, Condition X Com-

 

plexity interaction) but there was only a suggestion of age differences

in this effect (Condition X Complexity X Grade interaction). Fifth-

 

and second—graders did feel one of the high complexity forms longer
than one of the low complexity forms (p_< .05-.01), while kinder-
garteners did not.

To summarize, the time children looked at the standards was not

 

related to their age (grade), test condition, form complexity, or
whether the comparison forms subsequently presented were looked at or
felt. By contrast, children who felt the standards felt high com-
plexity forms longer than low complexity forms, more so in the T—V
than in the T—T condition. Further, this difference was more marked
among older than among younger children. Put another way, older
children varied their haptic exploration of the standard forms more
than younger children, and this variation depended upon the complexity
of the forms and the perceptual system in which comparison forms were
subsequently presented.

Finally, none of these age differences in exploration—time seem
to be related to age differences in correct judgements. Although grade
differences in exploration—time interacted with form complexity and
condition together, grade did not interact with condition alone.

Variation in correct judgements was related to grade and condition

84
alone but not in conjunction with form complexity. In view of these
inconsistencies between judgement and exploration-times, it seems un—
likely that exploration—time differences could have produced the
variation in the number of correct responses.
d. Effects of test session. It might be expected that children's
exploration would decrease with practice as they become more effic-
ient in the task. The results are consistent with this hypothesis:
children's exploration thme was less in Session II than in Session I
(Session effect, p_< .0005). The decrease was greater for high than
for low complexity forms (Session X Complexity interaction p_< .0005,

shown in figure 12). Forms at all complexity levels tended to be

 

Insert Figure 12 about here

 

explored less in Session II than in Session I although differences
between sessions were significant only for the five highest complexity
levels (2 < .01).

The results for correct responses show that children made more
correct identifications in Session II than in Session I, mainly it
seems, because their ability to identify more complex forms had im—
proved. Therefore, improved performance is associated with a decrease
in exploration time.

e. Effects of type of form Pair. A child would not have been able
to predict what type of comparison form would follow the standard
that he was exploring. It was therefore reasonable that a comparison
form, whether the same as or different from the standard it followed,

did not affect the length of exploration of the standard (Pair effect,

Time in Seconds.

85

 

50‘

30"

10 ‘

 

 

 

 

0 r”* I T I I I I
s .10 m 20 23 36 he

,,.. , . A ‘ ._.....‘-...¥xs...‘,

m-J

Fig. 12. Standard exploration time as a function of test
Session and form Complexity.

86
p_< .36). It was surprising, however, that type of Fair interacted

with Complexity QE_< .0005), Complexity and Session (p_< .0005) and

 

Complexity and Condition (p_< .0005). Examination of cell means in

 

these interactions revealed no consistent effect of type of Fair. It
is possible that standard exploration time might have been influenced
by the type of pair presented on the preceding trial, but the data
have not been examined to test this hypothesis.

3. Relation between duration of exploration of comparisons and

correct judgements. Many factors might determine children's ex-
ploration time for comparison forms but the most important is likely

to be whether or not the comparison is the same at the standard.

When forms are different, detecting any difference at all between them
is sufficient for a judgement to be made; but when they are the same,
the subject must be certain that they are the same in every respect.
Therefore, when comparison forms are the same as the standards, ex—
ploration should be of longer duration than when they are not. Com-
parison exploration ought also to vary as a function of whether the
same perceptual system was used to explore the standard: Comparisons
should be explored longer when children must use a different system
than when they use the same system as they used to explore the stand-
ards.

a. Effects of Grade and Conditon. Since comparison exploration should
be more thorough for same than for different pairs, exploration of same
comparisons ought to be more nearly like standard exploration than dif—
ferent comparisons. Therefore, comparison exploration times ought to
have varied like the standards, as a function of Grade and Condition

for same pairs but not necessarily for different pairs. Type of Fair

87
did not interact with.§£gdg and Condition however; variation in ex—
ploration as a function of Eggdg_and Condition apparently was the same
for both types of form pairs. There were essentially no age differ-
ences in comparison exploration times: The effect of Egad§_was only
marginally significant (p_< .078), and differences between grade
levels were not significant. Cell analyses within the significant
Condition effect (p_< .0005) showed that the comparison forms were
felt longer than they were looked at (T—T and V-T times exceeded V—V
and T-V times, p_< .01). Other differences between conditions were
small and not significant.

The differences between looking and feeling times for the com-
parisons were only marginally related to the ages of the children
(Condition X §52d3_interaction, p_< .085). The only significant dif-
ferences were between visual and haptic exploration of the forms:
haptic exceeded visual exploration times among fifth— and second-
graders QB.< .05) but not among kindergarteners. Finally, comparisons
were explored only slightly longer in the inter—system than in the
intra-system conditions and these differences were not significant.

In view of the marginal significance of these findings and the
fact that the pattern of results does not correspond to the pattern
of results for correct responses, there seems to be but weak evidence
to suggest that variation in exploration time of comparison forms as
a function of grade and condition produced the observed variation in
correct judgements.

b. Effects of form Complexity. Comparison exploration should have
been influenced by form complexity in much the same way as was stand—

ard exploration: Older children ought to have exceeded younger

 

 

88
children in their exploration of high complexity forms. Also variation
in form complexity ought to have had greater effects on haptic than
on visual exploration. The results showed that exploration time in—

creased as complexity increased (Cgpplexity effect, p_< .0005) and,

 

although differences between successive complexity levels were not
significant, the four most complex forms were explored longer than the
four least complex forms (p_< .01).

Generally, children of all grade levels explored the more com-
plex comparisons longer than the less complex ones; older children,
second- and fifth-graders, explored comparisons longer than kinder—

garteners, and did so more as complexity increased (Grade X Complexity

 

interaction, p_< .01), shown in Figure 13). Exploration times for

 

Insert Figure 13 about here

 

— .—

second- and fifth—graders were slightly but not significantly higher
than kindergarteners for the least complex forms and, as complexity
increased exploration times increased, more markedly for the older
children than for the younger ones. Fifth- and second—graders both
explored 48-sided forms longer than kindergarteners (p_< .05—.01).

As expected, the effects of form complexity on exploration were
produced when comparisons were explored haptically but not when they
were explored visually (Complexity X Condition interaction, p_< .005)

as shown in Figure 14. There were no differences within either visual

 

fiv v— v ‘— v v— v .—

Insert Figure 14 about here

 

Time in Seconds

89

 

 

 

 

 

50 4
140 " D—--—O 5'
H z,
O----<D K
30 ‘
20 “
10 T
O I I I ‘T I I I I
5 8 10 1h 20 28 36 he

"‘" ' - 7 _— O. —_—4 _ n— A"- __ _._ 7‘. 7

Fig. 13. Comparison form exploration time as a function
of Grade and Complexity.

in Seconds

Time

90

 

 

 

 

 

 

504
ho-I
30“
20‘
104
0 l l l V l I I 1*
5 ~ 8 10 1h 20 28 36 he
Complexity

Fig. 14. Comparison form exploration time as a function
of Complexity and Condition.

91
comparison-exploration conditions(V—V and T~V) across complexity
levels nor were there differences between these condition at any com-

plexity level. This was true regardless of age (Complexity X Con—

 

dition X Grade interaction, p_< .021) as can be seen in Figure 15.

 

f‘f wv— v—

Insert Figure 15 about here

 

Comparison forms were generally felt longer than they were looked at
across all complexity levels but significant differences were found
only for the most complex forms and only among second- and fifth-
graders (p_< .05-.01); differences between visual and haptic ex-
ploration were not significant for the four lowest complexity levels
at any grade level.

Within the haptic comparison—exploration conditions, children
increased their exploration times as form complexity increased. The
four most complex forms were felt longer than the four least complex
forms (p_< .01). This was a joint effect of the three grade levels;
within any single grade level, only the difference between S—sided
and 48-sided forms was significant and only for second-graders
(p <.075).

It seems reasonable that, when comparisons are explored with
the same system as the standards, exploration times should be less
than when they are explored with a different system. It is surprising,
therefore, that exploration times for comparison forms in inter-system
and intra-system conditions were not different at any grade or com-

plexity level.

v' I

Fig. 15. Comparison form exploration time as a function of Complexity,
Condition, and Grade level.

92

a: on as am :m

m ooeao I .

.Swm

.7... I. III. . .

N cache

3 a a a a, a

m: on mm om as
— — _ _ n

m 398

 

 

 

 

 

 

 

. OH

ON

on

3

spuooes up ompm

93

The relation between comparison exploration.time and correct
responses seems especially difficult to predict: Exploration of com—
parisons that is of short duration may yield inadequate information
and lead to incorrect responses. On the other hand, exploration that
is of extended duration may mean that the subject cannot find those
parts of the form he is looking for or it may mean that he has for—
gotten the standard and is looking for any part of the comparison that
reminds him of it. Thus long duration exploratory behavior may also
be associated with incorrect responses. Therefore, it is unclear as
to what relation, it any, should exist between exploration time and
correct responses for comparison forms.

The results for exploration of comparisons and for correct

responses do not suggest that the two are related. Grade, Condition,
and Complexity interacted to determine exploration time but complexity
was unrelated to correct judgements. These inconsistencies suggest
that, as with the standards, there is no direct relation between com—
parison exploration time.and response accuracy.
c. Effects of test session. As subjects' experience in the task in-
creased, their exploration times decreased for the comparison just as
was found for the standards (Session effect, p_< .0005). Most of the
reduction in exploration time appears to have occurred within the

haptic comparison presentation conditions (Session X Grade X Condition

 

 

interaction, p_< .023) but the few cell differences within this inter-
action preclude a clear interpretation of its meaning.

d. Effects of type of form Pair. It was expected that comparison
forms would be explored for less time when they were different from

standards than when they were the same and this expectation was

94
supported by the significant §§i£_effect (p.< .028). It was further
expected that more complex "same" comparisons should require more
lengthy exploration than less complex ones. Consistent with this ex-

pectation the Complexity X Pair interaction, shown in Figure 16 was

 

 

Insert Figure 16 about here

 

significant (p_< .001) though the form-complexity hypothesis cannot
be unequivocably supported inasmuch as both same and different pairs
were affected equally by the complexity variable: Exploration times
for the four least complex forms were significantly less than for the
four most complex forms for both same and different pairs (p.< .01).
At no complexity level was the difference between same and different
pairs significant, nor was there any consistent tendency at any com-
plexity level for one type of pair to have been explored longer than

the other. Type of Pair also interacted with Complexity level and

 

Session (p_< .02). This interaction seems to be of no consequence
and differences between.means within this interaction were not signi—
ficant.

In summary, there is little evidence here of a relation between
length of exploration of the comparison forms and accuracy of judge—
ment. Type of Pair, Grade, and Condition all interact to affect judge-
ment accuracy, but they do not interact to influence comparison-
exploration times. On the other hand, both Pair type and Complexity
are related to exploration times and to accuracy of judgements but not

in the same way. There appears to be no direct correspondence, then

Time in Seconds

95

 

25-

201

15‘

10 S

 

 

 

 

0 I r’ I I I I I l
5‘ 8 10 11+ 20 28 36 [+8

Complexity - "

Fig. 16. Comparison form exploration time as a function
of Complexity and type of Pair.

96
between the number of correct responses and the length of exploration
of the comparison forms.

4. Within—subject variation in standard and comparison exploration

 

times and their relation to correct responses. Although there ap—

 

parently is no evidence of a consistent relation between exploration
time for standards or comparisons and incidence of correct judgements

by groups of children in the various conditions, it seemed possible

that this relation would be found within individual subjects' scores.
Since the data were not in a form suitable for computer analysis, a
preliminary test of this relation was made with a stratified random
sample of 24 of the 144 children, two from each condition at each grade
level. Each child's average exploration time leading to correct respon-
ses was computed and compared with his average time leading to incorrect
responses for both the standard and the comparison forms.

With respect to the standards, correct response times exceeded
incorrect response times for 12 of the children while, for the other
12, the opposite was true and with approximately equal differences in
both cases. Analyses by age and condition were equally inconclusive.

For the comparison forms, similar results obtained: 12 children
were found to explore comparisons longer prior to making a correct
response while 12 children explored them longer prior to making an in-
correct response. There was no indication in these data of a relation
for individual children, between length of exploration and performance.

It was suggested earlier that the relation between exploration
time and accuracy of judgement may not be linear but instead may be
curvilinear. That is, if the comparison form is explored for too short

a time, the children.may be judging on the basis of too little

97
information; and if exploration of the comparison is too long, they
might forget the standard because too much time has elapsed since they
last examined it. Therefore the gapgg_of variation in exploration
times (as measured by their standard deviation) should be greater
prior to incorrect than prior to correct responses.

If anything, just the opposite seems to have been the case:
exploration times were more variable prior to correct judgements than
prior to incorrect judgements for 14 children, less variable for the
remaining 10.

In summary, there is no evidence for a relation between explora-

 

tion time and correct responses within individual subjects for either
standard or comparison forms.

5. Analyses of video-tape records of children's haptic exploratory
behavior and the relation between exploration and correct judgements.
Haptic exploration was recorded on video—tape for the first 16 trials
for all children (excepting those in the V—V condition, of course).
After preliminary study of these records it seemed that children's
exploration of the perimeter of the forms might change with age and be
related to improvement with age in the number of correct judgements
they made. Since the distinguishing characteristics of the forms used
in this study were located at the perimeter of the forms a child's
judgement accuracy ought to be related to his exploration of the
perimeter. Specifically, the more time children spent feellngEhe
perimeter of forms relative to the total time their hands were in
contact with the forms, the better their performance ought to have
been. Similarly, the morepperimeter the children explored relative

to the total perimeter comprising the forms, the better their

98
performance ought to have been. The children's behavior was scored
for both these measures.

To ensure reliable scoring of the children's hand movements, two
raters viewed and scored the records of twelve children, representing
all ages and conditions, for sixteen trials per child. Inter-observer
determinations of the children's total exploration time, total time
on the perimeter, and total distance along the perimeter that was
explored, were correlated.

The inter-observer Pearson Product-moment correlations were
r a .99 for total time, r - .98 for time on the edge, and r t .99 for
distance on the edge. These high correlations reflect the fact that
each child's tape record was played several times at normal speed and
frame-by-frame which tended to eliminate chance variation between
the observers. A single observer subsequently scored the records of
all other subjects in the manner just described.

Approximately 200 viewing hours were required to score the video
tapes; consequently, several months lapsed between the time at which
the inter-observer reliability estimates were made and the time at
which the last subject's data were scored. Therefore an estimate of
intra-observer reliability over this time and practice span was com-
puted. The records for six of the twelve subjects for whom inter-
observer reliability had been established were re-scored and yielded
intra—observer reliability correlations of r - .99 for total time,
and time on the edge, and for distance explored along the edge.

Four analyses of variance were performed on the scores obtained
from the video-tape records—«two for the exploration scores with the

standard forms, two for the comparison forms. One analysis for the

4'

 

 

 

99
standard and one for the comparison forms pertained to the percent of
the total exploration time the children spent on the perimeter; the
other analyses for the standards and comparisons pertained to the per—
cent of the total perimeter comprising the forms that the children
actually explored. These analyses are summarized in Tables 4, 5, 6,

and 7. Scores were analyzed as a function of the children's Grade

 

Insert Tables 4, 5, 6, and 7 about here

 

level, presentation Condition (T—V and T-T for standards, V-T and T—T

for comparisons), type of Pair (same or different), and Complexity

 

level (the lowest four levels vs. the highest four levels). Scheffe's
method was used for analyses of sub-effects.

a. Exploration time on the edge of standard forms. It was expected

 

that younger children would spend a smaller proportion of their ex-
ploration time in contact with the perimeter of the standard forms
than older children. Further, any such age difference in exploration
should be especially apparent with the high complexity forms, since
younger children should have more difficulty processing the infor-
mation in high relative to low complexity forms than older children.
These expectations were borne out by the results as can be seen in

Figure 17. The proportion of time children spent exploring the perimeter

 

Insert Figure 17 about here

 

v—‘w— W W hv V— . V—

of the forms increased with age (Grade level effect, p_< .01). Kinder—

garteners spent a lower prOportion of their exploration time feeling

.7

 

Summary of Analysis of Variance of Percent of Exploration—time

Source

Between
Condition (C)
Grade (G)

C X G

Error

Within
Pair (P)
C X P
G X P
C X G X P
Error X P
Complexity (L)
C X L
G X L
C X G X L
Error X L
P X L
C X P X L
G X P X L
C X G X P X L

Error X P X L

on the Perimeter of Standard Forms

df

66

66

100

TABLE 4

MS

432.153
5083.625
415.762

515.880

92.373
0.186
37.253
6.390
18.657
1403.464
172.691
210.018
90.416
72.538
1.386
19.799
1.754
28.394

18.836

I'Ti

0.838
9.854

0.806

4.951
0.010
1.978

0.342

19.348

2.381
2.895

1.246

0.074
1.051
0.093

1.507

.05

.01

.075

101
TABLE 5

Summary of Analysis of Variance of Percent of

Perimeter Explored of Standard Forms

Source <_1_f_ ES _F_‘

Between

Condition (C) 1 3714.338 3.827

Grade (G) 2 7796.375 8.033
C X G 2 521.711 0.538
Error 66 970.588

11221.2.

Pair (P) 1 350.849 13.029
C X P 1 10.276
G X P 2 66.683 2.476
C X G X P 2 72.016 0.382
Error X P 66 26.927

Complexity (L) 1 26772.379 91.670
0 X L 1 2.590 756.336
c'x L 2 1868.154 6.397
C X G X L 2 121.127 0.415
Error X L 66 292.053
P X L 1 0.496 0.012
C X P X L 1 61.978 1.457
G X P X L 2 28.565 0.672
C X G X P X L 2 1.217 0.029
Error X P X L 66 42.539

.01

.01

102

TABLE 6

Summary of Analysis of Variance of Percent of Exploration-time

Source

Between
Condition (C)
Grade (G)

C X G

Error

Within
Pair (P)
C X P
G X P
C X G X P
Error X P
Complexity (L)
C X L
G X L
C X G X L
Error X L
P X L
C X P X L
G X P X L
C X G X P X L

Error X P X L

on the Perimeter of Comparison Forms

df

66

66

MS

100.250
4707.414
115.143

558.687

1.254
9.374
10.822
0.186
18.147
2864.791
18.832
499.944
47.533
72.273
57.588
50.977
26.136
89.320

17.139

1".

0.179
8.426

0.206

0.069
0.516
0.927

0.010

39.638

0.260
6.917

0.658

3.360
2.974
1.525

5.212

 

I

 

103

TABLE 7

Summary of Analysis of Variance of Percent of

Perimeter Explored of Comparison Forms

Source d_f_ gig _F_

Between

Condition (C) 1 385.944 0.255

Grade (C) 2 7622.613 5.029
C X G 2 394.146 0.260
Error 66 1515.849

lama

Pair (P) 1 92.324 0.980
C X P 1 5.014 0.053
G X P 2 58.980 0.626
C X G X P 2 20.469 0.217
Error X P 66 94.160

Complexity (L) 1 58427.117 189.328
C X L 1 308.465 1.000
G X L 2 288.141 0.934
C X G X L 2 655.059 2.123
Error X L 66 303.603
P X L 1 63.402 0.809
C X P X L 1 58.287 0.743
G X P X L 2 260.312 3.320
C X G X P X L 2 138.974 1.772
Error X P X L 66 78.407

.01

.01

Percent Time

104

 

100'“

 

90'

 

80"?

707

60“

 

50 i

.f'

F r I

x 2 5
Grade

Fig. 17. Exploration time on the edge of Standard forms
as a function of Grade and Complexity.

 

105
the edge of the forms than did second—graders (p_< .01) and fifth—
graders (p'< .01) who did not differ from each other.
Children at all grade levels spent proportionately more time
exploring the perimeter of low than high complexity forms (Complexity
effect, p_< .01) and this difference became less pronounced as age in-

creased (Grade X Complexity interaction, p_< .075, figure 17), although

 

this is apparently due to a ceiling effect. Kindergarteners spent
proportionately less time exploring the perimeter of high than low
complexity forms (p_< .01) but second and fifth—graders explored both
about equally. Second- and fifth-graders spent a greater proportion
of their time feeling the perimeter of all forms than kindergarteners
(p_< .01) but the increase with age between the kindergarteners and
the second— and fifth-graders was greater for high than for low com-
plexity forms (p_< .10).

Inexplicably, the type of Fair also affected the proportion
of time spent exploring perimeter (p_< .05). »Children spent a
slightly greater proportion of time (1.1%) exploring the perimeter of
standards among different than.among same pairs. Since subjects did
not know whether a same or different comparison would follow the
standard they were exploring it is possible that their exploratory
behavior was somehow affected by the preceding trial.
b. Proportion of total perimeter explored of standard forms. It was
expected that the proportion of perimeter explored would vary with
children's age and form complexity in the same way as did perimeter
exploration time. Younger children should have more difficulty pro—
cessing form information than older children and, consequently,

should explore less perimeter. Further, this age difference should

106

be accentuated for high relative to low complexity forms. Figure 18

 

Insert Figure 18 about here

 

shows that the results are consistent with these expectations. Kin-
dergarteners covered less of the forms' perimeter than did the
second-graders (p_< .01) or fifth—graders (p_< .01) who did not dif-
fer from each other (9E2d2_effect, p_< .01). Furthermore, the peri-
meters of the.more complex forms were explored less completely than

were the perimeters of the less complex forms (Complexity effect,

 

2.< .01).

Finally, children at all ages explored a smaller proportion
of the perimeter of high complexity than low complexity fonms Qp <
.05-.01), and for kindergarteners this difference was larger than

for either second— or fifth-graders (p.< .01) (Grade X Complexity

 

interaction, p_< .01, Figure 18). For low complexity forms there
were no age differences in proportion of perimeter explored. For high
complexity forms, kindergarteners explored a smaller proportion of

the perimeter than did second- or fifthegraders (p_< .Ol).

Paradoxically, again the perimeters of standards comprising
different pairs were explored slightly more than the perimeters of
standards comprising same pairs (85.8% vs 83.6%, p_< .01).

To summarize, the older children spent a greater proportion of
their time feeling the perimeter of the standard forms and explored
a greater proportion of the available perimeter than did the younger
children. The older children, also, were more accurate. Whether

these findings imply a functional tie between exploration and

Percent Perimeter

107

 

 

 

 

 

100 "
fie LOW
90'1
,FF’ ------- -o PUSH
80 4 _ /
/
/
/
/
/
70" /
/
/
/
/
/

60 ~ d

50 “f

-O I I F l '

x 2 5
Grade

Fig. 18. Exploration of perimeter of Standard forms as
a function of Grade and Complexity.

 

 

108
judgements is still problematic, since the relation holds for these
but not other variables: Perimeter exploration varied as a joint

function of Grade and Complexity while correct judgements did not.

 

c. Exploration time on the edge of comparison forms. Although
correct responses appear to be unrelated to tgtgl_exploration time
for comparison forms they might nonetheless be related to the pro-
portion of exploration time on the perimeter. The more time a child
spends on the perimeter of comparisons relative to time spent in the
center areas of forms, the more time he spends obtaining relevant
information and the better his performance ought to be. Older
children ought to spend a greater proportion of their exploration
time obtaining relevant information than younger children and, con-
sequently, age differences in prOportion of time on the perimeter
were expected.

Information pertaining to haptic exploration of the perimeter
of comparison forms was obtained from children in the V-T and T-T
conditions. The results were as expected: Generally, older children
spent a greater proportion of their exploration time on the perimeter
of the comparison forms than younger children (Eggdg_effect, p_< .01).
Kindergarteners spent less time feeling the perimeter than did
second- and fifth-graders (p_< .01) who did not differ from each
other.

It was also expected that complexity would influence perimeter
exploration times because more information must be processed for
high than for low complexity forms. The more information to be pro—
cessed, the.more tentative and the less thorough exploration ought

to be. Since older children presumably can process more information

109
than younger children, it was further expected that age differences
would be more pronounced with high complexity than with low com-
plexity forms. The results are consistent with these hypotheses as

can be seen in Figure 19. Children at all grade levels spent

 

Insert Figure 19 about here

 

prOportionately more time exploring the perimeter of low than high

complexity comparison forms (Complexity effect, p_< .01) and this

 

difference diminished as age increased (Grade X Complexity inter-

 

action, p_< .01), although this is apparently due to a ceiling effect.
Kindergarteners spent prOportionally less time exploring the peri-
'meter of high than low complexity forms (p_< .01), while second- and
fifth-graders explored both about equally. The age differences in
perimeter exploration time were not significant for low complexity
forms, but for high complexity forms kindergarteners spent a smaller
proportion of their time feeling the perimeter than did second— or
fifth—graders (p'< .01) who did not differ from each other.

The prOportion of time spent on the perimeter of comparisons
was unrelated to whether they were the same as or different from
the standards: the §§l£_effect was not significant. This finding
was not surprising since it should be equally important for the
identification of both same and different comparisons that exploration
be directed to their perimeter and not to their center areas.
d. Prpportion of total perimeter explored of comparison forms. In
contrast to preportion of exploration time on the perimeter, the

prOportion of perimeter explored ought to have depended on whether or

 

 

 

Percent Time

100

90

80

7O

50

110

 

 

 

 

 

' 4. Low
___...—0 HIGH
J
{ l l l
K 2 S
iGrqde

Fig. 19. Exploration time on the edge of Comparison forms

as a function of Grade and Complexity level.

 

111
not the comparisons were the same as the standards. If same pairs
are to be correctly identified, the comparisons ought to be as
thoroughly explored as the standards; on the other hand, this need
not be the case with different pairs since a correct judgement can
be made as soon as any difference between them is detected. The re-
sults do not support this hypothesis, however. The only effect of

Pair occurred in interaction with Grade and Complexity (p_< .05).

 

Within this interaction the proportion of perimeter felt of same and
different pairs was about the same within each Grade and Complexity
level. No other Pair differences within this interaction seem.mean-
ingful statistical significance notwithstanding.

The §52d2_effect was significant (p_< .01) as might be expected
although the effect seems to have been curvilinear rather than linear.
Second-graders explored a greater proportion of the perimeter of com—
parisons than kindergarteners (p_< .05) while the exploration of
fifth-graders was in-between and not significantly different from
either age group. Finally, the perimeters of less complex forms were
explored more completely than the perimeters of more complex forms
(Complexity effect, p_< .01) but this complexity effect was not re-
lated to the ages of the children.

e. Analyses of differences in children's exploration of standard and
comparison forms on correct versus incorrect trials. There are several
ways to look at these data on children's exploratory behavior to try

to determine whether some relation exists between exploration and
correct responses. One way involves looking for concommitant variation
in the effects of certain independent variables on both correct

responses and exploratory behavior. The results of these analyses

112
have been reported above and offer little to suggest that differences
in exploration underlie correct response differences. Another way
to explore this hypothesized relationship involves comparing child-
ren's exploratory behavior on correct trials with their behavior on in-
correct trials. This method assumes that if exploration determines
response accuracy, than a subject's exploration ought to have been
different prior to correct than prior to incorrect responses. More
specifically, children ought to have spent a greater proportion of
their time on the perimeter of the forms and explored a greater
preportion of that perimeter before making a correct response than
before an incorrect response. Differences in exploratory behavior
between correct and incorrect response trials were represented by
difference scores and subjected to analysis of variance.

To arrive at the difference scores, the proportions of ex-
ploration time on the perimeter of forms were grouped, for each
subject, according to the complexity of the forms (lowest four and
highest four levels) and the type of form pair (same or different).
The mean proportion of time associated with correct judgements was
compared with the time associated with incorrect judgements. In
other words, the difference scores refer to the proportion of ex-
ploration time on the perimeter of correct-judgement relative to
incorrect-judgement trials as a function of complexity and type of
pair. Difference scores were similarly derived for the proportion
of perimeter the children explored.

Difference scores pertaining to haptic exploration time and
distance on the perimeter of the standards (children in the T—V and

T—T conditions) and comparisons (children in the V—T and T—T

113
conditions) were analyzed by analysis of variance. These analyses

are summarized in Tables 8, 9, 10, and 11. The significance of

 

Insert Tables 8 and 9 about here

 

differences between means within significant effects was tested by
Scheffe's method.

1. Evidence for differential exploration of standard forms. The

 

difference—score analysis of time on the perimeter of standard forms
showed significant effects Grade (p_< .01), Condition (p_< .01), and

Complexity (p_< .05), and no interactions. Generally, children

 

spent a greater proportion of their time feeling the perimeter prior
to correct than prior to incorrect responses. This difference was
greater for T—T than for T-V children, greater for second- and
fifth-graders (who didn't differ from each other) than for kinder—
garteners (p_< .05, p_< .01, respectively), and greater for low com-
plexity than for high complexity forms.

The difference-score analysis of the proportion of the perimeter
of the standards that the children felt exactly paralleled the re-
sults for time: Main effects of Condition (p_< .05), Eggdg_(p < .01),
and Complexity (p_< .05) were significant and interactions were not.
Again it is generally true that correct judgements were preceded by
more thorough exploration of the perimeter than were incorrect judge-
ments. This difference was greater for T—T than for T-V children,
greater for second— and fifth—graders (who did not differ) than for
kindergarteners (p_< .10, p_< .01, respectively) and greater for low

complexity than for high complexity forms.

Summary of Analysis of Variance of Difference-Scores

for Time Exploring the Perimeter of Standard

Source

Between
Condition (C)
Grade (G)

C X G

Error

Within
Pair (P)
C X P
G X P
C X G X P
Error X P
Complexity (L)
C X L
G X L
C X G X L
Error X L
P X L
C X P X L
G X P X L
C X G X P X L

Error X P X L

df

N

N

66’

66

114

TABLE 8

MS

14078.816

17795.141

2373.533

1927.774

54.088
6154.453
1585.628
2610.830
6799.195
7795.941

24.922
1997.440

14.422
1684.436

12.316
9102.930
1911.471
1737.959
2367.232

1

7.303
9.231

1.231

0.008
0.905
0.233

0.384

4.628
0.015
1.186

0.008

0.005
3.845
0.807

0.734

<.01

<.01

<.05

<.10

Summary of Analysis of Variance of Difference-Scores

for the Amount of Perimeter Explored of Standards

Source

Between
Condition (C)
Grade (G)

C X G

Error

Within
Pair (P)
C X P
G X P
C X G X P
Error X P
Complexity (L)
C X L
G X L
C X G X L
Error X L
P X L
C X P X L
G X P X L
C X G X P X L

Error X P X L

df

66

66

115

TABLE 9

MS

10840.504
13822.525
2211.014

2124.118

417.507
223.997
627.065
2429.441
5339.836
6065.270
498.320
2426.443
130.148
1441.725
129.106
531.800
1169.369
2545.322

2411.540

E.

5.104
6.507

1 .041

0.078
0.042
0.117

0.455

4.207
0.346
1.683

0.090

0.054
0.221
0.485

1.055

<.05

116

These analyses show only that the magnitude of exploration dif—
ferences is related to Grade, form complexity and presentation con-
dition. They do not tell us whether the exploratory behavior leading
to correct judgements is significantly different from exploration
leading to incorrect judgements. If exploration of the perimeter is
more thorough, as reflected in time and distance measures, prior to
correct than prior to incorrect responses, then mean difference
scores should be positive and significantly greater than zero.

To test for a significant departure of a mean difference score
from zero, confidence intervals were set up around zero for each
significant effect. The fact that all mean differences within all
significant effects were positive indicates that the exploration
leading to correct judgements exceeded that leading to incorrect
judgements. No mean difference within any significant effect was

glgnificantly greater than zero, however.

 

 

Insert Tables 10 and 11 about here

 

2. Evidence for differential exploration of comparison forms. A
difference—score analysis was carried out with the comparison forms
in the same manner as for the standards. The following results per—
tain to analyses-for time and distance on the perimeter of comparison
forms and can be seen to parallel the results for the standards.

The difference—score analysis of the proportion of timg_children
explored the perimeter of comparison forms showed a significant effect
0f.§EEES;(2.< .01); The proportion of time feeling the perimeter

prior to correct judgements exceeded that prior to incorrect judgements

Summary of Analysis of Variance of Difference-Scores

117

TABLE 10

for Time Exploring the Edge of Comparisons

Source

Between
Condition (C)
Grade (C)

C X G

Error

Within
Pair (P)
C X P
G X P
C X G X P
Error X P
Complexity (L)
C X L
G X L
C X G X L
Error X L
P X L
C X P X L
G X P X L
C X G X P X L

Error X P X L

df

66

66

913.

180.046
20411.969
1164.727

2620.429

7793.074
15782.656
4425.430
4845.852
4255.988
14687.254
4502.266
2064.736
378.684
1680.553
1246.477
1611.781
4480.523
242.643

2582.726

2'.

.069
7.790

0.444

1.831
3.708
1.040

1.139

8.740
2.679
1.550

0.225

0.483
0.624
1.735

0.094

<.01

<.10

<.01

Summary of Analysis of Variance of Difference-scores

for the Amount of Perimeter Explored of Comparisons

Source

Between
Condition (C)
Grade (G)

C X G

Error

Within
Pair (P)
C X P
G X P
C X G X P
Error X P
Complexity (L)
C X L
G X L
C X G X L
Error X L
P X L
C X P X L
G X P X L
C X G X P X L

Error X P X L

df

66

66

118

TABLE 11

DM

24.208
11320.613
498.822

1987.877

2695.967
13058.348
3815.193
1682.148
3239.424
24190.066
2074.473
2277.717
182.680
1384.530
40.465
5118.856
3592.258
70.266

1854.128

E

0.012
5.695

0.251

.832
4.031
1.178

0.519

17.472

1.498
1.645

0.132

0.022
2.761
1.937

0.038

<.01

<.05

<.01

119
by a significantly greater margin for second- and fifth—graders than
for kindergarteners (p_< .05 and p_< .01, respectively). The second—
and fifth-graders were not significantly different from each other.

The proportion of time feeling the edge prior to correct re-
1ative to incorrect judgements was greater for low than for high
complexity forms (Complexity effect, p_< .05). In other words, dif-
ferential exploration time is a more important factor in judgements
of low than of high complexity forms.

The results of the analysis of the prOportion of the distance
along the perimeter the children felt paralleled the results for time
on the perimeter. The significant Eggdg effect (p_< .01) shows that
fifth- and second-graders felt a greater prOportion of the perimeter
of comparison forms prior to correct judgements than prior to in-
correct judgements by a significantly greater margin than kinder-
garteners Qp<< .05); second— and fifth—graders did not differ from
each other.

The extent to which the proportion of perimeters explored of
comparison forms prior to correct judgements exceeded that for in-
correct judgements was greater for low complexity than for high com-
plexity forms (Complexity effect, p_< .01). In other words, dif-
ferential exploration of the amount of perimeter, like length of ex-
ploration time, is a more important factor in judgements of low than
of high complexity forms.

The interaction of type of Pgi£_and Condition was significant
(p_< .05) but no differences between means within this interaction

were significant and no explanation of this effect can be offered.

120

Despite the statistical significance of variation in the
‘magnitude of difference scores for comparison exploration as a
function of grade and complexity, none of these mean differences
was large enough to be significantly greater than zero. In other
words, for the comparisons as for the standards, there is a clear
tendency for older children to explore forms differently prior to
correct and incorrect judgements and for exploration differences to
be associated with the identification of less complex forms but these
differences are not large.

To summarize these findings for both standard and comparison

 

forms, the adquacy of exploration, as measured by time and distance
on the perimeter, tends to be more closely related to response
accuracy for older than for younger children; younger children's per-
formance appears to be unrelated to exploratory behavior. Also, dif—
ferences in exploration tend to be more closely related to response
accuracy for simple than for complex forms: correct versus incorrect
judgements of complex forms seem not to result from differences in
exploratory behavior. Finally, differential exploration of standards
tends to be more closely related to response accuracy for children

in the T—T than in the T-V condition: factors other than differential
exploration apparently underlie variation in correct judgements in
the T-V condition. With respect to all of these findings differences
in response accuracy were not associated with large differences in
exploration but it is important to note that, in all groups, more

and not less exploration was associated with correct than with in—

correct responses e

121
f. Analysis of exploration of specific parts of standard and com-

parison forms. The analyses of exploratory behavior presented thus

 

far have pertained to measures of thoroughness of exploration. There
are other aspects of exploratory behavior that were not explicitly
examined in these analyses, however, and that deserve attention.

One aspect of exploratory behavior has to do with whether or not
children can direct their exploration to areas of comparisons that
are comparable to areas explored on the standards. Performance ought
to be better when the perimeter explored on comparison forms is the
same as that explored on the standards than when it is different.

Age differences ought also to be found here because older children
should be better than younger children at locating areas on com-
parisons that are identical to areas explored on the standards. Com-
plexity effects might be expected to Operatelmre also since it

should be more difficult to locate identical parts of complex than of
simple forms. Finally, any complexity effects that obtain ought to
be more pronounced for younger than for older children.

To test these hypotheses, the video—tape records of haptic ex-
ploration of same pairs by children in the T-T condition were scored
for the proportion of perimeter explored on the standards that was
subsequently explored on the comparisons. Thus, eight "overlap"
scores were obtained for each of the twelve subjects at each grade
level. These scores were subjected to an analysis of variance as a

function of Grade and form Complexity.

 

The results of this analysis, which.is summarized in Table 12,

 

Yi‘ V v —v_ ‘fi

Insert Table 12 about here

 

.x

Summary

Source
Between
Grade (C)

Error

Within
Complexity (L)
G X L

Error X L

122

TABLE 12

of Analysis of Variance of "Overlap"

of Perimeter Exploration

df

33

14

231

mg
2183.617

1077.246

3535.339
515.582

364.120

I'll

2.027

9.709

1.416

$.01

123
revealed a non—significant Gmmdg_effect (0.75 < p.> .10) and reflect
the fact that there was no tendency for theanount of overlap to in—
crease with.age. The Complexity effect was significant (p_< .01)
showing that there was more overlap with simple than with complex
forms but this effect was not related to grade level.

An additional analysis was conducted where, for each subject,
percent of overlap was compared on correct and incorrect trials as a
function of Grade level. This analysis is summarized in Table 13.
The results show that, in general, there was more overlapping ex—
ploration on correct than on incorrect trials and also that overlap
differences were larger for older than for younger children. While
differences between means are all in the right direction none of the
effects was significant at an acceptable level of significance.

Insert Table 13 about here

 

While it seems reasonable that performance will be lower to the
extent that subjects explore less of those areas on comparisons than
they explored on standards, it also seems reasonable that performance
should be lower when subjects explore parts of a comparison that they
did not explore on a standard. On such occasions as these, they are
picking up information that is irrelevant to the discrimination,
noise in addition to signal so—to-speak, and this pick-up of irrelevant
information might be expected to impair performance. Specifically,
it seems likely that younger children should pick up more irrelevant
information than older children. Also, more irrelevant information

ought to be obtained from high than from low complexity forms,

especially by the younger children.

124

TABLE 13

Summary of Analysis of Variance of Overlap of Perimeter

Exploration Associated with Correct Responses

Source
Between
Grade (A)

Error

Within
Correct-Incorrect (C)
A X C

Error X C

df

33

33

‘mg
890.500

703.888

1650.237
1236.985

479.350

E. 2.
1.265

2.443 .10
2.580

125
To test these hypotheses, the video—tape records of subjects
in the T—T condition who explored same pairs of forms were scored
for the proportion of comparison exploration that duplicated the ex—
ploration of the standard. In other words, the prOportion of com-
parison exploration that was relevant to the standard was recorded;
an analysis of variance was conducted on these prOportion scores as

a function of Grade and form Complexity. This analysis is summarized

 

in Table 14.

The results of this analysis showed that the Emmdg_effect was
significant (p_< .05): Older children obtained less irrelevant in-
formation than did younger children; differences between grade levels

were not large enough to be significant however. The Complexity_

 

effect (p_< .01) showed that less irrelevant information was detected
on the low than on the high complexity forms; older children were
less influenced by form complexity than were younger children who
tended to pick up more irrelevant information on high than on low

complexity forms (Grade1X.Complexity interaction, p_< .01). These

 

findings suggest that younger children detect more irrelevant in—
formation because they are unable to restrict their exploration to
those form parts which yield only relevant information. Unfortu-
nately, the acceptability of this conclusion is diminished by the
fact that, since older children explored forms, especially high
complexity ones, more thoroughly than younger children, there was
less information on the comparisons that could be irrelevant for
them. Thus the possibility that an important difference between

kindergarten and fifth—grade children lies in their relative ability

126
to detect only relevant information in a haptic discrimination

cannot be decided by these data.

 

Insert Table 14 about here

 

f __

The prOportion of comparison exploration that yielded relevant
information was compared on correct and incorrect response trials
for individual subjects. An analysis of variance for exploration as

a function of Grade level and Correct vs Incorrect responses was

 

conducted (see Table 15) and showed that less irrelevant information
was obtained prior to correct than prior to incorrect responses

QB < .05). There was also a tendency for differences in the amount
of irrelevant information detected on correct versus incorrect trials

to be greater for older than for younger children although this

 

tendency was only marginally significant (Emmdg_x Correct-Incorrect
interaction, p_< .10). Thus, younger children tend to pick.up more
irrelevant information than older children, especially with high com—
plexity forms, but differences in the amount of irrelevant information
are related to correct responses only for older and not for younger
children. In other words, fifth-graders do not pick up much irrelevant
information, but what they do pick up impairs their performance.
Kindergarteners, on the other hand, pick up much irrelevant information
but their performance is no more impaired when they pick.up more

irrelevant information than when they pick up less.

r Y
"v—‘v a fi fifi

Insert Table 15 about here

 

Summary of Analysis of Variance of the Percent

of

Source
Between
Grade (A)

Error

Within
Complexity (L)
A X L

Error X L

127

TABLE 14

Irrelevant Perimeter Exploration

9:. 11.8. :r:
2 3248.764 4.771
33 680.977
7 933.807 4.607
14 561.198 2.769

231 202.684

<.05

<.01

<.01

128

TABLE 15
Summary of Analysis of Variance of the Percent of
Irrelevant Perimeter Explored Associated

with Correct Responses

Source if Mi _1'_‘_ p
Between
Grade (A) 2 425.054 .793
Error 33 536.265
83.91112
Correct—Incorrect (C) 1 2763.943 5.434 <.05
A X C 2 1449.658 2,850

Error X C 33 508.678

129

To summarize these results pertaining to haptic exploration of

 

specific parts of forms, there appear to be important differences
between kindergarteners and fifth-graders in their exploratory be—
havior but only for the older children is variation in exploratory

behavior associated with variation in response accuracy.

DISCUSSION

The results of this research bear on two fundamental questions:
First, "What process underlies inter—system perception?" and, second,
"What factors underlie developmental change in inter-system perception
between the ages of kindergarten and fifthegrade?" Essentially three
views are held currently as to the processes underlying inter-system
perception and the present findings can be related to them. Birch
and Lefford (1963) have said that inter—system perception requires
the development of neurological "linkages" between the perceptual
systems. A second view—-"the traditional sensation-based view"-—
holds that the association of "input" from the different systems is
required for inter-system.perception.

Given these first two views and a successive discrimination
task involving vision and touch, one would expect the performance of
subjects in the counterbalanced inter—system conditions (T-V and V-T)
to be the same since the same associations or linkages are involved
in both. The present research shows that V—T performance was better
than T-V performance. To explain these findings it is necessary to
suppose that another process, such as memory, may be involved in the

task and may be involved differently in the two conditions. Specifically,

re:

de:

821:

ch.

ar:

thl

f0:

fu:

111

In

be

as

01

mo

in

3)

it

130
children may not remember things they have felt as well as they can
remember things they have looked at. This is only an hypothesis and
deserves experimental test.

Differences in performance were also found between inter-system
and intra-system conditions: V—V and T-T performance was better
than V-T and T-V performance respectively. Since these differences
arise between conditions in which the standards were explored with
the same system, they cannot be attributed to differential memory
for things looked at and felt. Instead it seems necessary to make
further assumptions about the relative strength of associations or
linkages in intra-system and inter-system tasks. These assumptions
have not as yet been outlined by the adherents of these two views.
In summary, the associationistic or linkages views of perception can
be used to explain these results only when supplemented by other
assumptions whose validity remains untested.

A third view of inter—system perception, outlined by the
Gibsons (1966, 1969) seems to handle the present findings somewhat
more easily. According to their view, a perceptual system detects
information in stimulation some of which can also be detected by other
systems. It is the existence of trans—system information that makes
inter-system perception possible. Inter-system perception thus
depends on the availability of information in stimulation that is
relevant for the systems used in a particular perceptual task and
the ability of the organism to detect that information. In the
present study, it is clear that the forms used as discriminanda pro—

vided information that was relevant to both the visual and haptic

53"

f0

Oil

in

(A

131
systems, although the subjects were better able to detect form in—
formation visually than haptically.

The ability to detect information in a perceptual task depends
on the natural sensitivity of that system to the relevant information
in stimulation and on the subject's practice at detecting that infor—
mation. There is no reason to assume equal sensitivity or equal
practice at visual and haptic detection of form information so the
fact that visual performance exceeded haptic is not surprising. The
opposite results might have obtained had targets varying in texture
and not form been used if children's natural sensitivity for and
practice at detecting variation in texture favors touch.more than
vision.

In inter-system conditions (T-V and V-T) where both systems
are used, performance ought to depend especially on the system used
to explore the standards since, in a successive discrimination task,
adequate perception of and memory for the standard is essential in
order for the subject to be able to relate the comparison to it.
Since visual perception was better than haptic in intra-system tasks
it seems likely that performance should be better in an inter-system
task when standards are explored visually than when they are explored
haptically. Therefore, it might be expected that V—T would exceed
T—V performance as was found in the present research.

The finding that performance was worse in inter-system than in
intra-system.tasks can also be explained in light of Gibson's theory.
Since trans—system information is a subset of all the information
that a perceptual system can detect it seems reasonable that trans-

system information should be harder to detect than information relevant

to ju

SpeC'
to d
is a
wen

pre

thn
f0
th

56

132
to just one system alone. Therefore, it is not surprising that V-V
and T-T performance was better than V—T and T-V performance re-
spectively. The hypothesis that trans-system information is harder
to detect than is information relevant to a single perceptual system
is also supported by the results for exploration time: Standards
were explored longer when the comparisons were to be subsequently
presented in another system than when they were not.

In summary, Gibson's theory predicts and the results show
that performance in intra—system and inter-system tasks depends on
the relative efficiency of the perceptual systems used, especially
for exploring the standards, and whether or not it is necessary for
the subject to use another system to identify the comparison as the
same as or different from the standard.

What factors underlie improvement with age in inter-system
perception between kindergarten and fifthegrade? One hypothesis
holds that associations (linkages) are built up between the sense
data deriving from each perceptual system. The findings reported
here show that developmental change tends to be different in the T-V
and V—T conditions, though identical change would be expected since
the same associations (linkages) ought to be involved in both tasks.
In conjunction with the finding that the levels of performance in
the two conditions are different, it can be concluded that developmen-
tal change in inter-system tasks is unlikely to depend in any major
way on the develOpment of associations between the systems.

A second hypothesis holds that improvement in inter—system
perception simply reflects improvement in one or both systems involved

in the task. In the present study, improvement in visual perception

was
cond
Imp:
V-T
cap
the
dif

m8!

38
de
mi

if

133

was slight and not significant so the improvement found in the V—T
condition cannot be attributed simply to better visual performance.
Improvement was significant in the TrT condition, however, and so
V—T improvement might be attributed to improvement in haptic per—
ception. Two factors militate against this conclusion: First,
the rate of improvement in these two conditions was significantly
different; second, there ought also to have been significant improve-
‘ment in the T-V task if improvement in haptic perception is pervasive.

According to Gibson's theory, one would expect improvement with
age in trans—system tasks as a function of children's practice at
detecting information in one system that is relevant to another. One
might expect different rates of development in V—T and T-V conditions
if practice at detecting information visually that is relevant to
touch is not the same as practice detecting information haptically
that is relevant to vision. Given that the levels of performance
and developmental change in the V—V and T—T conditions are different
it is not surprising that there are also differences in developmental
change in the T-V and V-T conditions. The present research does not
provide conclusive evidence bearing on the hypothesis that develOp-
mental change in inter-system tasks reflects improvement in the
ability to detect trans-system information. A transfer-of—training
study involving practice at detecting trans-system information would
be more conclusive.

Another potential basis for age differences in performance in
perceptual tasks can be found in the greater "information processing"
capacity of older relative to younger children. In the present study

it was thought that older children should perform better than younger

134
children, especially with.high.complexity forms, because of their
greater capacity for assimilating and retaining information. In con-
trast to this prediction, the results showed no relation between age
and complexity for correct responses. Although this finding may
suggest that age differences in processing bear no relation to age
differences in correct responses, it is also possible that the forms
used in this task did not require different amounts of processing
for correct identification of them despite the variation in their
complexity.

Random shapes have been used in many different studies because,
presumably, more complex shapes require more processing than do less
complex shapes. In those studies, subjects have been asked to look
at, associate to, state a preference for, and identify forms varying
in complexity. Despite their use of identical form parameters, the
results have varied depending on the particular task. It seems in-
correct, therefore, to presume that differential information process-
ing is involved in all tasks that make use of forms that vary in
perimeter. Whatever information-processing is involved in a task
depends on the task itself and not on the fact that forms varying
in complexity are used.

It seems entirely possible that the present task made little
if any demand on differences in information—processing ability among
children of the different ages tested and, as a consequence, process—
ing differences were not related to response accuracy. This possibil—
ity exists because the forms used in this study, when they were the
same, had all aspects identical, yet had no aspects in common when

they were different. The forms were not identical in some ways and

135
different in others. Therefore, it was necessary to process only
parts of forms even when "same" pairs were presented.

Further evidence that the forms used in this study placed
little demand on subjects' information processing ability can be
found in the results for correct responses as a function of whether
the form pairs were the same or different. Same pairs ought to have
demanded more information—processing than different pairs and should
have been more difficult to correctly identify especially for younger
children and especially for forms of high complexity. Although the
results show that improvement with age in correct responses was
greater for same pairs than for different pairs, different pairs
were not easier to identify than same pairs at any age level and were
not related in any consistent way f0 form complexity and age. It
seems likely, then, that the current findings do not bear on the
question of the importance of information processing in perceptual
development because differential processing apparently was not re-
quired of the subjects.

Might age—differences in performance be attributed to differ—
ences in exploratory behavior? Gibson's theory is built on the notion
that performance in a perceptual task depends on the ability of the
subject to detect the information distinguishing the targets.
Developmental change in performance is thus seen to reflect change
with age in the ability to detect information and, since information
detection is seen by Gibson to involve the active use of perceptual
systems, exploratory behavior and develOpmental change in exploratory

behavior ought also to be related to performance. Past research too,

136
notably that of Vurpillot (1968), has implicated the importance of
exploration in age differences in performance in perceptual tasks.

In the present study several measures were made of exploratory
behavior. These measures consisted of total exploration time, per-
imeter exploration time, and the proportion of perimeter explored of
the standard and comparison forms. Older children's haptic explor-
ation was generally more "thorough" than was the exploration of
younger children, especially with the more complex forms. From
these observations one would expect older children to have made
fewer errors than younger ones especially in the T-T and T-V con-
ditions and especially with the high complexity forms. Correct
responses, while associated with age and condition, were not associated
with complexity, however.

It then becomes important to consider why there are age dif-
ferences in exploration if these differences do not relate to correct
response differences. It seems possible that fifth-graders are
generally more thorough in their exploration than kindergarteners
despite the specific demands made of them by the task. Therefore,
kindergarteners may have explored forms only to a certain extent
which, because of the characteristics of the forms, was sufficient
to identify even the most complex ones. Older children may have ex—
plored forms more thoroughly perhaps as a general strategy and thus
detected more information than was really necessary for them to be
able to respond correctly.

Although.the differences between younger and older children in
the'thoroughness" with which they explore were not related to correct

response differences, other differences in their exploratory behavior

137
looked likely to be of some importance. One's impression gained
from viewing the video—tape records was that older children explored
very systematically, feeling parts that were adjacent to one another.
Younger children, on the other hand, tended to explore non-adjacent
parts in what appeared to be a very unsystematic way. Therefore, it
seemed likely that there might be age differences in the extent to
which identical parts would be explored on comparisons that were
first explored on the standards. The analyses did not bear out this
hypothesis.

Other evidence suggesting that age differences in correct
responses relates to exploration differences comes from the analysis
of the amount of irrelevant information that children detect. Un-
fortunately, the importance of the finding that kindergarteners
detect more irrelevant information than fifth-graders is diminished
by the fact that the more thorough exploration of the fifth-graders
markedly reduced the possibility that as much information deriving
from their exploration could be irrelevant. The only way to adequately
test this hypothesis would be to match subjects of different ages for
the amount of the perimeter they explore of the standards and then to
look for differences in the amount of irrelevant information they de-
tect on the comparisons. The present data are too limited in quantity
to permit this kind of analysis.

It appears, then, that age differences in correct responses are
unrelated to differences in exploration and it is tempting to draw
the larger conclusion that variation in a child's exploratory behavior
is not related to his response accuracy. These data show convincingly,

however, that this conclusion is only valid for kindergarteners and

138
not for fifth—graders. In order to offer a parsimonious, though
admittedly post—hoc, explanation of these findings it is necessary
to distinguish between information about form obtained by exploration
and information mggd,

Correct identification of forms requires not only that a subject
detect information about them but also that he be able to make use
of the information he has detected. Since response accuracy is re-
lated to exploratory behavior for older children it appears that in—
correct responses can be attributed to faulty information detection.
On the other hand, for younger children, differences in exploration
are not so important for determining whether or not they will be
correct. Instead it seems that their m§g_of the information they
detect may determine their accuracy. The ability to use information
detected may be related to, or the same as, the ability to remember
information detected. Therefore, an important difference between
the younger and older children in this study may have been their re-
lative ability to remember the information they detected about the
forms.

It seems possible that memory may have been an important part
of the present task, since the forms to be discriminated were pre-
sented successively. Therefore it is not unreasonable that differ—
ences between kindergarteners and fifth—graders in their ability to
remember form may have produced these results. To test the hypothesis
that age differences in the ability to remember the information
detected is related to age differences in correct responses the

present study‘might be replicated and the time interval between

139
successive form presentations varied. The results of the present
study make this hypothesis very much worth testing and, if such re—
sults are positive, we can then ask why it is that kindergarten-age

children can remember some forms and not others.

APPENDICES

APPENDIX A

The forms used in this research.were generated with a matrix
of 100 X 100 squares on 8 squares to-the-inch graph paper. Pairs
of numbers between 0 and 100 were taken from a table of random
numbers and used as coordinates for plotting points within the
matrix. The number of pairs drawn for any particular form corresponded
to the number of sides the form was to have with the restriction that
any two plotted points had to be at least three squares apart. The
matrix was then bisected vertically and horizontally and the point
upper most and nearest to the vertical axis in the upper right quad-
rant was taken as the starting point. A line was then drawn from
this point to the next lowest point in the quadrant, and so on through
the lower right quadrant. After the lowest point in the right lower
quadrant was reached, the next point to which a line was drawn was
the lowest point in the lower left quadrant nearest the vertical
axis. Lines were then drawn to successive points moving in an up-
ward direction through the upper left quadrant. Finally, the upper-
most point in the upper left quadrant was connected by a line to the
starting point.12

This principle was used to generate 9 forms at each of 8 com—
plexity levels (5, 8, 10, 14, 20, 28, 36, and 48 sides). The perimeter

of each form was then.measured and the mean perimeter for each com-

plexity level was calculated. The mean perimeters for the forms at

 

12If three points fell along a straight line one of them was

moved 3 squares, the point and the direction in which it was moved
being such as to maximize irregularity in the contour of the form.

140

141
the successive complexity levels were found to differ by about 25%.
Then the perimeter of the forms at each complexity level were
altered by changing the position of as few points as possible so that
all forms at each complexity level came to have the same perimeter.
Thus, the perimeter increase from level to level was 252. An ex—
ception to this rule was made for the intervals between the 8— and

10- and 10— and l4—sided forms where the increase was 12.5%.

APPENDIX B

Four blocks of form pairs were generated with each block con-
taining one pair of forms at each of the eight levels of variability.
The ordering of the eight complexity levels in each block was only
quasi-random since some constraints were thought to be necessary:

The first form pair of a block could not be a 36- or a 48-sided one
since it seemed undesirable to present what was expected to be a
‘maximally difficult pair on the first trial. Also, care was taken
that, within each block, the first four and the last four trials
should contain both low and high variability forms. Finally, care
was taken that, when blocks were placed in succession, the last
trial of one and the first of a succeeding block did not call for
presenting forms of the same level of variability.

Determination of whether a particular pair was to consist of
same or different forms was also determined quasi-randomly. The
constraints imposed here were that the first two trials in each block
must consist of one same and one different pair of forms and that
no more than three successive trials could be comprised of the same
type of form pair (same or different). Four same and four different
pairs were presented in each block of eight trials.

The trial blocks produced were these:

142

143

Block 1 Block 2 Block 3 Block 4
Var. level S-D Var. level S-D Var. level S—D Var. level S-D

20 S 10 D 8 D 14 S
36 D 48 S 20 S 28 D
14 D 8 S 14 D 36 D
10 S 20 D 36 S 8 S
28 S 5 D 10 S 20 D
8 D 28 D 48 D 48 S
48 D 36 S 5 D 5 S

5 S 14 S 28 S 10 D

The orders in which the four blocks were presented were: Order I,

Blocks 1,2,3, and 4 and, for Order II, Blocks 4,3,2, and l.

REFERENCES

REFERENCES

Birch, H. G. & Lefford, A. Intersensory development in children.
Monographs of the Society for Research in Child Development,
1963, 28, (Serial No. 89).

Blank, M., Altman, L. D., & Bridger, W. H. Cross—modal transfer of

form discrimination in preschool children. Psychonomic Science,
1968, l9, 51—52.

 

Blank, M. & Bridger, W. H. Cross~modal transfer in nursery—school
children. Journal of Comparative and Physiological Psychology,
1964, 28, 277—282.

Brown, D. R. & Owen, D. H. Metrics of visual form: Methodological
dyspepsia. Psychologloal Bulletin, 1967, 68, 243-259.

Bruner, J. 8. On perceptual readiness. Psychological Review,
1957, 64, 123—152.

 

Bryant, P. E. Comments on the design of developmental studies of
crossdmodal matching and cross—modal transfer. Cortex, 1968,
._5, 127—137.

Cashdan, S. Visual and haptic form discrimination under conditions
of successive stimulation. Journal of Experimental Poychology,
1968, 16, 215-218.

Eastman, R. Relative cross—modal transfer of a form discrimination.
Psychonomic Science, 1967, 2, 197—198.

 

Ferguson, G. A. Statistical analysis in psychology and education,
New York: McGraw-Hill, 1971.

Gaydos, H. F. Intersensory transfer in the discrimination of form.
American Journal of Psychology, 1956, 92, 107-110.

Gibson, E. Perceptual development. in Stevenson, H. W. (ed) National
Society for the Study of Education Yearbook, 1963.

Gibson, E. Percoptual learniog and development. New York: Appleton-
Century-Crofts, 1969.

Gibson, E. J., Gibson, J. J., Pick, A. D., & Osser, H. A develop—
mental study of the discrimination of lettervlike forms.
Journal of Comparative and Physiological ngchology, 1962, 22,

Gibson, E. J. The senses considered as perceptual oystems. Boston:
HoughtonvMifflin, 1966.

144

145

Hays, W. L. Statistics for Poychologists. New York: Holt, Rinehart,
and Winston, 1963.

Hebb, D. 0. The organization of behavior. New York: Wiley, 1949.
Hermelin, B. & O'Connor, N. Recognition of shapes by normal and

abnormal children. British Journal of Psychology, 1961, 52,

 

Lavrent'eva, T. V., & Ruzskaia, A. G. Communication V. Simultaneous
intersensory comparison of form at a preschool age. Doklady
Akademii Pedagogicheskikh Nauk RSFSR, 1960. Trans. Soviet

Psychology & Psychiatty, vol. 1.

Lobb, H. Vision versus touch in form discrimination. Canadian
Journal of Psychology, 1965,‘l2, 175-187.

 

McCall, R. B. & Kagan, J. Attention of the infant: effects of com-
plexity, contour, perimeter, and familiarity. Child Develope
ment, 1967, 38, 939-952.

 

Munsinger, H. L. & Kessen, W. Uncertainty, structure, and preference.
Psyohological Monographs, 1964, 28, (9, whole No. 586).

Munsinger, H. L., Kessen, W., & Kessen, M. L. Age and uncertainty:
Developmental variations in preference for variability.
Journal of Experimental Child Poychology, 1964, l, 1-15.

Munsinger, H. L. & Kessen, W. Stimulus variability and cognitive
change. Psychological Review, 1966, 23, 164-178.

Piaget, J., & Inhelder, B. The child's conception of space. New
York: Humanities Press, 1956.

 

Pick, A. D. Improvement of visual and tactual form discrimination.
Journal of Experimental Psychology, 1965, 62, 331-339.

Pick, A. D., Pick, H. L., Jr., & Thomas, M. L. Crossdmodal transfer
and improvement of form discrimination. Journal of Experimen-
tal Child Psychology, 1966, 3, 279-288.

Pick, H. L., Jr. Perception in Soviet psychology. Psychological
Bulletin, 1964, pg, 31—35.

Rudel, R. G. & Teuber, H. L. Cross-modal transfer of shape discrim-
ination by children. Neur0psychologia, 1964, g, 1-8.

Sheffe, H. The analysis of variance. New York: Wiley, 1959.

Sherrington, C. S. The integrative action of the nervous system.
London: Constable, 1906.

146

Vaught, G. M; Form discrimination as a function of sex, procedure,
and tactual mode. Poychonomic Science, 1968, lg, 151-152.

Vernon, M. D. A further study of visual_perception. Cambridge:
Cambridge University Press, 1954.

Vurpillot, E. The development of scanning strategies and their
relation to visual differentiation. Journal of Experimental
Child Psychology, 1968, 6, 622-650.

Zinchenko, V. P. & Lomov, B. R. The functions of hand and eye move—
ments in the process of perception. Problems of Psychology 1.
1960, 12-26. (reviewed in Gibson, E. J., Perceptual develop—

ment. in Stevenson, H. W. (ed) National Society for the Study
of Education Yearbook, 1963.

"1111111111111“