Professional Documents
Culture Documents
tionless percept. It becomes difficult, The writer and collaborators have re-
however, for the case of the moving eye. cently given a mathematical descrip-
Why does the phenomenal world not tion of this kind of stimulation for the
move during an eye movement? The general case of what is called motion
eyes perform saccadic or exploratory perspective (8). The fact that it has
movements without ceasing during wak- to do with the perception of space has
ing life; they perform compensatory long been recognized, but the fact that
movements whenever the head moves; it also has to do with the perception of
and they perform pursuit movements locomotion is less well understood and
whenever a moving object catches the deserves emphasis. The visual field
attention of the observer. Since the during forward locomotion seems to ex-
image of the environment moves across pand radially from a point of focus on
the retina during all these responses, the line of locomotion. The optical
the world should seem to move. It may geometry of this expansion is perfectly
be noted that with certain unusual types definite. The retinal image undergoes
of eye movement an observer will re- a deformation that can be neatly speci-
port that the world does seem to move; fied in terms of differential angular
examples are the after-nystagmus caused velocities. This retinal motion reaches
by bodily rotation (or other causes) high magnitudes during rapid travel,
and the artificial movement of the eye and there is reason to believe that it is
caused by pushing it with one's finger the important factor in the perform-
(11). During normal eye movements, ance of landing an aircraft. The ap-
however, the world does not seem to parent expansion of the visual field has
move, and this poses a question. been noticed by nearly everybody in
There are still other reasons for re- driving an automobile. The question
jecting the simple hypothesis that a mo- that arises is why the visual world does
tionless image yields a motionless per- not seem to expand but instead seems
cept. They appear when we consider to appear rigid, with the observer mov-
what happens when the observer moves. ing instead. The flier is never con-
The third problem concerns the visual fused by the impression that his run-
perception of locomotion in a stable en- way is behaving like stretched rubber.
vironment. We perceive not only the It is worth noting that there are spe-
motions of objects but the movements cial cases of visual stimulation in which
of ourselves; the performance of field- it does become equivocal whether the
ing a baseball illustrates both. In the visual scene is moving or whether the
case of active locomotion, such as run- observer himself is moving. If one sits
ning, there is, of course, a large com- looking through the window of a sta-
ponent of kinesthetic stimulation from tionary railway train at another train
the proprioceptors that accompanies the on the adjacent track, and if one of the
purely visual stimulation from the reti- trains begins to move slowly, the im-
nas. But in the case of passive or in- pression of moving self with stationary
voluntary locomotion, such as riding in scene may give way to that of station-
trains, automobiles, and planes, the kin- ary self with moving scene, or vice
esthetic component may almost wholly versa.
drop out. The visual component of The three problems of the moving ob-
stimulation results from the fact of mo- ject, the stationary environment, and
tion parallax, and consists of differential the moving observer are evidently in-
motions of different parts of the image. terrelated. Objective motion is con-
306 JAMES J. GIBSON
nected with subjective movement,2 since presenting light to the eyes of the ob-
both stimulate the retina. The motion server, that is, methods of systemati-
of an object, the movement of the eye, cally varying his retinal images. In the
and the movement of the observer him- case of motion, not many such devices
self may alter the retinal image in dif- have been successfully built. The types
ferent ways, but they all alter it. They of apparatus for inducing controlled im-
are all inseparable from the problem of pressions of objective motion are ap-
how or why we see the environment as proximately as follows: (a) the strobo-
stationary both when its image is al- scope and the variants of this device,
tered and when it is stationary on the used to study apparent motion; (b) the
retina. One thing is clear at least: the moving belt viewed through a window
kinetic experience in general involves or aperture, used to study so-called
the problems of so-called space percep- "real" motion, or to induce the water-
tion. fall illusion; (c) the rotating disk with
a spiral, used to induce the impression
EXPERIMENTAL EVIDENCE ON THE PER- of an expanding or contracting object
CEPTION OF MOTION AND MOVEMENT and the negative afterimage of this im-
pression; (d) the device of casting the
A survey of the established facts shadow of a physically moving or rotat-
about the three problems may clarify ing object on a translucent screen, the
them and even point to solutions. The deforming shadow inducing the impres-
experiments are not numerous, some of sion of a three-dimensional object in
them are unfamiliar, and they have motion; (e) the device of rotating a
seldom been considered together. disk with spiral lines behind a slotted
screen, inducing the impression of ob-
Motion oj an Object jects moving along the slot. Practically
Apparatus employed. Experimental all that is established about the percep-
studies of visual perception necessarily tion of motion comes from one or an-
depend on devices for systematically other of these experimental methods.
2
However, a novel device for presenting
In this paper, for lack of a better termi- multiple complex motions on a trans-
nology, the word motion will always be used lucent screen has recently been de-
to refer to change in position of an object,
and the word movement will always refer to scribed by Johansson (12). One might
change in position of the observer's body in suppose that the animated motion pic-
whole or part, that is, a response. Both may ture would have been used for con-
be visually perceived. The responses with trolled experimentation by psycholo-
which we are concerned are chiefly eye move-
ments and locomotor movements. Movements gists, but it has scarcely been tried (5,
of the limbs and bands are also important ch. 2). There have also been a num-
since they constitute a large part of behavior ber of setups with luminous spots in a
(gestures, manipulation, tool-using), and most darkroom, one or more of which are
of these are also visually perceived In them, put into relative motion. This latter
however, the kinesthetic component, the mus-
cle sense, is obviously important, and the experiment, like the autokinetic illusion,
visual component cannot be isolated for analy- is relevant to the problem of the sta-
sis as it can for locomotion. They are prac- bility of the environment as much as it
tically never passive or involuntary, as loco- is to the motion of an object.
motion can be. They will not be considered
here. Nevertheless the writer believes that
the visual feedback is just as important for Stroboscopic motion. The only large
motor performance as the bodily feedback, body of evidence based on these devices
and that "visual kinesthesis" should be recog- comes from the stroboscope. It is said
nized along with classical kinesthesis. to yield "apparent" motion as distin-
VISUAL PERCEPTION OF OBJECTIVE MOTION AND SUBJECTIVE MOVEMENT 307
guished from "real" motion, and the moving belt is presented to the eye.
stroboscopic effect is often loosely re- The same thing is true for the rotary
ferred to as the phi phenomenon. The motion of the surface of a disk. For
stroboscope has evoked much research, both, there are lower thresholds for
probably because it demonstrates that velocity and also upper thresholds for
a physically moving object is not neces- velocity when motion turns into blur.
sary for an experience of motion, and Acuity for motion is high at the pe-
because this seeming paradox has riphery of the retina considering how
prompted psychologists to formulate weak it is for color and form. There
controversial theories in order to ex- occurs a negative afterimage of ve-
plain it. locity in a stationary visual field in that
The important fact about stroboscopic part of it which has previously been
motion, for present purposes, is that the stimulated by a moving belt or disk.
stimulus is intermittent but that when The afterimage may be linear or rotary
certain relations hold, the perception of or it may be one of expansion or con-
motion is the same as if the stimulus traction if the rotating disk bore a
•were not intermittent. As Troland as- spiral that contracted or expanded (Pla-
serted, "a perfect motion impression teau's spiral). The perceived velocity
can be aroused without any actual mo- of a moving surface tends to be con-
tion of an object by the discontinuous stant at different distances from the eye
substitution of one object for another although the retinal velocity of its image
at progressively different points in is inversely proportional to distance.
space" (18, p. 381). This situation Brown, however, discovered some other
has frequently been reduced for experi- puzzling facts about such apparent ve-
mental convenience to the case of tvro locities connected with the size of the
successive light sources at two sepa- frame or aperture behind which the belt
rated points in space, and this experi- moved and with the brightness of the
ment has resulted in an elaborate Greek- surface (2). Another fact, which is in-
letter phenomenology of motion impres- teresting for the problem of the connec-
sions (alpha, beta, gamma, delta, and tion between retinal motion and eye
phi). The results of this experiment movement, is that perceived velocity is
have been reviewed elsewhere (for ex- reported to be somewhat faster when
ample, 1, ch. IS) and will not be dis- the eyes are fixated on the aperture than
cussed here. The fact is that when an when they follow the moving pattern
adjacent order and a successive order from one side to the other and back
of discrete stimuli are correlated, a con- again. This has been called the Aubert-
tinuous impression of an object in mo- Fleischl paradox (2).
tion results. The main limitation seems
Deformation of shadows and the per-
to be that the interval between stimuli
ception of depth. Linear and rotary
must not be too disproportionate to the
motions presented to the eye by belts
separation between them. Hence strobo-
or disks occur in the frontal plane of
scopic stimulation differs from so-called
the observer and are so perceived. So
"real" stimulation only in being discon-
does the apparent expansion of a Pla-
tinuous when the latter is continuous.
teau spiral, and this is also perceived as
The relations of order are the same in
flat except for an occasional report that
both.
the afterimage suggests motion in depth.
Motion of a patterned surface. The But the shadow of a rotating object ob-
speed and direction of linear motion are served from the other side of a trans-
perceived with some accuracy when a lucent screen, although seen in one
308 JAMES J. GIBSON
voluntary action comes from studies of haps the latter are not so different after
pursuit tasks, reaction time, and the all. The facts of the experiments can
like, which might be said to deal with be explained by the hypothesis that the
manipulation rather than locomotion. retina responds to adjacent and succes-
A theory of movement with respect to sive order. If the orders correlate for
a goal or destination is obviously of the stroboscope and the object, the fact
great importance, but we are here con- that the former is a discontinuous emit-
cerned with the cues or stimuli for ter may be unimportant. The two
movement as such. This may be justi- retinal images are similar in that the
fied on the grounds that the flow of ac- relations of order are the same in both;
tions, choices, or decisions during, for for example, right-left and before-after.
instance, an aircraft landing cannot be The stimulus for motion, then, may be
understood unless the flow of informa- ordinal.
tion is understood. There is other evidence to suggest
that the stimulus for motion is also re-
IMPLICATIONS OF THE EVIDENCE lational. This means that it cannot be
There is plenty of evidence to indi- derived from the hypothetical "local
cate that visual motion is a "sensory" signs" of retinal receptors. The fovea
variable of experience. It has a kind of does not have a fixed value for breadth
intensity (speed) and a kind of quality and height when stimulated by a single
(direction). It has absolute thresholds, point of light. Moreover, as Duncker
both lower and upper, like pitch. Acu- proved, the motion of one point of light
ity depends on the part of the retina on the retina is perceived relative to an-
stimulated, like form. It has a negative other point of light, not relative to the
afterimage, like hue. It tends to mani- retina. The frame of reference for mo-
fest constancy, like size and shape. In tion (or stability) seems to depend on
the form of "pure phi" it can be ab- the array of stimulation rather than the
stracted from an object. But more than location of the receptors; it is trans-
any sensory impression, it fails to cor- posable over the retina. Just as a mo-
respond to the physical stimulus pre- tion for the physicist can be specified
sumed for it. Whatever the stimulus only in relation to a chosen coordinate
for motion might be, it is not simply system, so is a phenomenal motion rela-
motion in the retinal image. This seems tive to a phenomenal framework (13).
to imply that motion is not sensory. Perceived motion occurs in a percep-
Before concluding, however, that phe- tually stable space or environment. An-
nomenal motion is not a function of other way of saying this is to assert
stimulation, the stimulus conditions that the perception of stability is part
should be re-examined. and parcel of the perception of motion;
you cannot have the latter without the
The distinction between "real" and
former.
"apparent" motion is unfortunate and
has interfered with the search for the The optical stimulus conditions for a
essential conditions. It should be noted stable environment seem to be a retinal
that stroboscopic stimulation can yield image containing many elements rather
just as psychologically "real" a motion than a few or one. This can be de-
as does continuous stimulation, if cer- scribed as a differentiated or "textured"
tain relations are preserved. A strobo- image (7, 9). Perhaps stability goes
scope and a moving object are mani- with the perception of a surface or an
festly different, but they are the sources array of surfaces extending over most
of stimulation, not the stimuli, and per- of the field of view. The disappearance
VISUAL PERCEPTION OP OBJECTIVE MOTION AND SUBJECTIVE MOVEMENT 311
of the autokinetic illusion when the Brown's results [2] are accepted)
darkroom is even slightly illuminated is points in the same direction.
consistent with this hypothesis. So is Facts about the perception of bodily
the occurrence of the moon-in-the-clouds movement as distinguished from object
illusion. So also is the railroad train motion are scarce. They are enough to
illusion when we take the window-filling suggest, however, that the impression of
train on the next track to be motion- oneself being moved, like that of an ob-
less. Perhaps the textural background ject being moved, depends on the per-
image, whatever its relation to the ception of the space in which the move-
anatomical retina, always tends to de- ment occurs. Ego movement like ob-
termine the phenomenal environment, ject movement can be induced. The
and the more it approximates the total train illusion and the cylinder rotating
image the greater the stability? around the head are examples. The
perception of forward locomotion can
Common experience suggests that we probably be induced, and the experi-
can perceive the motion of an object in ment should be tried. This will re-
depth as readily as its motion at right quire optical stimulation governed by
angles to the line of sight, and the ex- differential angular velocities for many
periments with deforming shadows on a points in the visual field, i.e., motion
translucent screen tend to bear out this perspective or, crudely speaking, an ex-
suspicion. The kinetic depth effects panding image.
so far obtained depend on perspective
transformations of the shadows, and A promising hypothesis for research
yield impressions of changing slant or would be that any transformation of
rotation. There is no reason why they the total retinal image, as distinguished
should not also be obtained with size from a part image within it, tends to
transformations of shadows, which will yield an experience of a movement of
the observer, and the kind of movement
yield impressions of linear approach and
experienced depends on the kind of
recession. A general hypothesis is sug-
transformation. For example, a simple
gested by these experiments, namely,
translation of the image may contribute
that any regular transformation oj a
to the experience of an eye movement;
bidimensional image tends to yield a an expansion may contribute to the ex-
tridimensional motion in perception, and perience of forward locomotion; a con-
the kind oj motion perceived depends traction to the experience of backward
on the kind of transformation. This locomotion; and so forth.
hypothesis has the advantage of relat-
ing the experiments on moving shadows There is said to be a striking lack of
correspondence between the presumable
to experiments on shape constancy and
optical stimuli and the ensuing visual
size constancy, and suggests a principle
perceptions of motion or movement.
of space perception that may be com-
The evidence does indeed show what
mon to both. The fact that the trans-
appear to be obvious discrepancies. It
verse motions of a pair of belts ob- is certainly true that kinetic impres-
served at different distances can be sions are not copies of their stimuli.
judged equal in velocity when the sur- But it fails to follow that they are not
faces are actually equal in velocity (if functions of their stimuli. It cannot
8 simply be assumed that a movement is
This hypothesis is consistent with, if not the same thing in the object, the retina,
essentially the same as, the position taken by
Duncker in his admirable study of "induced" the brain, and consciousness. The fore-
movement (4). going hypotheses make it possible to
312 JAMES J. GIBSON
test for psychophysical correlations, al- which we are concerned are tabulated
though they do not imply any pictorial below:
correspondence, between the dimensions
of the stimulus and the qualities of Rigid motion
kinetic experience. 1. Translation
2. Rotation
HYPOTHESES ABOUT KINETIC
RETINAL STIMULATION Elastic motion
3. Size transformation
A psychophysics of kinetic impres-
4. Perspective transformation
sions would require a mathematical
5. Deformation
analysis and classification of the mo-
tions or transformations of a retinal Disjunctive motion
image. This is a complex and difficult 6. Multiple movements
task for the future. Some preliminary
assumptions are possible, however. These abstract mathematical motions
Geometrically, one can distinguish be- are interestingly related to optical
tween a rigid and a nonrigid motion of stimulation. Let us assume an eye and
a form or of a set of points. Transla- a reflecting surface, such as the face of
tion and rotation are the types of rigid an object toward the eye, and let us
motion with which we are concerned. consider the cross-section of the sheaf
The figure after displacement is con- of light rays to the nodal point of the
gruent or identical with the figure be- eye (18, pp. 326 f.). This is equiva-
fore displacement. The kinds of non- lent to the retinal image. What tridi-
rigid motion are diverse and are still mensional events produce these motions
being explored by the higher branches of the bidimensional cross-section?
of geometry. However, two classes Numbers 1 and 2 above correspond re-
exist, which may be called elastic mo- spectively to a lateral movement of the
tion and discontinuous or disjunctive eye (or the object) and a swivel move-
motion. In the former, the lines of the ment of the eye (or a rotation of the
geometrical form do not "break up" (or object). Number 3 corresponds to a
the set of points maintains the relations movement of the eye (or object) along
of neighborhood), whereas in the latter the line between them. Number 4 cor-
the form is ruptured ( or the points are responds to a planetary movement of
"scattered"). The class of elastic mo- the eye around the object or an inclina-
tions includes two types, the size trans- tion of the object to the line between
formations and perspective transforma- them. Number 5 corresponds to an
tions on the one hand and nonperspec- event confined to the object—a fluid or
tive transformations on the other. The elastic motion of its substance. Finally,
first type can be defined as a projection number 6 probably corresponds to an
of the form or pattern on a plane dif- event such as the shattering of a single
ferent from its own, either an enlarge- object or the interaction of multiple ob-
ment (or reduction) or a slant projec- jects. Some of these statements need
tion. The second type can be defined qualification in order to be exact, but
as a deformation other than these, but they may serve as preliminary general
for which the continuity of the form is rules. In other words, some very im-
preserved. The class of disjunctive mo- portant types of physical events corre-
tions includes many types, which do not spond to the geometrical types of mo-
need to be specified here, but all in- tion in the projection. It is a reason-
volve discontinuity. The six types with able hypothesis that the eye can register
VISUAL PERCEPTION OF OBJECTIVE MOTION AND SUBJECTIVE MOVEMENT 313
these geometrical types of motion when causes a rigid movement of the image,
they occur in the retinal image. and any transportation of the eye
It may have been noted that the causes an elastic movement of the
physical events corresponding to mo- image.6
tions number 1, 2, 3, and perhaps 4 are The causes in the environment and
ambiguous. Whether the eye moves or the results in perception of deforma-
the object moves, the result is the same. tions and disjunctive motions of the
The optical situation assumed in the image (numbers 5 and 6 above) are
previous paragraph consisted of an eye complicated. So far, we have been as-
and a single object (specifically a plane suming a solid environment. Nonper-
face of an object). A more typical spective deformations are caused by
optical situation would consist of an liquid or fluid motions of physical ob-
eye and an environment. Let us there- jects and surfaces. Rivers flow, smoke
fore assume instead an eye and an infi- swirls, rubber stretches, and above all
nite plane surface. This is a better ap- living organisms flex their surfaces in
proximation to the terrestrial environ- many ways. The faces of men, for in-
ment. Except for the "sky," the image stance, undergo an astonishing variety
of the surface occupies the whole of the of rubbery motions, which we call facial
retina, and it constitutes a textured expressions. We perceive these motions,
background image rather than a de- sometimes with great acuity. We do
limited object image. An infinite plane not seem to confuse them with the me-
surface would be physically stable and chanical motions of solid obiects which
would constitute an excellent frame of tilt, slant, advance, or recede with a
reference for visual perception. There kind of inanimate quality. There may
is evidence to suggest that a background be a basis in optical stimulation for this
image does help to determine the stable difference.
phenomenal environment. Ambiguity of Disjunctive motions of the image are
perception as to whether the eye moves caused by a still greater variety of
or the environment moves in this situa- events. Objects break, ants swarm,
tion would therefore tend to disappear.4 billiard balls collide, and men shake
The types of physical events produc- hands. Michotte believes that multiple
ing the geometrical types of motion of motions can yield immediate impres-
a total background image are fairly sions of causation that are specific to
univocal. Translation and rotation of the relations between them, and he has
this image can hardly be caused by fortified his belief by experiments (16).
anything but eye movements. Size and The possibility of isolating high-order
perspective transformations for the ele- variables of stimulation in such images
ments of an extended plane surface con- seems remote, but it should not be re-
stitute motion perspective (8) and this jected.
can hardly be caused by anything but In conclusion, the various motions of
locomotion with respect to the surface. objects in a stable environment and the
Certainly it is true that any eye move- various movements of ourselves in that
ment in an illuminated environment environment can both be visually per-
ceived. A psychophysics of such kinetic
4
If to our disembodied eye we add assump- impressions, however, is almost nonex-
tions about gravity, posture, muscles, and
8
kinesthetk stimulation, the ambiguity would The classification of the motions of a
certainly disappear. But we are here con- retinal image here given U considerably re-
cerned only with optical stimulation, admit- vised from that proposed previously by the
tedly an abstraction. writer (6, p. 131 ff.).
314 JAMES J. GIBSON