A Psychology of Picture Perception

Full text

The Scope of Outline Pictures 107
even some of his pets (Chapter 5). Why is it that mere lines can be so
versatile? What are some rules that could explain their usefulness to
p
eople from different backgrounds?
Elements of the Visible Environment
Logically, if lines can depict the basic elements that create a
visible environment, then lines could have the power to depic
t
anything that was visible. Perhaps every visible object and every
scene is made of the combinations and arrangements of a few es-
sential elements. The same essential elements and the same kinds o
f
combinations may occur throughout the world. Can lines depict such
elements? If so, the usefulness of line drawings across cultures
would be understandable. Such elements would be properties of the
environment that influence the light coming to the eye. They would
be the elements about which light might provide information. Vary-
ing the order or layout of the elements would result in differen
t
objects, different landscapes, the entire visible world. And if line
depiction is tied to some of the fundamental features of the visible
environment, that would help explain why vision research such as
Rubin's inevitably became caught up in problems of depiction.
In one conception, the elements of vision are patches o
f
color, stimulated by light of different wavelengths. This is a tradi-
tional and standard conception of vision, but it is not useful fo
r
analyzing depiction, because depiction, and outline depiction espe-
cially, usually violates the colors of the depicted objects. A black
and-white sketch might depict a rainbow among the clouds, over a
grassy field. Nowhere in the sketch would the colors of the rainbow
b
e portrayed. To match the colors of the depicted object is a rare
achievement in illustrations. It is necessary to find some other con-
ception than color to describe elements of vision.
Another conception of vision could treat each object (o
r
pattern) as an element. In this view, each object is an independent
unit, and an arrangement of objects forms a scene-just as a still life
is a scene made up of domestic objects, each independent of its
neighbors, as well as all being together in one place. Verbal dis-
course involves fairly independent units-the various words we use.
But depictions of two different objects are more alike than are two
Chapter Seven
The Scope of
Outline Pictures
People who live a few miles apart,
in neighboring countries (such as France and Germany), often do
not understand each other, for though they use the same elements
of speech-that is, the same sounds--the combinations of sounds
often mean nothing to the listener. Of course, the elements alone-
isolated sounds--mean nothing to either speaker or listener. The
same is true with written words. As separate elements, W, 0, R, D
are only nondescript letters, and together they mean word only to
someone familiar with English. It is clear that to be familiar with a
language one must acquire a huge vocabulary, for each word may
mean something different; to know the elements of a new word, or
words like it, will not help one to understand the word.
But in any country and in any age since the cave artists, once
a man has learned about sketching he can, if he wants, represen
t
things to his neighbors (Chapter 5), his children (Chapter 4), and
106

108 A Psychology of Picture Perception
different words. We can draw a picture of an unfamiliar object for a
friend, and he will recognize the object if it eventually appears.
When we simply tell our friend the name for the object, he cannot
usually recognize the object when it appears. A picture of an un-
familiar object can tell our friend how many legs the object has,
where its arms and neck are, and so on. Presumably, the picture tells
our friend about familiar visual elements in a new arrangement. I
f
so, pictures represent parts of objects, not just the whole object.
So, are the basic elements of pictures the parts of objects?
Think of pictures of animals; the parts include arms, hooves, necks,
ears, and the like. Indeed, these parts make up the animals, as ele-
ments of the whole animal. But different kinds of objects are made
of different parts: animals are made of hooves, limbs, torsos, and so
on; suitcases are made of flaps, handles, locks, and the like; wine
glasses are made of brims, stems, etcetera. There is an infinite
variety of objects and an infinite variety of kinds of parts of objects.
If the parts we have been mentioning were the basic elements in
outline pictures, a large vocabulary of parts would be involved in
understanding them. Perhaps there are more basic elements than the
kinds of parts mentioned so far, some kinds of basic elements that
are few in number.
If objects can be analyzed into a few basic elements, per-
ceivers might use a small set of units--a small vocabulary--to un-
derstand any and all outline drawings.
Is there a small set of basic elements in the visible world?
When the question is put another way, a small set of elements is
suggested. Consider: Are there just a few elements that create the
optic array at the eye? One such element is a contour, so popular in
figure-ground research, where the pigmentation on a surface varies
from one area of a surface to an adjoining area. Another is
shadowing, where the illumination on a surface varies because an
opaque object intervenes between part of the surface and the source
of illumination on the surface. Another element is given by variation
in the relationship between a smooth surface, the source of il-
lumination, and the location of an observer--yielding highlights. Yet
another is provided by variation in the inclination of a surface--some
facets of a surface may face the direction of illumination more
The Scope of Outline Pictures 100
directly than others and so receive stronger illumination. And still
another is created by varying surface texture.
These elements are the main sources of variation in an optic
array and are the basic features of the visible environment. Varying
the layout of surfaces creates hills and valleys and the shapes o
f
objects. Variation in pigmentation creates the coloring of a land-
scape and its objects. Variation in the locations and opaqueness an
d
smoothness of objects and the locations of sources of illumination
creates shadows and highlights. Variations in the material of objects
and the forces they are subject to creates different textures. Optic
arrays contain the information for a visible environment, and, in the
last analysis, it is variation in surfaces-their layout and composition
and their relation to sources of illuminati on-that create optic arrays.
Variation in surfaces and their relation to sources of illumi-
nation provide the basic elements of the visible world. If lines in
outline drawings can depict many or all of the sources of an optic
array, then the versatility of outline sketches is understandable. The
visibility of objects and landscapes, the entire visible world, rests in
the main on a few sources of optic structure. If lines can depict each
of the main sources of optic structure, then lines can depict almost
any visible object or scene. And these would be not simply normal
familiar, objects, or familiar parts of objects, but anything that is
visible except pure color and uniform surface-for arrangements o
f
color patches on a surface may be outlined, but pure is0lated colors
or uniform surfaces cannot be.
Can each source of optic structure be depicted in outline
drawings? Will an untrained subject be able to understand such
outline depictions? Let us consider the elements of the visible worl
d
one by one.
Layout of Surfaces
Surfaces are either plane or curved. They can face toward o
r
away from an observer, so a concept of a point of observation must
be incorporated into any description of arrangements of visible
surfaces.
One plane surface may join another plane Surface at an

110 A Psychology of Picture Perception
angle, with both surfaces being visible from a particular point o
f
observation. The two surfaces form a two-sided plane angle, a
dihedral angle as it is called in solid geometry. There are two types
of dihedral angles. One type is concave to the point of o
b
servation,
like the comer of a room (Fig. 24). The other type is convex to the
point of observation, like the comer of a building (Fig. 25). An
abrupt change in the inclination of visible surface occurs across a
dihedral angle--a plane through the point of observation and the
apex of the dihedral angle meets the two surfaces at sharply differen
t
inclinations.
One of two adjoining plane surfaces may not be visible fro
m
a point of observation because it is behind the other surface with
respect to the point of observation (Fig. 26). The visible surface is
the front surface, and it is said to occlude the other surface. A front
surface occludes the back surface, and the visible terminations of the
front surface are called occluding edges. The distance from the poin
t
of observation to sources of optic structure changes abruptly from
one side of an occluding edge to the other. On one side of the
occluding edge is the front surface and on the other side is back-
ground.
A different kind of occlusion occurs with curved surfaces.
Instead of two plane surfaces meeting at a dihedral angle, one sur-
The Scope of Outline Pictures 111
face gradually changes its inclination and joins the back side of the
object smoothly, as in a sphere or the brow of a hill (Fig. 27). Again,
the back side is occluded by the front surface. The surface layout is
convex and rounded, not convex and angular. There is no edge,
strictly speaking of a sphere; nevertheless, spheres can occlude. I
will call the visible terminations of the front surface of rounded
objects occluding bounds. A sphere has no edges, but it has occlud-
ing bounds. Any curved surface has occluding bounds when its tan-
gent passes through the point of observation. There is an abrupt
change in distance of surfaces from the point of observation on
either side of the tangent from a point of observation to an occluding
bound. Beyond occluding surfaces lie backgrounds. A backgroun
d
lies behind an occluding edge or bound and does not make contact
with the occluding surface. Occluding surfaces occlude not only
their back surface but also parts of background surfaces. In some
cases, the background is not another surface, as occurs when the sky
is background. Whether the background is a surface or the sky, there
is an abrupt change in distance from the point of observation on
either side of a plane through an occluding edge or bound of any
terrestrial object. On one side of the plane is the occluding surface,
and on the other is the distant background.
Plane surfaces, curved surfaces, and dihedral angles are
concepts of layout that are independent of any observer. When a
point of observation is introduced, occluding edges, occluding
bounds, and backgrounds result. Occluding edges and bounds and
dihedral angles are visible features of surface layout.. Arrangements
of these features make up the visible terrain and objects standing on
the terrain. Can an outline drawing depict some of these features?
Which will be recognizable?
Cross-cultural research and research on children finds that
the shapes of objects can be recognized in outline drawings. The
shapes of objects are created by variation in the arrangement o
f
plane and curved surfaces; shapes are arrangements of features o
f
surface layout. So it seems that some of the features of surface lay-
out are recognizable in outline drawings- without training. And de-
pictions of all the features of surface layout occur commonly in
newspaper and textbook illustrations, in drawings from many cul-
Fig. 24 Fig. 25
Fi
g
. 26 Fig. 27
FIGURE 24. Concave comer.
FIGURE 25. Convex comer.
FIGURE 26. Occluding edge.
FIGURE 27. Occluding bound.

112 A Psychology of Picture Perception
tures and times. Cave paintings are often in the form of outline
drawings depicting the edges of objects and occluding bounds o
f
objects. The most common use of outline depiction is depiction o
f
features of surface layout--this is the usual vocabulary of the lan-
guage of outline. It appears to be a language discovered by early
man, universal in its understandability and inherent in the nature o
f
man's visual perception, for it requires no training.
Fig. 28 is an attempt to incorporate into one picture outline
depiction of all the features of surface layout. It should be instantly
understandable to the normal Western reader. Words to describe i
t
may be restricted to one culture, and the objects it depicts may be
unfamiliar to some cultures, but the kind of features it shows should
b
e identifiable in some line drawing or another by anyone reared in a
world of solid objects that rest on a terrain that stretches to a
horizon.
Figure 28 shows a seascape and a landscape with rounde
d
hills and a house with a walled-in garden. Different segments of the
lines in the figure depict the following six features: (1) An occluding
bound with a background surface-the brow of a hill, with the surface
of another hill behind. (2) An occluding bound with no background
surface-the brow of a hill, with sky visible above the hill. (3) An
occluding edge with no background surface--the apex of the roof o
f
the house, with sky visible above the house. (4) An occluding edge
with background surface-the termination of a wall, where the
continuation of the surface of the wall is occluded by the nea
r
surface, with ground surface visible beyond the termination of the
wall. (5) A dihedral angle forming a concave corner--two visible
plane surfaces of wall meeting at an angle of less than 180 degrees,
measured through the air enclosed by wall. (6) A dihedral angle
forming a convex corner--two surfaces of the house meeting at an
angle of more than 180 degrees, measured through the air around the
surfaces.
As Fig. 28 shows, all the basic features of surface layout can
be depicted by lines in outline drawings. A segment of line can de-
pict an occluding edge or bound or a dihedral angle. What makes a
segment of line depict at one time an occluding edge and at anothe
r
time an occluding bound is the context in which it is viewed. The
context can be other lines patterned around the line segment, as in
The Scope of Outline Pictures 113
Fig. 28, or the set of the observer, as pointed out in Chapter 6.
Besides the six features listed above, outline is capable of a
seventh kind of depiction. Each of the basic features of surface lay-
out is depicted by single lines in Fig. 28. However, sometimes a
single line can depict more than one feature of surface layout. That
is, representation of several features of surface layout can
sometimes be achieved with the use of a single line. One example is
contained in Fig. 28--namely, the crack between the door and the
walls of the house. A crack results when two dihedral angles of two
FIGURE 28. The meanings of a line. Seven kinds o
f
referents are included in this fi
g
ure.
surfaces abut or adjoin each other. Two abutting dihedral angles can
be depicted by a single line.
Another example of a single line depicting more than one
feature of surface layout was shown in Fig. 19, in Chapter 6. That
figure depicted a fence with a gap between two of the boards
b
ridged by strands of wire, each strand depicted by single wires. I
f
the strands were depicted as very thick, as thick as ropes, two lines
would be present, one line for each side of occluding bound of the
thick strands. Depicted by single lines, the strands are shown as thin
wires. To depict a thick strand, two lines could be used. Similarly,
to depict a wide crack, two lines could be used. Dihedral angles and
occluding bounds are depicted by single lines only. Thus, there is a
critical difference between strands and cracks and single features o
f
surface layout. Strands and cracks are created by several features o
f
surface layout, and in principle each feature can be depicted by
single lines. Wires are cylinders, with parallel occluding bounds.
Looked at closely, the cylinder would become evident. Cracks are

114 A Psychology of Picture Perception
spaces, provided by parallel abutting dihedral angles. The interior or
b
ackground of the space could be made evident. But typically the
substance of the wire or the background of the space may be
indistinct. That is, optic structure from the material of a wire o
r
coming through a crack is often too fine to be registered. A general
rule suggested by line depiction of wires and cracks is as follows:
An arrangement of several features of surface layout can be depicted
b
y a single line if the features are close together, parallel, and the
optic structure from the region between the outer margins is not
distinct. This general rule could cope with line representation o
f
cracks and wires, and it provides for grooves and scars and railroad
tracks being depicted by single lines.
In sum, line segments in line configurations can depict the
basic features of surface layout-dihedral angles forming convex and
concave corners, and occluding edges and bounds, with or withou
t
background surfaces, and single lines can also depict com
b
inations
of features that are close together and parallel, with indistinct
internal detail. To rephrase, lines can depict discontinuities of depth
or slant, and single lines can depict combinations of these
discontinuities if the discontinuities are in parallel, close, and with-
out distinct internal detail.
The Substance of Surfaces
The substance of surfaces is often capable of structuring the
light that comes to the eye. A surface can be smooth or planar and
yet entail substances that have different capacities to reflect light.
This difference in reflectance may be selective with regard to the
wavelength of light, in which case one part of the surface is said to
have a different color than other parts. Or the difference in reflec-
tance may simply result in some areas being able to reflect more o
f
die incident light than others, in which case some areas are said to
be lighter than others. The differences in reflectance are said to be
due to differences in pigmentation. So a surface may have no layout
discontinuities of distance or inclination and yet provide pigment
discontinuities.
To show depiction of pigment discontinuities alone it is nec-
essary to represent layout in which the only discontinuities are pro-
The Scope of Outline Pictures 115
vided by change in pigmentation on a smooth surface. Pure cases o
f
pigment change without surface-layout discontinuities occur in the
coloration of animals. The wings of butterflies and the hides o
f
animals are often sources of pure pigment discontinuities that do no
t
correspond to surface-layout change. Can pigment discontinuities be
recognized in outline depiction?
Figure 29 is an attempt to show pure pigment change. The
object shown is an animal. If a subject recognizes Fig. 29 as an ani-
mal, a horse in particular, information about features of surface
layout is being recognized. If a subject sees the figure as depicting a
zebra, the interior lines of the figure are functioning as depictions o
f
the pigment-change on the hide of the animal. If the subject were to
see the figure as a horse with lines painted on its flanks, then the
interior lines would not be acting as depictions of boundaries o
f
areas of pigment. In informal fashion, dozens of adults and two
children have been shown Fig. 29. All the adults, usually faculty and
graduate students, have taken the figure to be a depiction of a zebra,
and thus the interior lines depicted margins of areas of pigment. The
two children, visitors to the Cornell psychology laboratory, had
interesting reactions. One, a five-year-old girl, thought of the figure
as a horse with lines on it. The other, her eight-year-old brother, no
t
only saw the figure as depicting a zebra, but even spontaneously
identified some areas as showing the shape of light or dark areas on
the pictured zebra. The guiding principle he offered was that zebras
are light on the underbelly, and from this knowledge he was able to
figure out where the dark and light areas should be!
The adults and children were simply shown Fig. 29 and
asked "What is this?" No explicit hints were given. Once they had
responded, they were asked "What is this line for?" They were asked
about lines for the back and legs of the zebra before being asked
about an interior line. The figure was recognized as a zebra by the
adults and by one child. It seems likely from this informal evidence
that lines can depict pigment discontinuities without training in any
convention.
Lines can depict changes in layout of surfaces and the layou
t
of pigmentation on a surface. The context for a line is a critical
factor in determining what a given line is depicting. The more un-

The Scope of Outline Pictures 117
a shape as an area of pigment one need not have all the coloring
reproduced. Features of shape, without color, can be adequate. In-
formative features of the figure probably include, for example, the
fact that none of the interior lines disturb the overall silhouette. The
interior lines simply terminate at the exterior lines; they do not make
exterior lines poke out, as gaunt ribs would. The exterior line is
smoothly continuous. Thus, the interior lines do not indicate ribs
and concavities, which would disturb the overall silhouette. The
overall horse shape is probably critical, too. If the overall shape was
not evident, the interior lines could easily be taken to be contours on
a map. Many features of this zebra depiction are shared with othe
r
figures, but the zebra figure seems to include enough distinctive
features to be specific to outline depiction of pigment areas. The
lesson to be drawn is that outline depiction capitalizes on-and, if a
depiction is to be unambiguous, is restricted to visible features of the
environment that have distinctive form. This lesson will be
reinforced as more figures are considered.
In sum, subjects recognize Fig. 29 as a zebra, taking some
segments of line as depictions of pigment borders. Lines can depict
discontinuities of pigment.
Layout and Illumination
From Surface layout and pigment layout we now turn to
variation in illumination on Surfaces and variation in illumination
from surfaces--that is, shadows and highlights.
Shadows. When shadows are present on a surface, it is no
t
uniformly illuminated. A shadow is cast on a surface when an opaque
b
ody is situated between a source of radiant light, or the direction o
f
illumination, and the surface. The opaque body intercepts light that
otherwise would reach the shadowed Surface. Removing the opaque
body allows the illumination to reach the surface. When an opaque
b
ody prevents illumination from reaching an area of a surface, the
area is said to be cast in shadow.
Generally, terrestrial Surfaces shadowed from the prevailing
illumination are not entirely without illumination. Usually, light
reflected from other objects and the sky reaches terrestrial areas
shadowed from the sun. Thus, there is not the extreme contrast
116 A Psychology of Picture Perception
FIGURE 29. An outlined zebra containing lines depicting
p
i
g
ment boundaries.
familiar the context--the depicted configuration--the more difficult
it may be to ensure that the observer sees only one definite feature
of the environment as depicted by a given line. Presumably, the
five-year-old girl who failed to see Fig. 29 as depicting a zebra was
comparatively unfamiliar with zebras. Perhaps a depiction of a
favorite pet cat with lines showing the shapes of the pet's markings
would make the transfer from the solid, colored, textured, real world
to the merely black-and-white line drawing easier. A drawing of a
favorite costume, with its color patterns depicted only by lines,
might be easily recognizable, too. Possibly the simplest thing would
b
e to show examples of flags, like the American, British, and
Canadian flags, in outline, omitting all color. Their distinctive pat-
terns completely depend on pigment differences in otherwise uni-
form cloth.
Outline drawings restricting the observer to one unam-
b
iguous perception must usually both replay a highly specific
optical structure and also capture the observer's understanding of
environmental structure. Figure 29 seems to be specific enough for
many subjects. Lines in the figure are identified as depicting
pigment areas. The figure provides information about pigmentation
without reproducing the full patchwork of a zebra. It seems that to
identify

118 A Psychology of Picture Perception
found with objects in empty space. Usually, too, terrestrial shadows
result from objects intervening between surfaces and extended
sources of illumination--not point sources. So there are penumbras
to most shadows-regions where part of the extended source does
offer illumination. Penumbras result in the softly blurred appearance
of the margins of cast shadows.
Can cast shadows be depicted in outline? Outline drawings
omit the brightness-darkness change entailed by shadows and the
colors of shadows beloved of Impressionists. Outlines are made with
fine, sharp lines and do not provide cross-hatching or other means to
create gradients that would be like the gentle gradients in penum-
bras. These gradients seem important in recognition of shadow
(Helmholtz, 1924; MacLeod, 1932). As a result, it may not be
surprising that cast shadows are rare in outline drawings. In a sample
of more than 10,000 line drawings from the Cornell Fine Arts
Library, there were no examples of depictions of cast shadows. The
majority of the drawings were recent European and American work,
b
ut collections from Japan, India and the Near East, and from Celtic
and Anglo-Saxon times were included.
The rarity of outline representation of cast shadow suggested
an experiment (Kennedy, 1970). Given this rarity, most subjects
p
resumably have never been instructed that outlines can represen
t
shadows. Would instruction be necessary for subjects to perceive
shadows when shown (probably for the first time) an outline
drawing in which cast shadows are represented?
To avoid problems with penumbras, photographs in which
distant shadows were present were sought. Distant shadows can be
far enough away that penumbra are not evident. For distant shadows,
the angles subtended by the penumbra are so small that the transition
b
etween light and dark seems sharp to the unaided eye. Only
p
hotographs of terrestrial shadows were considered, those involving
natural or ecological optics, not altered by darkroom or laboratory
trickery such as reversing or altering illumination on the picture
d
scene with lenses or mirrors or artificial sources of light. In many o
f
these photographs, shadows were easily recognized as such; thus, a
p
enumbra is not necessary for a shadow to be recognized.
A particular line drawing (Fig. 30) was prepared from one o
f
the photographs. The outline was made by "tracing" from selected
The Scope of Outline Pictures 119
p
ortions of the picture. In particular, discontinuities of pigment
representing shadows were traced from the photograph. In addition,
human figures and luggage cases and a pole were depicted by lines
in the drawing, traced from pigment arrangements in the
p
hotograph.
The photograph capitalizes on the laws of projection of light,
and so it is in projective correspondence with an environment. The
lines traced from the photograph are also in projective corres-
p
ondence with features of the same environment. Can enough
structure be retained in a projective line tracing to make the lines
informative about corresponding features? In particular, will the
p
ictured shadows be recognizable even to untrained observers?
FIGURE 30. An outline drawing depicting boundaries of
shadows b
y
lines.
Eight adult subjects--graduate psychology students at Cor-
nell--were shown Fig. 30. They were asked, "What is this?" ---
nothing more direct. Six immediately identified the picture as de-
p
icting shadows. If the subject did not specifically say where the
relevant lines were, and name the objects casting shadows, he was
asked, "Shadows of what?" and "Which lines represent shadows?"
The six subjects correctly identified the lines in the lower half of the
figure as depictions of the shadows of men in a row, carrying flags.

120 A Psychology of Picture Perception
Two subjects did not spontaneously identify the shadows.
One said, "I don't know what this is" while indicating the lower hal
f
of the picture. The other said, "Is it water?" Both of these subjects
were given a hint, the one word "Shadows." Both then correctly
identified the shadows and correctly said what was casting the
shadows. The hint did not indicate particular lines or suggest
p
articular objects casting shadows. Still less was it instruction in a
convention or an arbitrary code. The objects casting the shadows
were not depicted in the drawing, so subjects could not identify an
object and then guess, following the hint, that otherwise-meaning-
less scribbles "must be meant to be their shadows." Nor were there
any flags or men standing in a row anywhere in the scene.
Perception of shadows in the drawing is not due to a learned
convention. At the time, November 1969, there were few if any out-
line depictions of cast shadows on which a convention might have
been based. Since 1969, examples have emerged in recent "revolu-
tions" in graphic art, inspired by high-contrast photography (which
emphasizes the structure of optic arrays, often at the expense o
f
recognizability of the environmental source of the more demurely-
contrasted array).
It seems that the capacity of outline to depict recognizable
shadows is inherent in the perceptual skills of untrained adult ob-
servers. What were the important attributes of Fig. 30 that made i
t
depict shadows rather than any of the host of other possibilities,
from comers to color patches? Presumably, Fig. 30 must presen
t
aspects of shape that are distinctive to shadows.
What attributes does Fig. 30 have in common with shadows?
The figure presents some of the shapes of men, but there are
anomalies. Parts of the figures are surprisingly wide, and other parts
in comparison are unusually narrow. The nearest arm of the shadow
of the nearest man is a particularly clear case. Notice, too, that the
wide parts point toward the observer, and thin parts are at right
angles to these. And the shoulders of the figures are oddly skewed
with respect to the main axis of the figures. Thus, the lower figures
are projected as would be flat figures, not as voluminous solid ob-
j
ects. Their flat character distinguishes them from real men.
Also, there is no interior detail to the lower figures. The
flags being carried are continuous with the standard and the stan-
The Scope of Outline Pictures 121
dard with the human figures. Absence of internal detail is a common
characteristic of shadows, particularly shadows in a natural scene
with no artificial lighting. Further, the baggage, the pole, and the
erect human figures define a ground plane and its horizon. These
objects in these erect postures should be resting on a solid surface.
The three human figures project smaller subtended angles, as thei
r
feet are depicted higher in the picture plane. The location of the feet
and the decrease in size corresponding to height in the picture plane
is perspective information for the location of a surface and its
horizon. One could even work out what height the photographer
held the camera, which would be the height at which the horizon
intercepts the standing figures.
Just as the erect figures define a ground plane and an imag-
inary horizon, so, too, do the figures in the lower half of the illus-
tration. The figures decrease in size and width as they move up the
p
icture plane. The axes of the human figures and the standards
converge toward the horizon; continued, they meet in one point.
Thus, these flat figures provide perspective information for a surface
and a horizon that corresponds to the surface defined by the erect
figures. It is characteristic of cast shadows that they lie on surfaces,
the very surfaces defined by other features of the environment.
Familiar figures that are present in outline, appearing flat,
with no internal detail, merging smoothly with the shapes of othe
r
figures, lying on terrain without suggesting surfaces at variance with
the rest of the environment--in such ways Fig. 30 contains informa-
tion for shadows. Probably none of the subjects who identified shad-
ows in the illumination could have identified many of the attri
b
utes
of the outlines that specify shadows. The figures presumably
capitalize on intuitive or tacit understanding of features of the
visible world, just as in sound localization we can say where a sound
source is but not how temporal differences at the ears are used to
identify the location.
To support the analysis of the illustration of shadows, the fig-
ures were redrawn in various ways. The expectation was that if any
attribute was altered, no observer would identify the result as a
depiction of shadows. The attributes were divided into (a) those fo
r
flatness, (b) those for absence of internal detail, and (c) those for
concurrence between the terrain and the surface bearing the flat

The Scope of Outline Pictures 123
122 A Psychology of Picture Perception
figures. An illustration in which all three attributes have been altered
is Fig. 31, which shows erect solid figures, men complete with
internal detail. Compared to Fig. 30, attributes (a), (b), and (c) have
been changed in Fig. 31. In Fig. 32, only (a) and (b) are changed.
Nine subjects (summer school students at Cornell) were
shown Figs. 30, 31, and 32. They were simply asked to comment on
the things pictured. Eight of the nine identified the shadows in Fig.
30. All nine identified the figures in Fig. 31 as being men standing
in a row and did not mention shadows. No subject mentioned shad-
ows for Fig. 32, seven of the subjects saying the figures were supine
men. Two thought the men were strangely flattened--"cutouts,"
said one; "flattened," said the other.
When internal detail was added, but the flatness (or con-
currence with a single plane) retained, subjects often commented on
the flatness of the figures, saying for Fig. 33 that there was an
impression of figures "painted on the ground" or "totally flat" o
r
"flat. and unreal." Again, no one mentioned shadows. However,
when the internal detail was removed, as in Fig. 34, even if the
outline was for a rounder, fuller figure, six out of nine subjects still
mentioned shadows. The information for the solid silhouette of a
man was not preventing subjects from using the lack of internal
detail and location of the figures as information for a shadow.
Could the kind of difference between Fig. 30 and Fig. 34 be
used by subjects? Perhaps the subjects were simply using lax criteria
for fonn, assuming that the drawings are made roughly with no great
emphasis on niceties of form. If so, subjects could be asked to
compare two drawings including the kinds of difference
distinguishing Fig. 30 and 34. Two extra drawings of vertical fig-
ures, different only in outline information for fullness or solidity
(Fig. 35 and Fig. 36), were shown to all nine subjects. They were
asked to say which looked more "bulky" and which looked more
"flat." All nine subjects chose correctly--Fig. 35 was said to be more
"bulky."
It seems that absence of internal detail, concurrence with
surface, and flatness are distinguishing features of shadows. These
aid subjects in recognizing outline depiction of shadows. Informa-
Fig. 32
FIGURE 31. The foreground figures are solid, detailed, and erect.
FIGURE 32. The supine figures are solid and detailed.

124 A Psychology of Picture Perception The Scope of Outline Pictures 125
Fi
g
. 33 Fi
g
. 35
Fi
g
. 36
Fig. 34
FIGURE 33. The supine figures are flat and detailed.
FIGURE 34. The information for solidity--rather than flat
shadows in the supine figures is often not detected.
FIGURE 35. The erect figures contain information fo
r
roundness or bulkiness.
FIGURE 36. The erect figures contain information for flatness,
unlike these in Fi
g
ure 35.

126 A Psychology of Picture Perception
tion for flatness may be the least helpful aid, if subjects are not dis-
couraged from making allowances for imprecise drawing.
The general aim of the studies was fulfilled. They showed
that outline can depict shadow, and shadows in outline depictions
can be recognized without training in a convention.
Highlights. Shadows can be depicted in outlines. Can high-
lights, another phenomenon created by the relation of surface layout
to illumination, also be shown in outline?
Highlights depend on surfaces being polished or smooth.
Highlights appear when the relationship between three factors-a
source of radiant light, a polished surface, and a point of observa-
tion--is just right. Shadows depend on surfaces and light sources and
are independent of a point of observation; highlights are not. As an
observer moves around, the surfaces showing highlights change.
Just looking first with one eye and then with the other can produce
remarkable changes in highlights.
In general, incident illumination is reflected off a surface
either in a specular manner (minor reflection off a smooth, polished
surface) or by scatter reflection. If the surface is polished or smooth,
illumination is reflected at an angle equal to its incident angle. In
contrast, a rough surface scatters incident illumination in many
directions. Often there is a compromise between scatter and specular
reflection, and the surface is said to be partially polished.
If the surface is at all polished, a station point will receive
p
articularly strong illumination from a direction meeting the surface
at an angle equal to the incident angle of prevailing illumination
(FIg. 37). Station point i in Fig. 37 receives light from points a and b
on a surface, but the light from a is particularly intense. Station
p
oint i lies in a direction from a where incident illumination is being
p
articularly strongly reflected. For station point ii, the illumination
from b is particularly intense, for similar reasons. For station point i,
there is a highlight in the direction of a. For station point ii, there is
a highlight in the direction of b. If one eye were at i and the other at
ii, the direction of a highlight would seem to shift from a to b, as an
observer looked with one eye and then the other.
So discontinuities in optic arrays--discontinuities in the
intensity of light from a surface to a station point--occur, given the
The Scope of Outline Pictures 127
right arrangements of sources of light, polished surfaces, and station
p
oints. "' ,
Are highlights recognizable in outline drawings?' They ,are
not uncommon in line drawings. Objects such as balloons and
bottles, rounded, with smooth surfaces, are often drawn to include
outline depiction of highlights (Fig. 38). Lines mark the margins o
f
directions of discontinuities of illuminaton.
Cross section of
a partially
p
olished surface
Length of these
arrows indicates
amount of light
reflected in the
direction of the
arrows
Fi
g
. 37
Fi
g
. 38
FIGURE 37. If a surface is partially polished, most of the incident
light is reflected at an angle equal to the incident angle.
FIGURE 38. Highlights on balloon and bottle.

The Scope of Outline Pictures 129
found in garments. The hem of a sweater, for example, is often
different from the body of the sweater, only in the arrangement o
f
the strands that have been knit to form the sweater. The coloring is
often identical in the body of the sweater and the hem. Except when
the hem is folded up or rolled up, the surface layout remains at one
level. The level of illumination from the sweater is the same from
the body and hem. The distribution of tiny pockets of shadow, from
the individual strands of the material, vary in the body and hem,
creating a visible texture, but the general level of illumination re-
mains the same. The discontinuity due to a hem is a pure case of a
texture discontinuity.
Figure 39 is an outline drawing of a sweater, traced from a
p
hotograph, in which the junction of the hem with the body of the
Sweater is depicted by a line. No difference in coloring marks the
b
oundary of the hem. Differences in weave are not indicated by
differences in stippling or hatching. Is the line for the edge of the
hem an effective depiction of a texture discontinuity, so that there is
no need for captions or training in a convention? Consider replies
from eight adult subjects (students at Cornell) asked about Fig. 39.
128 A Psychology of Picture Perception
Since outline depiction of highlights is fairly common, it is
probably not necessary to test adults to see whether such outlines are
recognizable, but it may be instructive to test children. My two child
visitors (five and eight years old), mentioned previously, failed to
identify the highlights depicted in Fig. 38. Asked what a highlight
was, the younger was nonplussed. The older said, "Well, you take a
light and put it up high!" Perhaps in that odd way that we can fail to
notice our own shadow during the day, so children can fail to notice
highlights until they are pointed out or become a source of play in a
contemplative moment. At some point, tacit understanding is able to
support outline depiction; the age at which that understanding is
reached is still a matter for conjecture.
To summarize, line segments can depict discontinuities o
f
illumination due to highlights or cast shadows, drawing on adult
tacit understanding of the ecology of light.
Changes in Texture
A surface texture occurs when a unit is repeated over an area
with stochastic regularity. A texture can be described in terms of the
unit being repeated, the number of units in a given area, and the
distribution of the unit. An arrangement of small areas of pigment
may constitute a texture, as in a slab of speckled quartz. Small
mounds and depressions may constitute a texture, as in ripples on a
b
each. Visible texture is created by arranging units defined by
pigment or layout or illumination changes.
A discontinuity in texture occurs where a terrain abruptly
changes its texture. On a beach, a pebbly area might adjoin an area
of sand. A grassy area might change to lichen as underlying soils
varied. As texture changed on a beach or area of vegetation, so, too,
would color and general layout, in many cases. Finding a pure case
of a texture change in nature is not easy.
Can a texture change be recognizably depicted by a line in an
outline drawing? Since layout and pigment and illumination dis-
continuities can be depicted in outlines, it is necessary to find a pure
case of texture change--one where no other kind of discontinuity is
p
resent. Otherwise, observers may recognize one kind of discontinu-
ity being depicted and infer the other. So far I have discovered only
one kind of pure texture change that is at all common; it is a kind
FIGURE 39. An outlined sweater, with line depiction of the
u
pp
er border of a hem
,
to show chan
g
e in texture.
The subjects were shown the drawing and asked nondirec-
tive questions. First they were asked, "What is this?", when the
drawing was presented to them. They generally replied "Its a draw-
ing of clothing--a sweater" or words to that effect. If the subject did
not mention clothes-as in the case of one who said, "It looks like a
p
age from Sports Illustrated!"--he was asked to say a little more. All
subjects mentioned garments or sweaters specifically. The second
question was about a specific line. The subject was asked, "What is

130 A Psychology of Picture Perception
this?", as the experimenter pointed to the line depicting the outer
edge (outer occluding bound) of the arm, just below the shoulder.
All subjects identified the line correctly. They were then asked,
"What is on this side of the line?", as the experimenter pointed to
the interior of the represented arm. All subjects identified the area as
the interior of the arm. They were then asked, "What is on this
side?", and the experimenter pointed to the exterior region adjoining
the line. Subjects said "Nothing" or "Air" or "Background" or words
to that effect.
The next line they were asked about was the topmost hori-
zontal line (for the neck) of the same sweater. Again the first ques-
tion was, "What is this?", and the next two questions were, "What is
on this side?", as the experimenter pointed to the two areas on eithe
r
side of the line. All subjects mentioned the neck of the sweater and
the neck of the wearer and mentioned the neck of the wearer was
not specifically represented.
What these questions established was that subjects under-
stood the questions despite their very general nature and that the
p
icture depicted a sweater in the appropriate orientation to the
subjects. Line depiction of layout was clearly effective for all these
subjects. The line of questioning was clear and meaningful to all the
subjects.
The next question centered on the line depicting the hem and
was again nondirective. The experimenter pointed to the line that
was a tracing of the upper boundary of the hem as shown in the
original photograph. Subjects were asked, "What is this?" Six of the
eight immediately mentioned a change of weave or knit or a change
of pattern or said "It's the top of the hem" and, asked to enlarge,
mentioned a change of weave. Of the other two subjects, one said
"It could only be ribbing" and, questioned, explained that a rib was
an elongated ridge knit into a sweater, and that the line in question
represented a rib that ran parallel to the bottom edge of the sweater.
The subject was adamant that ribbing was the only thing the line
could represent. The other subject said that the line represented the
j
oin of the hem to the body of the sweater. This subject was unable
to enlarge upon her answer, repeating, "the top of the hem" when
asked to explain. Asked to explain further, she said she could not.
Asked to explain what a hem was she could only say it was the
The Scope of Outline Pictures 131
bottom band of a sweater. It did not seem possible to have the
subject use terms like texture (for example, "weave," "knit,"
"pattern," "matting," "braiding," "reticulation," "plaiting") without
suggesting them directly, so the subject was not questioned further.
Whether she was thinking of a hem in terms of weave but could not
express this remains unsure.
So, six out of eight identified the line depiction of a texture
discontinuity. What makes the line provide information about a
texture change rather than any of the other things a line can depict?
Why was it not the top of the rolled-up edge of the sweater? Pre-
sumably, the line pattern in the sketch must contain distinctive
features of a texture change.
The figure provides a familiar configuration--a clothed body.
The lines fit the overall shape of a human torso but also have
characteristics of sweaters-curves representing folds, proportions that
are bulkier than a nude body, and extra lines cutting across the limbs
and body to depict cuffs for sleeves and the termination below the
waist. Consider the curves of lines that depict the sides of sleeves
and the lower body of the sweater and the neck. The neck line
changes direction; it swerves to show the silhouette of the bottom
section of the folded-over roll neck. The cuff line swerves to indicate
the end of the sleeve. These swerves depict layout edges--edges that
can be silhouetted. Similarly, the bottom edge of the body of the
sweater is marked by a line that comes down the side of the body
and then swerves to become horizontal before it divides in two, a
horizontal continuation and the vertical continuation. The lines for
changes of layout swerve to silhouette each layout change. But lines
for the side of the sweater do not swerve as they approach and form
a junction with the line for the upper part of the hem. At that
junction, the lines for the sides of the sweater continue undeviat-
ingly. Accordingly, that line does not suggest a layout change.
The line for the texture discontinuity occurs in the midst of
information for a garment and is parallel to and a short way above a
line for a change of surface layout, the end of the material of the
sweater. The line does not indicate a change of surface layout (like
a line for the top of a belt, for example), since if it did, the line for
the side of the body would change direction when close to it. The

The Scope of Outline Pictures 133
outline is a language of discontinuities and distinctive features o
f
shapes. Cave artists mastered the vocabulary of surface layout, and
still today that is the great domain of outline, the common pictorial
language of many cultures. It seems other basic features of vision
can be a part of that domain, too, when need arises.
In sum, outline drawings capitalize on ecological informa-
tion provided by distinctive features of shape and permit observers
without training or captions to identify basic discontinuities o
f
shape, slant, pigment, illumination, and texture. Outline can depict
discontinuities without reproducing the colors or textures or inten-
sities that define each discontinuity, by presenting the informative
variables of shapes that help distinguish each discontinuity.
132 A Psychology of Picture Perception
line is in the appropriate place for the border of a hem; such borders
in sweaters are typically texture discontinuities.
Hems are sometimes marked by color change as well as tex-
ture change. Hems are not exclusively dependent on texture discon-
tinuities. It is a little surprising that no subject mentioned a color
change. Perhaps there is some feature of shape that distinguishes a
p
igment change from a texture change, and a drawing that includes
such features would be a useful tool to explore the tacit knowledge
underlying subjects' judgments.
In summary, subjects can recognize outline depiction o
f
texture discontinuities, without training or captions. A drawing that
subjects identify as depicting texture change by outline contains
some distinguishing shape features of pure texture change.
Recognition of texture depiction by outline adds one more
ecological phenomenon to a long list that can be identified in outline
drawings.
Conclusion
Corners, whether convex or concave, occluding edges and
occluding bounds, with or without backgrounds, parallel combina-
tions of features of surface layout like wires or cracks, edges of shad-
ows, highlights, and pigment boundaries--all yield to outline along
with texture discontinuities. Perhaps abrupt change is the. facto
r
tying all these phenomena together. Abrupt change of depth or slant
with respect to the station point is the result of a feature of surface
layout. Abrupt change of illumination underlies shadows and
highlights. Abrupt change of reflectance defines pigmentation
change. And abrupt change in weave was depicted at the hem of a
sweater.
The rule that follows is that lines can depict discontinuities,
any of the visible discontinuities of surface, pigment, illumination,
and texture layout. These are the basic features that create the visible
environment. It follows that anything that has distinctive features o
f
shape and is visible should be identifiable in outline drawings. The
p
ower of outlines does not rest on showing whole objects, which, o
f
course, they can do, but on being able to present information for the
fundamental features of the visible environment. The language of