Brain (1995), 118, 1105-1127

Motion and shape perception in cerebral
akinetopsia
Matthew Rizzo,1 Mark Nawrot1 and Josef Zihl2
1

Division of Behavioral Neurology and Cognitive
Neuroscience, Department of Neurology, The University of
Iowa College of Medicine, Iowa City, USA and the 2Max
Planck Institute Fur Psychiatrie, Miinchen, Germany

Correspondence to: Matthew Rizzo, MD, Division of
Behavioral Neurology and Cognitive Neuroscience,
Department of Neurology, The University of Iowa Hospitals
and Clinics, Iowa City, IA 52242, USA

Summary
Motion cues serve many purposes in primate vision.
Consequently, akinetopsia, a defect of movement perception
due to cerebral lesions, would be expected to comprise a
range of motion-related defects. To address this issue we
explored further the perceptual profiles in akinetopsic subject
L.M. who has motion perception deficits due to extensive
bilateral lesions of the dorsolateral visual association cortex
that spare primary visual cortex, area VI. We used several
different experimental procedures. Using random-dot motion
stimuli, we showed that L.M. can still perceive global coherent
motion and discriminate motion direction, yet these abilities
fail even at moderate levels of background noise. L.M. also
viewed a Mo-frame apparent (phi) motion stimulus known

as a Ternus display. Her performance on this test suggests
defective 'long-range' apparent motion mechanisms. These
results are clearly in agreement with previous reports.
Additional experiments showed that L.M. can still perceive
2-D shape and 3-D structure-from-motion (SFM). However,
like motion direction discrimination, these abilities broke
down at moderate levels of moving and stationary noise.
Surprisingly, L.M. also had trouble perceiving 2-D shapes
defined by non-motion signals including 'on' and 'off'
transients, dynamic and static binocular disparity, and static
texture cues. Our findings highlight the role of the visual
association cortex in extracting salient information from
noise.

Keywords: akinetopsia; area MT (V5); form-from-motion; motion perception; structure-from-motion; visual cortex.
Abbreviations: AFC = alternative forced choice; AM = apparent motion; BVRT = Benton Visual Retention Test; CFT =
complex figure test; CS = contrast sensitivity; HVOT = Hooper Visual Organization Test; ISI = inter-stimulus interval;
LCD = liquid crystal display; MT = mesial temporal; RDC = random-dot cinematograms; SFM = structure-from-motion

Introduction
Ten years ago, Zihl et al. (1983) reported the case of patient
L.M., who had a severe disturbance of movement vision
following damage to the posterior sectors of both hemispheres. L.M. showed a deficit for movement perception in
several tasks. She could see movement in light targets moving
horizontally or vertically below 10-14°/s, but with higher
velocities she was uncertain whether targets moved or were
stationary. In such conditions she reported the appearance of
the target at successive positions with no movement in
between. Similarly, she could follow with her eyes a target
moving at 4°/s, but at 8°/s had saccadic pursuit. She could
not track the trajectory of her own right index finger if it
moved 'too fast'. L.M. underestimated the speed of targets
moving at 127s and 247s, but not at 37s or 67s. She also
had a defect of the phi mechanism, which normally allows
© Oxford University Press 1995

us to perceive apparent motion between successive static
frames in an animation sequence, such as a Hollywood
movie. Furthermore, L.M. had reduced perception of motion
after-effects, and the spiral after-effect elicited no sense of
motion-in-depth. Finally, she reported seeing changes in
position but not movement-in-depth when a black cube lying
on a table was moved toward and away from her.
The motion vision disturbance in L.M. was relatively
selective: she had good visual acuity and no reported visual
field defect for form or luminance. Thus, she could detect
stationary light targets (0.3-320 cd/m2) presented binocularly
in a Tubingen perimeter. Tactile and acoustic motion
perception were reportedly normal. Similarly, thresholds for
resolving the temporal flicker of a light presented at several
locations in the fovea and peripheral fields, and the amplitudes

1106

M. Rizzo et al.

and latencies of visual evoked potentials generated in response
to contrast reversing checkerboard stimuli were all normal.
Static binocular stereoacuity and manual reaction times to
visual targets were modestly impaired, yet there was no
neglect of visual targets flashed simultaneously in both
hemifields, and electro-oculography showed accurate
localization of visual targets by saccadic eye movements.
Moreover, the striking preservation of face and object
recognition, reading, and colour vision distinguished L.M.
from subjects with damage in ventromesial visual association
cortices, located below the calcarine fissure in the occipital
lobe and in adjacent inferotemporal regions. Zeki (1991)
labelled L.M.'s deficit 'cerebral akinetopsia', mimicking the
term 'cerebral achromatopsia', used to describe the human
cortical colour processing defects.
The case of L.M. was important in spurring further research
on a cortical substrate for motion perception. The lesions in
this benchmark case (Zihl et al., 1991) and in subsequent
subjects with reported motion perception deficits (e.g.
Thurston et al., 1988;Vaina, 1989; Vaina etai, 1990; Beckers
and Homberg, 1992; Morrow and Sharpe, 1993; Plant and
Nakayama, 1993; Plant et al., 1993) include structures in an
occipito-temporo-parietal and a parieto-occipital location.
These dorsolateral visual association cortices coincide with
loci of activation identified in PET studies in normal subjects
viewing moving stimuli (Corbetta et al., 1991; Watson
et al., 1993). These cortical areas include Brodmann's (1909)
classical cytoarchitectonic area 19 and adjacent area 37. From
criteria of myelogenesis (Flechsig, 1901), myeloarchitecture,
and callosal connections, part of these areas may resemble
the monkey mesial temporal (MT) area (Allman, 1977; Clarke
and Miklossy, 1990), also known as V5 (Dubner and Zeki,
1971; Zeki, 1974). Area MT (V5) contains neurons that
are sensitive to stimulus direction, speed, orientation and
binocular disparity (Maunsell and Van Essen, 1983 a, b),
that are active during perceptual decisions on these properties
(Salzman and Newsome, 1994), and that probably contribute
to motion direction discrimination (Newsome and Pare, 1988),
and even shape-from-motion perception (Schiller, 1993).
While the anatomical locus and extent of brain lesions in
patients with defective motion vision is an active research
topic, the range of perceptual deficits and residual abilities
in affected patients remain under-explored. Moreover, the
main indices used to diagnose defective motion processing
have varied widely. For example, some investigators have
used smooth pursuit eye movements alone to infer a motion
vision defect while performing no afferent testing of motion
(Thurston et al., 1988). Furthermore, motion plays many roles
(Nakayama, 1985), and consequently cerebral akinetopsia
should comprise a range of motion-related defects. Profiling
these defects would be expected to provide more insights on
the nature of cerebral akinetopsia and on the role of a human
cortical network for motion perception. To date, L.M. is
perhaps the akinetopsic subject whose visual deficits have
been studied in the greatest detail (Zihl et al., 1983; McLeod
et al., 1989; Hess et al., 1989; Baker et al., 1991). The

experiments of Hess et al. (1989) and Baker et al. (1991)
are of key importance and raised several issues that motivated
the current study.

Motion discrimination
Hess et al. (1989) noted that normal motion vision requires
first that spatial and temporal filters be intact, and second
that outputs of these filters be available for comparison. They
determined that L.M.'s contrast sensitivity (CS) for simple
detection of sine wave gratings showed only small loss even
for moving gratings, but CS dependent on discrimination of
motion showed a large loss. Also, spatial frequency and CS
discrimination were much better than temporal frequency or
velocity discrimination for the same moving and stationary
stimuli. The findings suggested that L.M. has a motion
performance deficit affecting judgment of stimulus and
motion attributes, not the detection of motion per se.
To determine whether L.M.'s residual performance was
due to inferences from non-motion cues, such as counting
the passing bars per unit time in drifting gratings, Hess et al.
(1989) used double flash random-dot cinematograms (RDC).
These stimuli depicted two sequential arrays ('frames') of
spatially random dots, the second being identical to the first
but displaced slightly to the left or right. L.M.'s judgment of
the apparent motion in these stimuli exceeded chance only
over a restricted range of spatial displacements. Also, her
performance deteriorated rapidly with frame durations of
<500 ms, whereas normal subjects do well with exposures
of much more than 100 ms. However, inter-stimulus intervals
(ISIs) of >40 ms degraded performance both for L.M. and
for normal subjects. Since this latter behaviour characterizes
'short-range' motion mechanisms, the authors inferred that
the 'residual motion perception in L.M. is carried out by
short-range mechanisms'.

Trouble separating signal-from-noise
Unfortunately, the two-flash RDC used by Hess et al. (1989)
contain spurious spatiotemporal frequency components
(noise) which might have interfered with the perception of
motion (Baker et al., 1991). Conversely, the frame durations
of 500 ms or more at which L.M. 'saw' motion might permit
indirect inference of motion by scrutiny of random-dot
clusters, another case of 'seeing' motion mediated by nonmotion processes. Consequently, to improve the estimates of
residual motion processing abilities in L.M., Baker et al.
(1991) used several additional versions of random limited
lifetime RDC stimuli. A 'Movshon noise' stimulus plotted
spatially random 'motion' dots every 200 |is over a total
stimulus duration of 1000 ms. Between frames, a dot would
'live' or 'die' with a certain probability or 'coherence value'.
A dot that lived was replotted a specific distance to the left
or right of its previous location; one that died was replotted
at a new random location. Using this procedure, Baker et al.
(1991) found that L.M.'s lesions caused a tremendous increase

Motion and shape perception
in the signal needed for L.M. to determine the direction of
global flow compared with that needed by normal subjects.
Surprisingly, when the motion stimulus used stationary dots
for noise, L.M. displayed a severe defect over a range of
lifetimes and displacements. To explain this result, Baker
et al. proposed that stationary dots might effectively reduce
the contrast of motion signal dots even though L.M. has near
normal CS for static stimuli.
Effects of dot noise and lifetime were also examined using
more explicit definitions of 'signal' and 'noise'. When signal
dots jumped together for a specific number of frames (a
lifetime) and noise dots jumped to new random locations in
each frame, L.M.'s performance declined precipitously as
noise was added, even for longer dot lifetimes. To determine
whether the trouble was due to decreased signal or increased
noise, L.M. viewed an identical display void of noise dots
('invisible noise') wherein the notion of low coherence refers
to a decreased number and density of signal dots. The results
showed dramatically better performance extending down to
low dot densities. Additionally, when the effects of dot
lifetimes were assessed at 100% coherence, L.M. had trouble
for short dot lifetimes, but not enough to account for her
trouble on the motion noise experiment. In other words,
L.M.'s deficit was due more to trouble handling degradation
by noise, than to trouble perceiving short lifetimes.

Specific aims
The current study follows up on the issues raised by previous
studies of L.M. and also examines several additional questions
to better characterize the perceptual deficits and mechanisms
affected in cerebral akinetopsia. We have four specific aims,
(i) To test the hypothesis of Baker et al. (1991) that
cerebral akinetopsia is preceded by an inability to extract
signal-from-noise. To address this issue we evaluate the
perception of global coherent motion and direction
discrimination in L.M. using random-dot stimuli constructed
with background noise. The stimuli in Experiment I are
similar to the limited-lifetime random-dot stimuli used by
Baker et al. (1991) but provide a more precise definition of
signal and noise and use a single dot movement speed. Other
data relevant to this hypothesis are collected in Experiments
2 and 3, which vary the signal-to-noise ratio in tasks of 2-D
and 3-D object perception.
(ii) To test the perception of 2-D shape-front motion and
3-D SFM in cerebral akinetopsia. There is ample evidence
that the visual system can use motion cues to extract
information on 2-D figure-from-ground (Frost and Nakayama,
1983; Regan and Beverley, 1984; Braddick, 1993; Stoner
and Albright, 1993) and even 3-D kinetic depth (Wallach
and O'Connell, 1953; Braunstein, 1962). We hypothesize
that the global motion processing disturbance in cerebral
akinetopsia should also impair these perceptual abilities.
Experiment 2 evaluated the perception of SFM using dynamic
random-dot and random-line depictions of 2-D and 3-D
objects. Experiment 3 evaluated the perception of 2-D shapes

1107

using a letter identification task in which the letter targets
were defined by motion, luminance transients, static or
dynamic stereopsis, or differing static textures.
(iii) To test if cerebral akinetopsia affects visuoperceptive
abilities other than motion. Relevant evidence on this issue
was collected in Experiment 3 as well as in Experiment 4
which used a battery of standardized tests of static visuospatial
perception.
(iv) To evaluate the abnormalities in the apparent (phi)
motion perception previously reported in L.M. (Zihl et al.,
1983). We were interested in the hypothesis of Baker et al.
(1991) that L.M.'s trouble reflects an impairment of 'longrange' rather than 'short-range' mechanisms (Braddick,
1980). We addressed these issues in Experiment 5 by varying
the inter-stimulus interval between the presentation of two
frames in a bistable apparent motion stimulus known as the
Ternus display (Pantle and Picciano, 1976). We also wanted
to determine the relative contribution of the long- and shortrange apparent motion systems to the bistable percepts
experienced with this display.

General methods
Subject
The case of experimental subject L.M., a right-handed woman
who had a secondary education, is fully described in previous
articles (Zihl et al., 1983, 1991; Hess and Baker, 1989;
McLeod et al., 1989; Baker et al., 1991; Long, 1992). In
brief, L.M. is now in her early fifties, and well adapted to
her visual deficits. These deficits are the sequelae of cerebral
infarction in 1978, probably due to venous sinus thrombosis.
L.M. has no known underlying systemic disorder to explain
her problem, and has had no new cerebral infarctions to the
time of testing. She is intellectually intact apart from her
vision impairments, and subsequently has been able to adjust
to the demands of daily life, including widowhood. She
lives independently in her own apartment, makes her own
decisions, and leads an active social life, although she has
not been able to return to her premorbid activity of running
an electrical shop. L.M. negotiates stairs, and even rode the
underground by herself, to participate, with her consent, in
a week of experiments. She learned the faces of her new
examiners (M.R. and M.N.) after first exposure, and
recognized them days later, even out of context, from visual
cues alone.

Experimental procedures
The procedures for Experiments 1-5 are outlined in Table 1.
In testing L.M., we took the approach of Hess et al. (1989)
and Baker el al. (1991) in that most tasks used an alternative
forced choice (AFC) psychophysical procedures. Whenever
possible, stimulus duration was unrestricted to avoid
assumptions on L.M.'s short-term memory needed for AFC
procedures, or focal attention needed for brief exposures.

1108

M. Rizzo et al.

Table 1 Experimental procedure
Experiment 1. Perception of a global coherent motion direction.
Experiment 2. Perception of SFM.
Screening: identification of 3-D SFM figures depicted rotating about the an axis (random-dot spheres, cubes, and a rndom-wire
figure)versus non-figure foils (2-D and 1-D random noise in a circular aperture, a coherent sheet of moving dots, transparent limited
version of 2-D random noise).
Experiment 2A. Identification of 3-D shapes amidst varying levels of background noise.
Experiment 2B. Perception of dynamic stereopsis.
(i) Perception of stereoscopically presented random-dot or random-wire figures rotating about their vertical axis (a sphere, a cube,
a cube canted onto a corner to look like a diamond, and a random-wire figure),
(ii) Perception of stereo versions of the low-, medium- and high-noise stimuli from Experiment 2A.
Experiment 3. Perception of 2-D letter shapes from motion and non-motion cues.
Experiment 3A. Relative motion cues.
Condition 1. Letters defined by stochastically moving a proportion of dots falling within the letter region against a static random-dot
background at different proportions of signal.
Condition 2. Like Condition 1, except stationary dots removed from background.
Experiment 3B. Transient cues—letter shapes defined by 'on' and 'off ' transients falling within the letter region against a static
random-dot background at different proportions of signal.
Experiment 3C. Static texture cues
Condition 1. Letter shapes differentiated from background by orientation of line segments.
Condition 2. Letter defined by small triangles against oriented line segments.
Condition 3. Letter shape defined by differences in dot density.
Experiment 4. Visuoperceptive tasks.
(a) Rey-Osterreith Complex Figure Test—copy and matching versions
(c) Benton Visual Retention Test—Revised. Form C and recognition versions F and G.
(b) Hooper Visual Organization Test.
(d) Mooney's Closure Faces Test.
(e) Cookie theft picture.
Experiment 5. Perception of bistable movement with the Ternus display.

Data were collected in short sessions with frequent intervals
of rest. Successive blocks of data were accumulated to yield
psychometric functions for the different tests. All stimuli,
except for those in Experiment 4, were generated by an
Apple Macintosh computer and presented in black and
white on a monochrome monitor at 100% contrast (white
19.5 cd/m2; black 0.0 cd/m2). The viewing distance was
57 cm. Stimuli were also viewed in similar conditions by
normal control subjects with no cerebral lesions.

Experiment 1: perception of a global coherent
motion direction signal
The study of motion perception requires a psychophysical
stimulus and task design that relies exclusively on the
perception of movement. An observer's ability to infer
movement from noticeable changes in the visual scene, the
way we 'see' movement in the hands of a clock, or between
two successive photographs, must be eliminated. For such
reasons, we test motion perception using computer generated
animation sequences known as RDC. These stimuli present
a motion signal among spatially random background noise,
allow variation of spatial displacement and temporal intervals
at programmable exposure durations, and provide a more

objective measure of residual motion perception than do
measures of Dmm or Dmin.
Random-dot cinematograms depict a number of small
dots moving around within an imaginary aperture upon the
computer screen. While a proportion of the dots move in a
single, cardinal direction (signal dots), the other portion
moves randomly (noise dots). When the proportion of
randomly moving noise dots is low, the entire stimulus
appears to have a global flow in the direction of the
unidirectional signal dots. In this case it should be easy to
determine the direction of signal dot movement. However,
as the proportion of noise increases, the perception of global
flow decreases, and it becomes more difficult to determine
the direction of signal dot movement. Motion perception can
be assessed by determining the proportion of signal necessary
for the perception of the global movement. A larger proportion
of signal indicates that the motion processing mechanism is
less able to integrate the local motion vectors from the local
dot movements into a global coherent motion flow. Therefore,
a high threshold for direction discrimination indicates a
disruption in the neural mechanisms for normal motion
processing.
In Experiment 1, we attempted to confirm the Baker et al.
(1991) result using an experiment including more directions

Motion and shape perception

% Signal

Direction of
Global Flow

0%

o
o

50%

100%

Fig. 1 The random-dot motion stimuli were composed of
randomly positioned dots moving at a uniform speed. Signal dots,
depicted black here, move in one particular direction (rightward
shown). Noise dots, depicted white here, make up a flat
distribution of movement directions so that their net movement is
zero. Net or global stimulus movement is influenced by the
proportion of signal to noise dots, with greater movement
perceived with a higher signal proportion.

of movement and a more precise signal and noise definition
whereby each stimulus comprised a single dot movement
speed.

1109

smaller proportions of signal made the direction discrimination task more difficult.
The perception of coherent motion, and, therefore, correct
responses on the experimental task, depend on a motion
processing mechanism's ability to integrate local motion cues
across the display. This task cannot be completed by scrutiny
of individual dots or groups of dots since the dots were
small, frame and stimulus durations were brief, and the
assignment of dots to a signal or noise distribution displacement varied between frames. A low dot density
minimized the possibility of accidental correspondence
between signal and noise dots on successive RDC frames
(Williams and Sekuler, 1984) and similarly prevented a range
of noise velocities which gives unreliable masking of the
motion signal by the noise signal (Bravo and Watamaniuk,
1992).
Subject L.M. viewed the computer display from 57 cm
while seated in a dimly lit testing room. Trials were initiated
by the experimenter only after L.M. fixated a small cross at
the screen centre. The RDC was presented in the centre of
the screen and L.M. had to indicate which direction of signal
motion was displayed.
A 'warm-up' and learning phase was implemented in two
steps. The first step was to present a group of 15 small
(24'X 10'), like oriented, arrows to determine if L.M. could
report their uniform direction with brief stimulus
presentations. The second step used RDC stimuli ranging
from 90 to 100% to determine if L.M. could perform the
motion perception task. The ratio of signal to noise dots was
varied within a method of constant stimuli. The experimental
phase followed and used RDC stimuli ranging from 10 to
100% signal. By varying the dot displacement (6' and 10')
and frame rate (30 and 15 ms), two speeds were tested,
3.37s and 117s, respectively. These tests were completed
with brief stimulus durations varying from 450 to 195 ms in
different conditions. L.M. completed 24 trials in each step
for each of the two speeds tested. Responses were verbal
or gestural and were recorded on the computer by an
experimenter.

Method
Each RDC animation sequence contained up to 30 frames
depicting 150 randomly placed, small (2X2 min) black dots
moving within a 4°X4° square area. To generate the dot
movements, each dot was displaced a small constant distance
between cinematogram frames. The direction of this
displacement came from one of two different distributions.
A proportion of dots was given a unidirectional displacement.
These displacements were either upward, downward, leftward
or rightward and they gave the signal direction to the stimulus.
The remaining dots were given displacements from a flat
distribution of directions spanning all 360°. This generated
the noise in the stimulus (see Fig. 1). The subject's task was
to indicate the direction of global coherent motion, which
should be determined by the direction of the signal
displacements. The proportion of signal depicted in the motion
stimulus was the main independent variable. Remember that

Results
First, L.M. was able to determine accurately the direction
indicated by the static arrow cues with the brief 450 and
195 ms stimulus durations (94 and 82% correct, respectively).
This demonstrates that L.M. had no overt behavioural deficit
that precluded testing with this particular paradigm. Second,
L.M. was able to detect motion direction at the high
proportions of signal presented in the warm-up and learning
phase (e.g. 94% correct in the 117s condition). This
demonstrates again that L.M. has no difficulty with this
particular motion testing paradigm.
Data from each of the short testing blocks were compiled
to generate one data set per condition. Thresholds for direction
discrimination were determined by a probit analysis.
Figure 2A demonstrates the threshold calculation with the

M. Rizzo et al.

1110

movement (Fig. 2B). Thresholds for older normal subjects
are -15% signal for a 195 ms presentation duration. L.M.'s
thresholds are >5 SDs above normal.

Discussion of Experiment 1

40

Baker el al. (1991) hypothesized that L.M.'s performance
deficit was due to decreased number of neurons carrying
directional information or an impairment in processing stimuli
with low signal-to-noise ratio. This lead us to test the motion
direction discrimination performance in L.M. by using RDC
stimuli that provided a more precise definition of signal and
noise. Our objective in Experiment 1 was to minimize any
accidental correspondence between signal and noise dots
which might produce a false low assessment of L.M.'s motion
direction discrimination ability and to further reduce any
non-motion cues which might allow correct inferences of
global motion direction.

60

% Signal

50 r

50 r-

40

40

i/5 30

30

I
120

10

10

450 375 300 195

Norms

Stimulus Duration (ms)

450 195

Norms

Stimulus Duration (ms)

Fig. 2 (A) L.M.'s performance in the 1 l°/s and 450 ms condition
of the direction discrimination task. The solid line shows the bestfit psychometric function. The value of this function at the point
where it crosses statistical discriminability (horizontal line) is
used to determine the percent signal at L.M.'s threshold for
direction discrimination (vertical line). (B) Left, L.M.'s thresholds
for various stimulus durations in the 3.3°/s condition; right, the
thresholds for the 11 °/s conditions. The performance of normal
controls (n = 22, mean age 70 years) are shown for comparison.
Error bars denote 2 SDs. The results confirm that L.M. has a
persistent and severe motion direction discrimination deficit
(Baker et al., 1989).

arbitrarily chosen 1 l°/s, 450 ms condition. In this condition,
L.M. requires -35% signal dots to determine correctly the
direction of signal dot movement. Thresholds were similarly
determined for other conditions and for normal controls.
Results from the experimental phase confirm that L.M.
has a persistent and severe motion direction discrimination
deficit, similar to what Baker et al. (1991) showed. In both
the 3.3°/s and ll°/s conditions L.M. needed -40% motion
signal to determine consistently the direction of signal dot

Like the earlier stimuli of Hess et al. (1989) and Baker
et al. (1991), our RDC depicted small signal and noise dots
moving within a circumscribed square target area. Also, since
ISI was equal to frame duration time (15 ms) and each
animation sequence contained up to 30 frames, overall
stimulus duration of -500 ms fell squarely in the range of
residual motion performance in L.M., determined by Baker
et al. (1991) as did the size of the dots and amplitude of dot
displacements we used.
However, unlike those earlier stimuli, the RDC in the
current study varied the assignment of a dot to the signal or
noise distributions between every frame. This made it more
difficult to follow the successive jumps of a single signal dot
and maximized the likelihood that a correct response
depended on the ability to pool the local directional motion
cues across the stimulus. As our dots randomly switched
between expressing signal and noise information in different
frames, most signal dots in our RDC had short lifetimes,
particularly at low percentage signal coherence. However,
Baker et al. (1991) had already demonstrated a limited
importance of dot lifetime compared with dot noise effects.
L.M.'s results using our stimuli showed the presence of
severe motion discrimination deficits, resembling the findings
of Baker et al. (1991). Using a similar paradigm in monkeys,
Newsome and Pare (1988) found that the proportion of signal
needed for such a task increased significantly when monkeys
were given acute lesions to area MT. The MT lesions disrupted
cortical motion processing in the monkeys, producing a
performance deficit in the psychophysical task.
However, we did find residual motion direction discrimination in L.M. at the 117s velocity, higher than was
expected. Moreover, at a quantitative level, our results show
a less severe deficit with a threshold of 40% signal compared
with the 80% thresholds measured by Baker et al. (1991).
Consider also that normal thresholds were different: -15%
in our study and only 4% in the Baker et al. (1991) study.
These differences indicate at least two things: (i) that the
non-correspondence of velocity or spurious masking of signal

Motion and shape perception
by motion dots played some role in the signal from noise
deficit measured by Baker el al. (1991); and (ii) that while
L.M. does have a severe motion perception deficit, she has
considerable remaining motion vision, even more than Hess
et al. (1989) and Baker et al. (1991) found. As a consequence,
we wondered if this residual motion perception might be
used to recover SFM or whether this ability was lost with
cerebral akinetopsia. This is examined in the next experiment.

Experiment 2: perception of SFM
The perception of depth or SFM is one commonly cited
ecological use of motion perception (Gibson, 1950;
Nakayama, 1985). Since L.M. demonstrates a remarkable
impairment in the perception of movement, we wondered if
she might also show deficits in the perception of depth or
structure from motion. Several earlier findings in L.M.
suggested this might be the case. First, L.M. could not sense
movement-in-depth on viewing a black cube moved toward
or away from her and reported seeing changes in position
but not velocity (Zihl et al., 1983). Secondly, the spiral aftereffect elicited no sense of motion-in-depth. Thirdly, we note
with great interest the findings of Baker et al. (1991) in their
experimental manipulations of dot noise in a motion direction
discrimination task using RDC comparable to the task we
used in Experiment 1. L.M. saw motion at 100% coherence
but her performance rapidly declined {see Baker et al., 1991;
fig. 4, p. 457). Owing to the design of their stimulus, in the
high percentage coherence range where L.M.'s performance
declined, normals easily saw two transparent surfaces that
were separated in depth. L.M., however, did not perceive
this kinetic depth (incidentally, these cues were not available
in the motion stimuli we used). Thus, Baker et al. (1991)
produced a very important, if unintended, demonstration of
possible abnormal SFM in L.M.

1111

Part of the design was to determine whether L.M. would
spontaneously generate such reports. Each of these stimuli
was presented continuously until L.M. gave a response.
L.M. could readily report the structure of the SFM figures,
similar to the patient A.F. of Vaina et al. (1990) who gave
an accurate report of the structure of a rotating cylinder at
high signal coherence. L.M. effortlessly described the spheres
and cubes as voluminous objects and described the random
line figure as rigid object resembling 'a bunch of rotating
triangles'. L.M. reported that these figures appeared solid
and to have depth. Of equal importance, L.M. did not report
seeing a rigid 3-D structure in the presentations of the foils
and saw no depth or structure in viewing the 2-D noise. She
accurately described the pair of random-dot sheets, but did
not report perceiving any depth between them.
While some normal observers report seeing a gelatinous
sphere interspersed with dots in the 1-D random-noise display
(Williams and Phillips, 1987; Treue et al., 1991), or a barber's
pole rotating and moving upward in the directionally limited
2-D random-noise display (Williams and Phillips, 1986),
L.M. did not. The 1-D random-noise display is a randomdot display wherein dots can move randomly back and
forth along one dimension, for instance, horizontally. In the
directionally limited 2-D random motion, dots move randomly
in directions 65° either side of upwards. In these cases, the
perception of a rigidly rotating figure has been taken to
suggest cooperative interactions among velocity-selective
motion processing mechanisms (Williams and Phillips, 1987).
L.M.'s inability to perceive depth or form in these stimuli
underscores her inability to integrate stochastic motion
information over space. However, in more orderly displays,
L.M. has no difficulty in recovering 3-D SFM. The following
experiments more closely examine L.M.'s perception of SFM
in stimuli with a varying noise component.

Screening procedures—identification of 3-D
SFM figures among non-depth foils

Experiment 2A: two-AFC shape identification
with varying amounts of background noise
Method

In preliminary testing, L.M. correctly identified several 3-D
SFM figures presented among several non-depth foils.
Computer generated SFM figures were presented in
orthographic projection on the face of the computer monitor.
The figures included: transparent random-dot spheres, opaque
random-dot spheres, random-dot cubes (lying flat or standing
on a vertex), and a multisegmented 'random-wire' figure.
The figures were shown rotating about either the horizontal
or vertical axes. The non-figure foils included: 2-D random
noise within a circular aperture, 1-D random noise in a
circular aperture, a coherent sheet of moving dots, a pair of
transparent random-dot sheets moving across each other in
opposite horizontal directions, and a directionally limited
version of 2-D random noise. As discussed in detail below,
some of these foils may elicit a report of non-rigid structure
or depth between the different directions of dot movement.

For an objective test of L.M.'s ability to perceive SFM, we
developed a two-AFC shape identification paradigm where
the observer had to report the shape of the object presented
in each particular trial. Accurate performance on this task
depends on the subject's perception of the figure's shape
from motion. To prevent shape identification from nonmotion cues, such as edges or dot density, and also to index
the difficulty of the task, varying amounts of random-dot
noise were added to a square background region surrounding
the target.
The SFM figures were a random-dot sphere and a randomdot cube canted 45° about the x- and z-axes so as to stand
on a corner. The figures rotated about either the horizontal
or vertical axes. The two stimuli were of comparable size
(-2.8° visual angle in diameter) but could be easily
distinguished by the SFM cues in low noise conditions. These

1112

M. Rizzo et al.
100

the performance of normal controls (n = 8, mean age 62
years). Normal threshold level for discrimination, shown with
the horizontal line at 75% correct, is slightly more than 10%
signal. In contrast, L.M. did not begin to discriminate reliably
between the two SFM figures until the signal was near 30%.
This is a significant threshold elevation compared with
normals. The high stimulus noise and low stimulus signal in
this set of stimuli severely affects L.M.'s ability to recover
SFM. Of course, it is important to note that in less noisy
conditions, L.M. performs normally. Unless the stimulus is
obscured by a high proportion of noise, L.M. has no difficulty
perceiving SFM.

o
(0

u

90

iC

2

80

in

70

o
U

60

50
10

15

20

25

30

% Signal (100 - %Noise)
Fig. 3 For the kinetic depth shape discrimination experiment,
normal discrimination performance is shown by the filled circles
and the best-fit psychometric function. Normal threshold is at
-11% signal. L.M.'s performance is shown by the bold line. Her
performance fluctuated at about chance (50% correct) but her
psychometric function shows evidence of shape discrimination by
30% signal where performance exceeds 75% correct (dotted line).

figures were presented against a square background region
8°X8° visual angle. Within this background region either 20,
200 or 1000 small (2'X2') white dots were presented, giving
a nominal dot density of < 1,3 or 16 dots/square degree. Noise
dots moved with a speed of ~3°/s, which is approximately
the speed that figural dots moved when fronto-parallel to the
observer.
The signal to noise ratio of dynamic random-dot elements
comprising the figure and background noise were varied. The
low (10%) noise stimulus had 200 figure dots and 20
randomly moving noise dots comprising the background. The
medium (50%) noise stimulus had 200 figure dots and 200
randomly moving background noise dots. The high noise
(70-95%) condition had a 1000 noise dot background and a
SFM figure composed of 50-300 dots.
In the low, medium and high noise conditions, L.M. was
given four presentations of each of the two shapes (cube and
sphere) rotating about each of the x- and y-axes. This gave
16 observations in the low and medium noise conditions and
48 observations in the high noise condition (six signal levels
between 5 and 30%). In each presentation, the subject viewed
one complete revolution of the figure, depicted in 180
sequential cinematogram frames presented with a 30 ms
frame rate. The stimulus presentation lasted 5.4 s. Within
each condition, stimuli were presented in a predetermined
random order.

Results
With only 10% noise in a sparse 220 dot display and
50% noise in a 400 dot display, L.M. performed at 100%
discrimination between the two shapes. However, with a
higher dot density, and fewer dots defining the object, L.M.
performed poorly. Figure 3 shows L.M.'s performance and

Experiment 2B: dynamic stereopsis
The perception of structure in a moving display is aided by
binocular disparity cues. The perceived depth magnitude of
displays containing both disparity and SFM is reported to be
much greater than perceived with either of these cues alone
(Tittle and Braunstein, 1993). Considering that L.M. has only
marginal static stereo acuity (140 min measured with the
Randot stereotest) and shows some deficit in the perception
of SFM in noise, we wondered if L.M.'s performance in the
previous shape discrimination task might be aided by the
addition of disparity information (Rouse et al., 1989).
Alternatively, it is possible that L.M.'s deficits in motion
perception and static stereopsis might also include dynamic
stereopsis. In this case disparity information would not confer
an advantage for the perception of SFM. To address L.M.'s
use of dynamic disparity information, we examined whether
L.M. could use dynamic stereopsis for: (i) disambiguating
the direction of rotation in an otherwise rotationally
ambiguous SFM figure; and (ii) improving performance on
the high noise condition of the shape identification task
used previously. In the second condition, binocular disparity
information should make the noise dots appear at a plane
behind the rotating figure. With this depth separation between
signal and noise, figural discrimination should be easier
and performance improved compared with the condition of
motion alone.

Method
In both conditions, computer generated figures were viewed
through a liquid crystal stereoscopic shutter system. This
stereoscopic viewing system allowed the presentation of
different views to each eye as alternating stereo frames were
drawn to the computer screen. The alternating frames had
slightly different views of the stimulus display. Although the
frames were presented asynchronously to the two eyes, this
system generates the essential disparity conditions necessary
for binocular stereopsis.
In the first experiment, L.M. viewed several stereoscopically presented random-dot or random-wire figures
rotating about their vertical axis. The figures were a sphere,
a cube, a cube canted onto a corner to look like a diamond,

Motion and shape perception
and a random wire figure (Nawrot and Blake, \99\b). The
relative disparity between the front and rear surfaces of the
figure was 4', 8' or 12', or the front and rear surfaces had
2', 4' or 6' crossed and uncrossed disparity, respectively. By
virtue of the disparity information, the previously ambiguous
rotation of these SFM figures is rendered unambiguous: the
disparity information specifies a particular depth arrangement
in the stimulus and therefore a particular direction of rotation.
L.M.'s task was to indicate the direction that the figure
appeared to rotate. From this response we can determine
which surfaces of the figure were perceived as front and rear,
and whether this corresponds to the disparity information
presented in the stimulus.
In the second experiment, L.M. viewed stereoscopic
versions of the low-, medium- and high-noise kinetic depth
stimuli described earlier in this experiment. The stimulus
parameters were essentially identical, with the same dot
numbers, densities and stimulus sizes. Again, white dots were
presented against a black background. For the stereoscopic
presentation, two stereo-pair animation frames were generated
for each figure position in the animation sequence. First,
background dots were given 6' of uncrossed binocular
disparity so that all the noise dots appeared on a plane behind
the random-dot figure. The dots on the plane still move about
randomly on the plane, but did not move in and out of the
plane in depth. Secondly, dots defining rotating figure were
drawn with varying amounts of binocular disparity such that
the front surface of the figure had 6' crossed disparity, the rear
surface 6 min uncrossed disparity, and a smooth continuum in
between along the surface of the figure. Again, the disparity
information defined two different directions of rotation about
either the x- or >>-axis. Figure 4 illustrates how the randomdot figures appeared in front of a plane of noise dots. Of
course, in the actual experiment the figure rotated, and the
noise dots moved randomly about their plane.
In all conditions, L.M. viewed one rotation of the figure,
which gave a stimulus presentation duration of 5.4 s. Using
the liquid crystal display (LCD) shutters, each eye was
presented their respective frame of the stereo pair for 15 ms,
while the other eye was occluded. An entire stimulus comprised
360 frames, of which each eye viewed only their respective
180 frames.
In the first direction of rotation condition, L.M. viewed 16
figures with each of the three amounts of disparity for 48
presentations total. In each of these presentations, L.M. had
to signal the direction that the figures appeared to rotate. In
the low- and medium-noise conditions, L.M. viewed 16
presentations of the various stimuli (two trials for each shape,
axis and direction of rotation). In the high-noise condition,
L.M. viewed 24 presentations with only the y-axis of rotation.
In this experiment, L.M.'s only task was to identify whether
a cube or a sphere shape was being presented.

Results
L.M. reported little appreciation of a solid 3-D object in
the dynamic stereopsis displays. She showed 0% rotation

1113

20% Signal

JV;;.K^vv::i^>;:
50% Signal

""

•

„ • • ' • •*• . • * . ' • "

'

:

.>'*'•'.•

- ' " * '

••

•

'

' - "
' • • •

•

•

•

•

.

.

'

Fig. 4 For the binocular stereopsis shape discrimination
experiment, a spherical or cubic shape was hidden amongst a
background of noise dots. With binocular viewing the figures
appear in front of the noise and the shapes are easily
discriminated. For illustration, the stereo-pairs here may be
viewed with crossed eyes (free fusion) for appreciation of the
depth and shape discrimination task used in the experiment. (A)
A cube depicted with 20% signal dots and 80% noise dots; (B) a
sphere depicted with 50% signal dots and 50% noise dots.

direction discrimination for all disparities tested. In all trials
of the first experiment, L.M. reported that the front of the
figure was moving leftward (clockwise rotation of the figure
in top-view). The disparity cue had no effect on the reported
direction of rotation. We noted in further testing, however,
that L.M. reported spontaneous reversals in the direction of
figural rotation of a (non-stereoscopic) SFM figure when
presentation durations were 15 s or longer. In other words,
like normals, L.M. can perceive both rightward and leftward
rotation in SFM figures. She is simply unable to use dynamic
binocular disparity information to disambiguate direction
of rotation.
In the stereoscopic version of the shape identification task,
L.M.'s performance was similar to that in the kinetic depth
version. L.M. had no trouble identifying the objects in the
low- and medium-noise version of the task (100% and 88%
correct, respectively). In the high-noise version of the test,
L.M.'s managed reliable discrimination between the objects
at 30% signal. This is similar to the threshold obtained in
the kinetic depth stimulus in Experiment 2A. In other words,
the addition of dynamic disparity cues in this experiment
failed to convey any advantage in the perception of 3-D
form. As the information from the kinetic depth experiment
was available and L.M.'s performance was similar, motion

114

M. Rizzo et al.

information must have been the sole source of shape
information that L.M. used. Therefore, it appears doubtful
that L.M. uses dynamic stereoscopic information in the
perception of depth.
In an ancillary test of L.M.'s use of binocular disparity
information, L.M. was presented with a static stereoscopic
version of the medium noise stimuli using the LCD shutter
system again. Here, with 16 presentations again, L.M.'s
discrimination performance dropped to 69% correct. With a
two-alternative force choice procedure, this is only 38%
discrimination. Normal discrimination is near 100%.
Therefore, it appears that while L.M. has difficulty recovering
SFM information, binocular disparity information conveys
no additional benefit.

Discussion of Experiment 2
The ability to recover form- and depth-from-motion allows
us to segment images into different objects at different
depths based on relative movement. This perceptual ability
is comparable in man and monkey (Siegel and Andersen,
1988), and probably depends on the operation of motion
processing units in visual association cortex. In fact, a
possible network for the processing of kinetic depth can be
constructed from motion processing units whose properties
are modelled after those of neurons identified in the monkey's
area MT complex (Nawrot and Blake, 1991a). Damage to
these processing units in primate area MT leads to errors in
recovering shape-from-motion (Schiller, 1993). Consequently, it was important to examine the ability to recover
form- and depth-from-motion in L.M. who has a welldocumented motion perception deficit due to damage in
neural units responsible for processing motion.
Based on the results of Experiment 2, we find evidence
that L.M. does have residual perception of 3-D SFM. L.M.'s
ability to identify rotating wire frame figures from successive
2-D projections is not simply a question of interpolating
perceptual constancies from a number of learned 2-D retinal
views. This is because she can perceive the volume even in
a completely unfamiliar figure. Moreover, she can perceive
3-D volume from 2-D RDC displays in which there are no
successive familiar views. However, the performance in L.M.
is not normal for the same reason that she failed to perceive
a directional motion stimulus in Experiment 1. Namely, her
perception of 3-D form is very sensitive to the presence of
noise amongst signal, which was evident when she had to
identify a sphere or cube defined by moving dynamic random
dots in Experiment 2A.

Comparisons with other cases
Aspects of the current findings in patient L.M. resemble
those in patient A.F., who reportedly had bilateral lesions of
the temporo-parieto-occipital junction (Vaina et al., 1990)
and could still perceive 3-D SFM despite other motion
processing deficits, and appear to differ from the report of

four patients with right parieto-occipital lesions (Vaina, 1989)
with total inability to perceive 3-D SFM. These latter patients
failed to identify and name a 3-D object defined by the
projection of randomly spaced dots lying on the surface of
an imaginary rotating cylinder, which A.F. could still do.
Instead of reporting a rotating cylinder they reported ' "a
bunch of dots moving", "mosquitos or flies", "birds high
in the sky" ' (Vaina, 1989), ' "snow blown by the wind",
or "ants crawling on the ground" ' (Vaina et al., 1990). The
findings were discussed in relation to the 'parieto-occipital'
lesions in L.M. (Vaina, 1989; Vaina et al., 1990); however,
we find it difficult to make direct comparisons.
The seemingly worse performance on a 3-D SFM task in
Vaina's right parieto-occipital patients may reflect several
factors. First, the epoch of testing was only 6-9 weeks post
ictus versus several years in L.M.: testing in the chronic
phase generally shows some recovery. Secondly, there are
issues of anatomy: the right parietal lesion localization in
Vaina's cases relied on brain CT, but no images were
published and no other localizing details were given. The
lesions may have actually formed part of larger unilateral
or even bilateral hemispheric lesions not detected by CT
performed just after a cerebrovascular event, as was ultimately
reported in A.F. A third factor is the design and
implementation of the psychophysical task. Vaina tested her
patients with only a single trial at one low signal density.
The entire stimulus comprised only 16 random dots.
Performance on more trials at higher densities may have
revealed residual ability as we found in L.M.
A.F. was tested on multiple trials using a more sophisticated
version of a rotating cylinder. The stimulus plotted 128 dots
in a 3°X3° field. In one version, percent signal was varied
at 400 ms dot lifetime. In a second version, stimulus duration
(dot lifetime) varied at 100% signal. Compared with II
control subjects, A.F. could perceive 3-D SFM, but this
performance was sensitive to noise, and deteriorated at 42%
signal or less, resembling the current findings in L.M.
Performance in A.F. deteriorated below a stimulus duration
of 400 ms, resembling the findings in the dot lifetime
experiments in the motion direction discrimination
experiments in L.M. by Baker et al. (1991). However, there
are questions concerning the localization of the lesions that
underlay the motion deficits in A.F. which hinder anatomical
comparisons with the case of L.M..
A.F. was a 60-year-old left-handed man who had an acute
right hemisphere haemorrhagic infarction by CT
(unpublished), a dense left visual field defect for luminance
and form, impaired visuoconstructive ability, and impaired
static stereopsis, but preserved visual acuity, contrast
sensitivity and colour discrimination. An MRI was obtained
3 months after the ictus and selected T2-weighted transverse
sections were reported to show bilateral lesions of the cortex
at the temporal-parietal-occipital junction (Vaina et al., 1990,
p. 354). Yet, the published images (Vaina et al., 1990, fig.
1, panels C-F, p. 355) show no lesion reaching the lateral
surface of the left hemisphere. There is a haemosiderin

Motion and shape perception
deposition in the right hemisphere consistent with chronic
haemorrhagic changes, but the cortical component of the
lesion is dorsal and posterior to the temporal-parietal-occipital
junction and its location is mostly parieto-occipital (sections
E and F). Most evident are bilateral patchy subcortical white
matter signals consistent with A.F.'s history of long-standing
untreated hypertension. Perhaps these lesions undercut white
matter connections to and from a human MT (V5) homologue,
as reported in a patient with a unilateral right hemisphere
lesion (Rizzo et al., 1992). Yet, A.F.'s right parieto-occipital
lesion should have been sufficient to impair motion perception
judging by the report of Vaina (1989). Consequently, why
A.F. performed better on a 3-D SFM task than Vaina's four
right parieto-occipital patients is puzzling, since A.F. also
had a right parieto-occipital lesion. One possible explanation
lies in the different 3-D SFM stimuli administered. Another
possibility is that Vaina's (1989) patients had lesions that
were more extensive than the initial CTs showed, similar to
the pattern reported in A.F. based on the later MRI. In any
case, A.F.'s lesions differ from L.M.'s in that AF's were
primarily subcortical and probably more diffuse.

Experiment 3: perception of 2-D shape from
motion and non-motion cues
Having determined in Experiment 2 that L.M. has residual
but deficient processing of 3-D SFM, we undertook to
evaluate her ability to perceive 2-D shape from motion
and non-motion cues. A great many visual cues provide
information about 2-D shape. These cues include luminance,
texture, colour and even motion boundaries. Of these cues,
shape from motion is found to be compromised following
posterior parieto-temporal brain lesions (Regan et al., 1992).
Given the location of L.M.'s cortical lesions and the extent
of her motion perception deficits, we wanted to determine if
her ability to detect 2-D shape from motion was likewise
impaired.

Method
To minimize changing procedural demands on the observer,
a single behavioural task was used to test the perception of
2-D shape from several different types of visual cues. The
task, an adaptation of the procedure introduced by Regan
and Hong (1990), was a simple five AFC letter identification.
In each condition, a single independent variable (the
proportion of a particular visual cue) was used to define a
single letter amongst a background of noise. The letter shape
was defined by texture, luminance, motion, motion direction,
motion speed, luminance transients, and static and dynamic
stereo (Nawrot et al., 1995).
A dark square background region, 4.2°X4.2°, was filled
with 4000 randomly positioned 1' XI' white dots.
Camouflaged within this background region was one of five
letters, H, O, T, E or L. The length of each letter stroke was

1115

2° and the width was 0.47°. Each letter was made from two,
three or four of these strokes. All letters were block type and
contained within a 2°X2° area which had five possible
positions within the background region.
To make these letter-shaped areas visible, a local change
in some particular aspect of the dots, or their movement, was
applied to the otherwise random-dots falling within that area.
For instance, movement of dots within the letter-shaped area
makes the letter visible against the stationary dots in the
background region. To reduce the utility of any incidental
cues, letters could appear in the centre or offset 0.66°
diagonally towards one of the corners of the background
region.
In a warm-up and learning phase, L.M. could effortlessly
identify black block letters against a white background in
this experimental design with a 500 ms stimulus presentation.
Given this performance for identifying simple letter shapes
in brief presentations, it was clear that L.M. did not have
attentional difficulties to interfere with her performance of
this task.

Relative motion cues
Condition 1
Letters were defined by stochastically moving a proportion
of dots falling within the letter region against a static randomdot background (Fig. 5A). The proportion of moving dots,%
signal, was varied across trials between 70, 90 and 100% in
a method of constant stimuli. L.M. completed 75 trials in
this condition.

Results
Normal subjects require only 1% signal to perform the task
accurately. L.M.'s performance was 32, 52 and 68% correct
for the three signal levels. This gave L.M. a threshold of
close to 100% signal for letter identification with this cue.
Quite clearly, L.M. has a very significant deficit in the
perception of 2-D shape from motion against a stationary
background.

Condition 2
In condition 1 above, L.M. not only gave poor performance
but also complained that the stimulus movement appeared
'troubling to look at'. To pursue this 'troubling' aspect of
L.M.'s perception of the stimulus, we removed the stationary
dots from the background region. Using only a signal
proportion (moving dots) of 100% within the letter-shaped
area, this new stimulus appeared to be a stationary lettershaped aperture in front of a field of dots moving together
in a single direction (Fig. 5B). The stimulus had a plain black
background. L.M. completed 25 trials in this condition.

1116

M. Rizzo et al.
shape, even though stationary dots in the surround were
removed. In another attempt to address this interesting aspect
of L.M.'s perceptual difficulties, the contingencies of the
previous experiment were reversed. Here the letter-shaped
area remained blank, as if occluded, while dots in the
background region moved together in one of the four cardinal
directions (Fig. 5C). Like the previous condition, large-scale
luminance cue and accretion/deletion boundaries could be
used for letter identification. However, this stimulus had no
dots or dot movement within the letter-shaped area to be
identified. L.M. completed 25 trials in this condition.

. • : • • • • • • • • • : • . . • •

C
i 1 , 11

L~= .I".".",'!, 1 , 11 , •:.'•.••.•;:
' l I I ' I 'l1 l | l | | l

: -

* . ' •••.'

• '

D
•

. • : • • • • . • • . - . • •
•• • . • • • ' •
•
•••.
• • • •• • • • • •
• •
• • • • • % • •
••
•
• * • • • • • •
•

• • • . • : • : : • . : : • . •
• • • •
"

• •
•

•

•
• •

•

1

*

•

•

• •

:

•>

• • • •
•
• • ••

•

•

m

•

•••••.
• • •

4

•

•

.

•

•

: • . • : : : • :
•

*

•
• •

•••• •

•
•

,

* • •

•

•
• •

•
•

•
• •

Fig. 5 Experiment 3 used several different shape-from-cue stimuli
to examine L.M.'s ability to recover 2-D shape from motion and
other cues. The letter shapes were made visible by: movement
against a static background (A); movement against a blank
background (B); a blank letter area against a moving background
(C); different orientations of texture elements (D); different types
of textures (E); different local dot densities (F); binocular
disparity (G). A representation of depth in G may be seen by free
fusing the pair of figures. An 'H' should appear in front of the
background.

Results
Here L.M.'s performance improved compared with condition
1. She got 80% of the trials correct compared with 68%
correct in the previous condition. As normal observers have
no difficulty performing at 100% correct, this performance
still indicates some difficulty in shape perception. Of course,
a number of cues other than the movement of dots might be
used for the identification of letter shape. For example, large
scale luminance differences would have been an effective
cue, as would be the accretion and deletion boundaries caused
by the dots appearing and disappearing at the edge of the
letter shape. It appears that the 'troubling nature of movement'
was again a factor in L.M.'s performance; i.e. while L.M.'s
perception of motion is clearly impaired, the mere
presentation of a moving stimulus to L.M. interferes, at some
level, with the processing of other visual information.

Condition 3
In the previous condition, movement within the letter-shaped
area had an effect on L.M.'s ability to identify the letter

Results
L.M.'s performance improved compared with the previous
conditions to a near perfect 96% correct (one error). Her
performance in this condition suggests that movement in the
background region does not interfere with letter identification
as does movement in the letter-shaped area itself. Since we
assume L.M. was attending to the letter-shaped region, one
possible explanation is an accompanying attentional deficit
related to the perception of motion. Movement within L.M.'s
focus of attention disturbs other visual processing, while
movement in the surround may not. As we summarize below,
this tenuous explanation requires more study.

Other conditions
Several other relative motion conditions were attempted with
L.M. She was unable to perform the letter identification task
when dots in the background and letter moved in opposite
directions, orthogonal directions or at different speeds. L.M.
simply found these conditions too difficult, and again too
troubling to even complete the small number of trials required
for each test.

Transient cues
This condition of the letter identification experiment used
transient visual cues that did not have a movement component
per se (Kulikowski and Tolhurst, 1973). The stimuli were
similar to those used in the previous conditions. The
background region was filled with 4000 randomly placed
dots. An instantaneous accretion or deletion (in different
conditions) was applied to a proportion of dots within the
letter-shaped area. This addition or deletion of dots was
applied such that the maximum dot density within the lettershaped region was equal or lower to that in the background.
To perform the letter identification task, the observer had to
detect an instantaneous transient signal mid-way through the
500 ms stimulus presentation. Preliminary testing indicated
that it was necessary to use a signal proportion (the proportion
of dots appearing or disappearing) of 50%. L.M. completed
25 trials in each of the two conditions.

Motion and shape perception

Results
Similar to the relative motion conditions, L.M. performed
poorly compared with normal controls. With 50% of the dots
within the letter region being deleted, L.M. performed at
only 44% correct. She performed at 32% correct when the
dots were being flashed on. The threshold for normal subjects
is in the order of 2-5% signal. L.M.'s threshold could not
be determined. At the high proportions of dot accretion or
deletion needed by L.M., relative density cues were a
significant confound. Her correct responses could have been
based on static textural cues created in the background by
the absence of a large percentage of dots either before
(accretion case) or after (deletion case) the transient signal
was applied. An ancillary condition with 70% signal, appears
to confirm this as L.M.'s performance increased to 80%
correct. L.M. could not use a non-motion transient signal for
letter identification, but instead based her responses on
relative density cues.

Static texture cues
While the warm-up condition with the black block letters
demonstrated that L.M. could easily perform the letter
identification task, the goal of these static non-transient
control conditions was to determine whether L.M.'s trouble
in the previous conditions are the result of a generic problem
in isolating any type of signal in a noisy background.

117

would improve with a larger difference between the two
textures. In this condition, the letter-shaped region was filled
with small triangles (10' per side). The element density was
reduced by two-thirds to preserve the overall pixel density
within the letter-shaped region thus avoiding a potential
large-scale luminance cue. The background comprised the
same oriented line segments of the previous condition
(Fig. 5E).
The difference between these two textures is best
understood in terms of Julesz's texton theory of texture
perception (Julesz, 1986). The texture of a visual scene is
analysed by the observer's array of textons. One type of
texton might be simple non-overlapping line segments having
a length, orientation and width. Another type of texton is a
terminator (ends-of-lines). Yet another type of texton is a
blob. In the stimulus used here, the letter-shaped region
differs from the background in at least two textonal properties,
the number of terminators, and blobs. First, the background
line segments each have two terminators (ends) while the
triangles have none. Secondly, the line segments making the
triangles should also form blobs. Furthermore, the length and
width dimensions of the lines and blobs differ considerably.
Therefore, discrimination between the two textures could be
based on a couple of different features, whereas in condition
1, it would be based on orientation alone. Would it make
any difference in L.M.'s performance? Here, the proportion
of signal was set 100% for every trial and the presentation
duration was set to 1 s. Again, L.M. completed 50 trials.

Condition 1

Results

Letter-shaped areas were differentiated from the background
region by the orientation of line segments. The stimulus
background was filled with 1000 small lines, 2'X12', all in
either a vertical or horizontal orientation. Within the lettershaped region, the orientation of a proportion of lines was
changed by 90° (Fig. 5D). Like previous conditions in this
experiment, a larger proportion of the signal in the lettershaped region made the letter more easily identifiable. In this
condition, the proportion of signal was set at 100% and the
presentation duration doubled to 1 s. L.M. completed 50 trials.

L.M.'s letter identification performance increased to 64% in
this texture condition at 100% signal. L.M.'s improvement
in this condition, compared with the previous condition, is
most likely due to the additional information available through
the different type of texture. Of course, her performance was
still very different from normal.

Results
At 100% signal, L.M.'s letter identification performance of
26% correct was only slightly better than chance. Normal
performance at 500 ms presentation duration is characterized
by a threshold of -32% signal, and near perfect performance
at 100% signal. Therefore, in addition to her difficulties with
moving and transient stimuli, L.M. also has difficulty in
determining letter shape from texture differences from
orthogonal line orientations.

Condition 3
In a final static texture condition, a dot density was used to
generate a texture difference. First, the background region
was filled with 4000 small dots, the same dot density as used
in the relative motion experiments above. Within the lettershaped region, the local dot density was increased in some
proportion to the background region. A 100% increase would
mean double density (Fig. 5F). Of course, local luminance
differences mirrored the local density differences: a higher
density of light dots generates a higher local luminance. The
percent signal, local dot density as an increasing function of
the background density, was varied according to the method
of constant stimuli between 100% and 250%. Presentation
duration was 500 ms. L.M. completed 100 trials.

Condition 2

Results

Since L.M. had difficulty with letter identification from very
simple texture boundaries, we wondered if her performance

L.M.'s threshold in this density condition was 163% signal.
This means that the density of dots within the letter region

1118

M. Rizzo et al.

had to be nearly 1.5 times the background density before
L.M. could reliably discriminate the letter. The average
normal threshold for age matched controls is in the order of
120%. The difference between L.M.'s performance and
normal performance is very small in comparison to the size
of the deficit L.M. exhibits in the other cue types in this
experiment. The result from this condition shows that L.M.
can perform the letter identification task if given a static
luminance cue, even if it is made out of small dots. As long
as the dots are not moving, she has less difficulty with the task.

Binocular disparity cues
To test letter identification from static stereopsis, a randomdot stereogram was generated. Background dots had zero
binocular disparity. A proportion of dots within the lettershaped region were given 6' crossed binocular disparity. The
two frames of the stereogram were viewed alternately by the
two eyes through a LCD shutter system linked to the 67 Hz
vertical retrace of the monitor. The letters appeared to be
made of dots closer to the observer than the background
dots. In this condition, the proportion of dots within the letter
shapes with binocular disparity was set to 100%. Figure 5G
shows a static stereopair with 100% signal depicting an H
in the centre of the stimulus. To give L.M. ample time for
stereofusion with the LCD goggles, the presentation duration
was 2.1 s. L.M. completed 50 trials in this condition.

Results
L.M. performed at only 12% correct, even though the
stimulus was at 100% signal. Average normal performance
is characterized by a threshold of 23% signal. Normal
performance is typically perfect at 100% signal. Apparently,
L.M. had trouble with letter identification when static stereo
cues defined the shapes in the random-dot stimulus. This
result confirms that L.M.'s does have some impairment of
stereoscopic vision. While the disparity in these stimuli was
above L.M.'s stereoacuity of 140', she was unable to group
the dots into a structure she could identify as a letter shape.
To further underline her difficulty with moving stimuli, L.M.
was unable to complete testing when a moving version of
this stereoscopic stimulus was presented. In this stimulus,
the dots moved about randomly, but any moving to a position
inside the letter-shaped area was given a crossed binocular
disparity so that a scintillating letter appeared in front of a
scintillating background of dots. L.M. found this stimulus
too troubling to continue beyond a few trials.

Discussion of Experiment 3
2-D shape-from-motion
To summarize, L.M. had reduced ability to perceive 2-D
shape from motion. She could easily read block letters, but
threshold requirement was nearly 100% signal when the

letters stimuli were defined by moving dots among a static
random-dot background compared with I % signal for normals
(Experiment 3A, Condition 1). This difficulty with 2-D shape
perception in the presence of stationary noise resembles the
pattern described in L.M. by Baker et al. (1991) of poorer
perception of a directional motion signal in the presence of
stationary noise. This signal-from-noise problem may even
mask available stationary textural cues that could be used
for 2-D shape identification, similar to the troubles that
Merigan et al. (1993) reported in monkeys with area V2
lesions. Moreover, the masking effects are worse when the
motion signal occurs within the boundaries of the target
shape (Experiment 3, Conditions 1 and 2) than when in the
surround (Condition 3), suggesting a link to L.M.'s focus of
attention. Also, motion masks motion, whether the signal
and noise are orthogonal or opposite to each other. A
remaining challenge is to be able to explain these findings
in terms of the known centre-surround properties of receptive
fields of neurons in primates visual association cortex such
as simian area MT (V5) and their additive and inhibitory
interactions. We do not yet have the solution.

Perception of 2-D shape from non-motion cues
L.M. had difficulties perceiving 2-D shape from motion with
cue types other than motion including transient cues, textures,
and dynamic and static binocular disparity. The only condition
in which L.M.'s performance approached normal was in the
case of the texture density cue. Therefore, it is clear that
L.M. can perform the task correctly, and even well in the
case of a static luminance cue for 2-D shape. However, with
most other spatial cues, L.M. showed severe deficits in
recovering 2-D shape from these non-motion cues in the
presence of noise (Merigan et al., 1993).

Experiment 4: visuoperceptive tasks
The results in Experiment 3 show that L.M. has trouble
perceiving 2-D shape from non-motion cues. The question
arises whether this form perceptual trouble in L.M. is
restricted to 'early vision' tapped in Experiment 3, or whether
higher order visual cognitive functions are also affected. To
investigate this issue it would be necessary to present stimuli
that tap higher order visual processes. As well, these stimuli
must contain sufficiently high levels of signal to override the
signal-from-noise deficit measured in previous experiments.
This is analogous to using oversized letters to show that a
person with low visual acuity still has preserved lexical
functions. Thus, in Experiment 4 we used five different
standardized neuropsychological procedures and their
variations. All stimuli consisted of high contrast static blackand-white figures which are designed for viewing in a
relatively wide range of ambient room lighting conditions.
These tasks provided an index of L.M.'s 'visuoconstructive'
abilities, referring to a combination of processes which allow

Motion and shape perception

1119

us to perceive all the features of a display and to judge the
spatial relationships among those features.

Rey-Osterreith complex figure test (CFT)
This test requires subjects to copy a complex geometric
figure. Good normative information is available (Spreen and
Strauss, 1991). An extensive scoring system is provided, and
response time is recorded as a second measure of subject
performance. The CFT copy provides a reliable index of
visuoconstructional ability, independent of memory function.
We also administered a version of the CFT in which drawing
is not required and performance depends on recognition
rather than reproduction. This version uses 10 pairs of Rey
figures. One member of the pair is always identical to the
target item (the standard Rey figure) while the other differs
with respect to a single detail.

Benton Visual Retention Test—Revised (BVRT)
The BVRT requires a subject to reproduce, by drawing, 10
line drawings depicting geometrical designs (Benton and Van
Allen, 1985). Typically, the subject is shown each plate for
10 s, the plates are withdrawn, and the subject is asked to
reproduce the figure by drawing on a white paper, after a
15-s delay. The first two designs consist of a major geometrical
figure and the remainder contain two major figures and a
peripheral figure. The BRVT is sensitive to impairments of
visual memory, perception and visuoconstructive impairments. It is well standardized, and has three different, but
equivalent forms (A, B and C). It is easy to administer, and
takes -10 min in most subjects. We administered form C.
We also administered versions F and G of the task in
which drawing is not required and performance depends on
recognition rather than reproduction (Benton, 1990). The
subject's task is to select which of four plates is identical to
a previous drawing which was viewed for only 10 s. Versions
F and G each contain 15 different items.

Hooper Visual Organization Test (HVOT)
In this task (Hooper, 1958; Western Psychological Services,
1983), the subject is asked to identify 30 different items (e.g.
scissors, shoe, broom, fish) from cut up, rearranged linedrawings of those objects. This test is memory dependent
since it depends on a history of exposure to the items
represented in the test.

Mooney's Closure Faces Test
This test (Mooney and Ferguson, 1951; Lansdell, 1970;
Newcombe, 1974) is a 'closure test' which provides
information similar to the VOT. The subject is presented
with 44 different black and white incomplete cartoons of
faces. The subject is asked to judge the age and sex of the faces

Fig. 6 The cookie theft picture. L.M. reported all the major items
in this high contrast black and white line drawing, just as normal
individuals do. She reported 'a scene' of 'a woman washing
dishes', and the 'water running over'. She also saw 'a girl', 'a
boy reaching and giving something to the girl' and noted the 'boy
is in trouble because he is not allowed in the cookies' and 'will
fall because the chair is falling over.' (Reproduced courtesy of
Harold Goodglass.)

depicted. Performance on this task is memory dependent, but
the recognition of previously familiar faces or famous faces
is not required.

Cookie theft picture
This black on white line drawing of a complex picture from
the Boston Diagnostic Aphasia Examination (Goodglass and
Kaplan, 1983) is designed with a balance of information
among the four quadrants and tells a story (see Fig. 6). The
subject is asked to view the picture and report what they see.
Scoring is qualitative and depends on a report of the major
items in the display.

Results
Complex figure test
Despite liberal scoring, severe impairments were noted on
two occasions {see Fig. 7). This included scores of four out
of 36 at >10' and 8.5 out of 36 at >15'. Importantly, L.M.
recognized that there were differences between the test figure
and her own reproduction of it. We did not undertake to
repeat the test for memory because of the poor performance
to direct copy. However, when L.M. was administered the
matching variant of this task, she performed much better.
She matched nine out of 10 items correctly, albeit slowly
(mean time per item: 71.2 s, SD 39.4). The performance of
two female subjects aged 52 and 53 years was 100% correct
and required 8.3 s (SD 2.5) and 7.5 s (SD 2.4) on average
per item.

1120

M. Rizzo et al.

A

slow, requiring 4.4 and 4.5 min, respectively, to perform
tasks that took the controls 2.1 and 2.3 min on average to
complete.

Hooper Visual Organization Test
L.M. achieved a score of 18 out of 30 with a corrected Tscore of 64. While this score indicates a moderate impairment
according to the normative data of Hooper (1983), it is
proportionally much better than the scores L.M. achieved on
the CFT-copy, and BVRT form C.

Mooneys Closure Faces Test
For this task, L.M. achieved 26 correct out of 44. This
performance is below expectations for L.M.'s age. As with
the HVOT, L.M.'s performance for this task was
proportionally much better than the scores she achieved on
the CFT-copy, and BVRT form C, and resembled more her
performance on the CFT and BVRT matching tasks. In fact,
L.M. was able to discriminate difficult items that many
normal subjects, (including the readers) have trouble with
(see Fig. 8).

Cookie theft picture

Fig. 7 (A) The Rey-Osterreith Complex figure is depicted. (B)
L.M.'s ability to copy this complex geometric figure was severely
defective. However, when L.M. engaged in visuoperceptive tasks
that required the recall rather than the reproduction of complex
visual patterns, her performance was much improved. L.M.'s
copying difficulties could have been related to her previous
complaints that she could not track the trajectory of her own right
index finger if it moved 'too fast' (see text).

Benton Visual Retention Test
Like the CFT, this test was given with no time restriction.
When performance depended upon reproduction by drawing,
L.M.'s score was two correct and 20 errors (normal for
L.M.'s age is six correct and six errors). This performance
was defective (Sivan, 1992). However, when performance
depended on recognition rather than reproduction, L.M.
again performed much better. The cut-off score for normal
performance on VRT versions F and G is 11 out of 15 correct
on each version (Benton, 1990). L.M. scored 12 out of 15
correct on version F, and 11 out of 15 correct on version G.
In comparison, the scores of two female control subjects (the
same tested in 5A) were 13 and 14 out of 15 on F and both
14 out of 15 correct on G. However, here again, L.M. was

L.M. spontaneously reported all the salient features in the
display, just as individuals with unimpaired vision and
language do. Specifically, she reported 'a scene' of 'a woman
washing dishes', and the 'water running over', exactly as
depicted on the right side of the display. On the left side of
the display she saw 'a girl', and 'a boy reaching and giving
something to the girl.' Furthermore, the 'boy is in trouble
because he is not allowed in the cookies perhaps' and 'the
boy will fall because the chair is falling over.' In short, L.M.
had a normal performance on this task.

Discussion of Experiment 4
L.M. was unable to reproduce line drawings of geometrical
figures printed at 100% contrast on the BVRT and CFT (see
Fig. 7). Such visuoconstructive difficulties are commonly
observed in patients with lesions of the posterior right
hemisphere. Patient A.F. reported by Vaina et al. (1990) is
an example (see fig. 2, p. 356, in which A.F. attempts to
draw a house to direct copy). These defects are especially
marked in individuals with 'visual disorientation' or
'simultanagnosia' due to bilateral lesions of the dorsolateral
visual system as reported by Balint (1909).
However, we do not think L.M. has 'visuoconstructive'
deficits like those described in right hemisphere cases or in
Balint syndrome. Her impaired performances on the CFT
and BVRT were evidently task dependent, and not revealing
of any inability to perceive or process certain dimensions or
features of the high contrast black-and-white static stimulus
patterns. Thus, L.M. produced an accurate performance on

Motion and shape perception

1121

Fig. 8 L.M. can perform sophisticated judgments on complex static stimuli, as long as the stimuli are
presented in high contrast, low-noise displays. She correctly judged the age and sex in different black
and white incomplete cartoons of faces depicted on the Mooney's Closure Faces test. For example, she
correctly identified the figures depicted in plates 29 (left) and 44 (right), which some normal subjects
mav even lind difficult.

these tusks when the mode of response was changed from
reproduction by drawing to forced choice pattern recognition.
Indeed, we lind ample evidence that L.M. can perform
sophisticated judgments on complex static stimuli, as long
as the stimuli are presented in high contrast, low-noise
displays, for which her performance on the HVOT. Mooney's
Closure Faces Test and cookie theft picture, was evidence.
As well. L.M. can still read. i.e. she does not have pure
(acquired) alexia. She can recognize objects and faces and
thus does not have agnosia, defined in the sense of Milner
and Teuber (1968) as a recognition defect in which normal
or near normal percepts arc stripped of their meanings.
Interestingly the lesions described in Balint's case overlap
those described in L.M. (compare the anatomical drawing of
Balint (1909) with Zihl et id. (1991). tig. 11. and the cases
do show some behavioural similarities. Balint's man had
optic ataxia. a defect of hand control under visual guidance,
and ocular apraxia. an inability to move the eyes to objects
of interest. L.M. also had trouble with eye-hand coordination
and was unable to track her own linger movements under
visual guidance (Zihl et id.. 1983). We believe L.M.'s trouble
copying patterns such as on the CFT could have been due.
in part, to such a deficit. Furthermore, we would not be
surprised if Balint's patient also had both a disturbance of
smooth pursuit eye movements and of movement perception
as L.M. does. Balint did not comment on the former defect,
and if he ever considered a movement perception defect in
his patient, he probably lacked a means to test it. Descriptions
of the phi phenomenon occurred only after Balint's report.
However, judging by her preserved report of the cookie
theft picture, L.M. does not have the spatial disorder of
attention known as visual disorientation (Holmes. 1918) or
simultanagnosia (Wolpert, 1924). Simultanagnosia is the

perceptual underpinning of Balint syndrome and probably
represents a variety of combined perceptual defects including
VI type scotomata, defective visuoconstructive ability and
defective visual attention (Rizzo. 1993). The effect is a
marked reduction of the useful field of view. In fact, Balint
claimed that his patient could only see one object at a time,
no matter what size. Actually, simultanagnosia is not really
an agnosia in the sense of Milner and Teuber (1968) and is
better covered by the term 'apperceptive agnosia", used by
Lissauer (1890). Unlike L.M.. simultanagnosics fail to report
the major objects and relationships among objects in displays
such as the cookie theft. If L.M. has trouble attending to
simultaneous stimuli, it is because they are moving (McLeod
el id.. 1989). In contrast, the deficit described by Balint in
his patient and by subsequent authors such as Hecaen and
Ajuriaguerra (1954), indicates that the problem perceiving
objects in simultanagnosia is more generalized and occurs
whether visual targets are moving or are stationary.

Experiment 5: perception of the bistable
movement with the Ternus display
The Ternus display (Ternus. 1938) is a two-frame animation
sequence depicting movement in a set of three evenly spaced
identical bars. Between frames, the set of bars is shifted
laterally by one cycle, such that two of the bars lie on the
positions occupied by their neighbour in the previous frame
(see Fig. 9A). A variable length LSI is interposed between
the presentation of the two frames. The movement in this
display has two possible perceptual interpretations, depending
on the ISI duration (see Fig. 9B). With long ISIs (100 ms),
the set of bars appears to move back and forth en masse:

1122

M. Rizzo et al.

Ternus Display
Frame 1

ISI

Frame 2

III

III

Element
Movement

Group
Movement

L.M.
- - Normals (1 SD)

movement has the longer perceived movement excursion in
this display. The short-range system generates the condition
for element movement by signalling the 'stationarity' or no
movement of the middle two elements (Petersik and Pantle,
1979; Braddick, 1980). As Braddick (1980) points out, 'Any
system of directionally selective detectors must have a
characteristic null response when no movement occurs.' This
null response is responsible for the perception of no movement
by the overlapping (middle) elements in the Ternus display.
The perceived movement of the outermost element is then
likely the product of the long-range system.
The perceptually bistable display described above provides
a well-defined paradigm with which to study L.M.'s motion
perception. We wondered if L.M. would perceive these two
different types of movement with this display, and whether
the parameters necessary for seeing each type of movement
would be similar to those for normal observers. Not only
would this experiment tell us something about L.M.'s
perceptual mechanisms, but more generally, L.M.'s performance might provide insight into the different mechanisms
underlying the two perceptual interpretations of this display.

Method
15

30

45

60

ISI Duration (ms)
Fig. 9 (A) The three frames of the Ternus display are depicted.
Frames 1 and 3 have one element transposed to the opposite end.
Frame 2 is a variable ISI during which the screen is blank. (B)
With brief ISls, typically one element appears to move from end
to end, termed element movement. With longer ISIs, typically all
three elements appear to move together back and forth, termed
group movement. (C) The proportion of element movement
reports for L.M. and a group of normal observers as a function of
ISI duration. Error bars denote I SD. L.M. transitioned from
element movement to group movement reports at a shorter ISI
than normals.

this is called group movement. However, with brief ISIs
(10 ms), one bar appears to move back and forth between
the two ends, while the middle two bars appear stationary;
this is called element movement. The shift from reports of
element movement to reports of group movement takes place
typically around 40 ms ISI.
These two mutually exclusive percepts are believed to
illustrate the dichotomy between the short-range and longrange processes in apparent motion (AM) perception
(Braddick, 1980; Petersik, 1989). Pantle and Picciano (1976)
suggest that group movement is subserved by the long-range
AM process. Both long-range motion and group movement
in the Ternus display are perceived under similar conditions.
These conditions are: (i) long ISIs; (ii) dichoptic viewing of
AM frames; and (iii) contrast reversals between AM frames.
Conversely, element movement is thought to be subserved
by the short-range AM system, even though element

The Ternus display used in this study had a three-frame
animation sequence depicting three vertical black bars against
a white background on the first and third frames (see Fig. 9A).
The middle frame was blank, and was presented during the
ISI. Each bar was 2.5° tall and 30' wide. The bars were
interspaced by 1 ° (two-bar widths). To generate the movement
in display, one bar was displaced from one lateral extreme
to the other in alternating frames. The interval between
frames (ISI) varied as did the frame duration. The ISI was
characterized by a solid white background.
Since the display was shown on a 67-Hz monochrome
monitor, all stimulus parameters were a function of the
15 ms frame rate. The ISI varied between 0 and 60 ms, in
15 ms intervals. Frame duration was 60, 90, 120 or 150 ms.
The selection of these parameters was based on some brief
pilot testing. L.M. completed 10 trials for each ISI-frame
duration combination for 240 trials in all.
In each trial, the subject viewed 12 cycles of the two
frames and the ISIs. The subject's task was to report if the
entire set of three figures appeared to move together backand-forth (group movement), or whether the outside figure
jumped from end-to-end. Trials were initiated by the
experimenter after signalling for the subject's attention to the
screen. The subject's reports were both verbal and gestural.

Results
Since L.M. showed no effect for frame duration the data for
ISI across all frame durations is shown in Fig. 9C. For
comparison, the performance of six normal observers (mean
age 50 years) is shown with error bars denoting 1 SD. The
performance of our normal observers is similar to that

Motion and shape perception
previously reported by Pantle and Picciano (1976) and
Petersik and Pantle (1979) with a 200 ms frame duration.
L.M.'s performance was broadly similar to that of normal
controls. She reported a greater proportion of element
movement with shorter ISI and more group movement with
longer ISI. Also, like normal controls, L.M. continued to
report group movement with longer ISIs. However, there was
an important difference in L.M.'s performance. In particular,
the transition in reports of element to group movement
occurred at a very short ISI. In fact, L.M.'s reports were
predominantly of group movement at ISI of only 15 ms. The
transition in normals occurs between 30 ms ISI (the current
study) and 40 ms ISI (Pantle and Picciano, 1976).

Discussion of Experiment 5
L.M.'s phenomenal reports of the stimulus movement are
interesting. She reported no perception of element movement
in these displays with brief ISIs. Instead, L.M. described that
the outer element appeared to 'blink' on and off, on opposite
sides of the middle overlapping elements. She perceived no
movement of this element between the two ends. This
coincides with her reported loss of long-range movement
perception (Hess et al., 1989). Her phenomenal descriptions
in conditions of group movement were similar. With the
longest ISIs, L.M. reported that the group of elements
appeared mostly to 'flash' at the two locations without a
percept of movement between them. With shorter ISIs, L.M.
tentatively reported some movement of the group of elements
between the two locations, similar to that which normal
observers would report. This suggests that L.M. may have
some ability to see some long-range movement or that these
limited reports of movement were mediated by a short-range
mechanism. Regardless of what limited motion perception
remains, it is apparent from L.M.'s ease in performing the
task that she could infer movement from the 'flashing' or
'blinking' of the stimulus elements even if she did not
perceive actual movement.
One alternative explanation for L.M.'s performance is that
L.M. simply detected flashing in the middle overlapping
elements and gave an appropriate response. This simple
strategy is possible even though the subject did not know
the purpose of the test and the 'normal' pattern of results;
i.e when the middle elements appear to flash (a long ISI),
the more appropriate response is 'group movement'.
However, when the middle elements do not appear to flash
(a short ISI), the more appropriate response is element
movement. By detecting flashing, a subject could produce a
ISI dependent change in reports of element and group
movement.
However, this pattern of responding is unlikely to generate
the results shown for L.M. in Fig. 6. She demonstrates normal
flicker fusion frequency of -30 Hz. This means stimuli at
15 ms ISI should appear almost continuous (i.e. not flash),
especially with these comparatively long frame rates of
60-150 ms. Therefore, L.M. was not simply detecting and

1123

reporting on stimulus flashing. If she had been, her transition
from reports of element movement to reports of group
movement would have been at a longer ISI, and appeared
more similar to normal observers. On the contrary, another
alternative explanation is that L.M. has unusually brief visual
pattern persistence which might not have been detected upon
testing her flicker fusion frequency. Breitmeyer and Ritter
(1986) found that increased pattern persistence extends reports
of element motion to longer ISI durations, supposedly by
contributing to the temporal integration of pattern and to the
perceptual stationarity of the two inside elements. As L.M.
transitioned to reports of group movement at a very brief
ISI, it remains a possibility that she exhibits very short
pattern persistence, the reverse of Breitmeyer and Ritter's
finding. Inasmuch as visual persistence contributes to
temporal integration of pattern (Di Lollo and Woods, 1981),
it is important to consider that such a deficit might interfere
with the integration of local dots movements in to a percept
of global flow in the RDC used to assess motion perception.
This possibility was not pursued in this set of experiments.
We believe the pattern in which L.M. reported group or
element movement supports the notion of Hess et al. (1989)
that motion perception in L.M. depends primarily on residual
short-range processes. Long-range motion processing appears
to be limited, and L.M. does not even report seeing phimovement (Zihl et al., 1983). With the Ternus display, it
appears that L.M.'s short-range system operates at 0 and
15 ms ISI to signal the stationarity of the overlapping
elements. By 30 ms ISI, L.M. reports only group movement,
whereas normal subjects still report element movement on
half the trials. This difference may result from the limited
nature of L.M.'s residual short-range motion processing,
which fails to detect whether the overlapping elements are
stationary at longer ISI. In such cases, L.M. reports group
movement.
The results of this experiment suggest that actual perception
of both element and group movement in the Ternus display
is mediated by the long-range motion system. For the most
part, L.M. saw neither. However, the difference between the
two percepts might be based on the activity of the shortrange system as previously proposed. When L.M.'s limited
short-range system was presented with an appropriate spatiotemporal stimulus, it actively signalled a null response. A
longer ISI made it an ineffective stimulus, L.M.'s short-range
system did not signal 'stationarity', and therefore group
movement was reported.

Conclusion
Summary
L.M. has a severe, persistent defect of motion perception.
While L.M. can still perceive a directional motion signal,
the results in Experiment 1 indicate this ability is highly
sensitive to noise, as first indicated by Baker et al. (1991).
Similarly, L.M. can perceive 2-D shape and 3-D SFM, but

1124

M. Rizzo et al.

Table 2 Magnitude of perceptual deficits measured in subject L.M.
Perceptual function

Deficit magnitude

Global coherent motion direction discrimination

Severe—at moderate noise levels

Perception of 3-D structure
from motion
from dynamic stereopsis

Severe—even at 100% signal
Severe—at moderate noise levels

Perception of 2-D shape
from motion cues
from transient ("on' and 'off ) cues
from static texture cues
from static stereo cues
from dynamic stereo cues
'Visuoconstructive' abilities
for 2-D stimulus presented at maximum contrast
Apparent (phi) movement in Temus display

even moderate levels of noise causes a breakdown in these
abilities (Experiments 2 and 3). L.M.'s perceptual reports in
viewing the Ternus display (Experiment 5) were similar to
normal, however, it appears that L.M. does not perceive longrange motion. Moreover, L.M.'s perceptual deficits extend
outside the processing of motion (Experiments 3 and 4).
L.M. also has marked difficulty perceiving 2-D shapes defined
by non-motion signals including 'transient' on-off cues,
dynamic and static stereo cues, and different static textures
(see Table 2).

Akinetopsia and area MT (V5)
The pattern of perceptual deficits in akinetopsic subject L.M.
does not resemble the pattern of M-pathway, P-pathway, or
area V4-type dysfunction observed in monkeys (Schiller and
Logothetis, 1990; Schiller, 1993). Baker et al. (1991) likened
L.M.'s poor CS for the direction of drifting sine wave grating
and decreased direction discrimination of RDC to the pattern
in animals lacking direction sensitive neurons in visual cortex
due to rearing under stroboscopic light. More recently, Plant
and Nakayama (1993) compared L.M.'s underestimation of
target speed to the errors reported by Newsome et al. (1985)
in monkeys with area MT (V5) lesions. Perhaps so, but the
perceptual profiles measured in this study indicate that the
pattern of perceptual deficits in L.M. extends beyond motion
processing and area MT (V5) dysfunction.
Another distinction between L.M.'s deficits and simian
area MT (V5) dysfunction concerns the longevity of the
deficits: L.M.'s motion trouble has persisted for over a
decade, whereas area MT (V5) dysfunction in monkey
recovers rapidly. A likely reason is that area MT (V5)
dysfunction is created by making millimetre-sized chemical
lesions in simian cortex using ibotenic acid (Newsome et al.,
1985), which largely spares the white matter. On the other

Severe—at low noise levels
Severe—at low noise levels
Severe—for certain textons
Minimal—for dot density cues
Severe—even at 100% signal
Severe—even at 100% signal

None to moderate—as long as
drawing not required
Mild to moderate at short ISI and severe at long IS1—suggesting
absent 'long range' mechanisms

hand, L.M.'s deficits are caused by cerebrovascular lesions
which are larger and more destructive. They are bilateral,
and widely surround the location of a putative human MT
area (V5) homologue as depicted in a PET study by Watson
et al. (1993). Due to their extent, the dorsolateral lesions in
L.M. may also include homologues of simian middle superior
temporal area, floor of the superior temporal sulcus and
lateral intraparietal area (Ungerleider and Desimone, 1986;
Boussaoud et al., 1990; Tanaka et al, 1993) which also
process motion, and affect the white matter connections
underlying these regions. Interestingly, Pasternak and
Merigan (1994) reported that simian MT area lesions produce
deficits that are graded in relation to lesion extent.

Residual movement perception
Despite extensive and critically located lesions in the
dorsolateral visual pathways, L.M. is not fully motion blind.
This observation is compatible with the idea that the
representation of motion in the primate visual system is
redundant (Ferrera et al, 1994). A human homologue of the
monkey's area MT (V5) complex may play an important role
in motion processing, but it does not operate in isolation.
Rather, it should function in concert with other structures in a
network for human motion perception, including the superior
colliculus, frontal eye fields, parieto-occipital cortex,
dorsolateral pontine nucleus, and even the cerebellum
(Nawrot and Rizzo, 1995). Residual motion processing in
L.M. might also depend on portions of the cuneus, precuneus
(Shipp et al, 1994), and ventromesially located structures
such as the fusiform and lingual gyri, which were spared in
L.M. The latter regions are thought to contain a human area
V4 homologue (Zeki, 1990), and the evidence in monkeys
indicates V4 possesses a rudimentary motion processing

Motion and shape perception
capacity (Schiller, 1993). Based on Experiment 5, such a
residual capacity includes a combination of short-range and
defective long-range apparent motion mechanisms. Of course,
L.M. can always infer motion by making use of any available
information on object displacements, but this is not true
motion perception.

Two visual systems
The hypothesis of 'two visual systems' in primates suggests
a 'what' pathway for perceiving and recognizing objects,
and a 'where' pathway for locating those objects in space
(Ungerleider and Mishkin, 1982; Damasio, 1985). In the
human brain, such pathways are thought to segregate along
a ventromesial to dorsolateral axis (Rizzo, 1994). The
dorsolateral pathway begins above the calcarine fissure in
occipital areas 18 and 19 and extends into adjacent parietal
and temporo-parieto-occipital regions located on the lateral
surface of the brain. These regions include areas 5, 7, 39, 37
and superior 19, portions of which were damaged in
akinetopsic subject L.M. A ventromesial visual pathway
begins in the human occipital lobe beneath the calcarine
fissure, connects with adjacent temporal regions, and may
include human homologues of the monkeys' areas V4 and
IT. L.M. has no evidence of damage to these latter structures,
yet she can be construed to have a 'what' deficit based on
her trouble perceiving 2-D shapes and 3-D objects defined
by a variety of different, motion and non-motion cues. These
findings indicate that the ability to segregate figure-fromground is not a unique property of the ventromesial visual
association areas. Rather, the dorsolateral association areas,
which should contain a human homologue of siman area MT
(V5), are also capable of making a significant contribution
to this process, especially when dynamic cues are involved.
The notion that the processing of information on textures,
patterns, shapes and objects depends exclusively on a
ventromesial or 'what' pathway in the human brain is an
oversimplification because parallel pathways have cross-talk
and converge (Martin, 1992; Merigan and Maunsell, 1993).

The role of visual association cortex in
extracting signal-from-noise
As mentioned, the motion direction discrimination in L.M.
is critically sensitive to noise, both moving and stationary, a
result in clear agreement with Baker et al. (1991).
Interestingly, so is the perception of 2-D shape and 3-D
SFM. The last result resembles the finding in motion deficient
patient A.F. reported by Vaina et al. (1990) who had trouble
perceiving 3-D SFM at lower motion signal to noise ratios.
Together these findings highlight the role of visual association
cortex, including a possible homologue of the monkey's area
MT (V5) complex and connections, in 'sharpening' our
picture of the world and thereby increasing the accuracy
of our perceptual judgments and decisions (Salzman and
Newsome, 1994). This function is accomplished by reducing

1125

the effects of stationary and motion noise which is ever
present in the visual scene, inherent in the operations of
visual neurons, and also caused by the ceaseless jitter of any
normal observer's eyes. The latter role would depend on
efference copy, the monitoring of corollary discharge from
eye movements, perhaps in connection with the cerebellum.
The cerebellum plays a role in motion perception (Nawrot
and Rizzo, 1995) and motor control including eye movements,
contains cells in its vermis which are sensitive to the direction
of movement of visual targets resembling those in area MT
(V5), and forms part of a CNS network that includes MT
area (V5). Interestingly, the most recent anatomical studies
of L.M. both show a significant midline cerebellar lesion
(Zihl et al., 1991; Shipp et al., 1994).

Acknowledgements
This work was supported by NIH PO NS19632, McDonnellPew Program in Cognitive Neuroscience, and MRC Canada.
References
Allman JM. Evolution of the visual system in early primates. In:
Sprague J, Epstein AN, editors. Progress in psychology, physiology,
and psychiatry, New York: Academic Press, 1977: 1-53.
Baker CL Jr, Hess RF, Zihl J. Residual motion perception in a
'motion-blind' patient, assessed with limited-lifetime random dot
stimuli. J Neurosci 1991; 11: 454-61.
Balint R. Seelenlahmung des 'Schauens', optische Ataxie, raumliche
Storung der Aufmerksamkeit. Mschr Psychiat Neurol 1909; 25:
51-181.
Beckers G, Homberg V. Cerebral visual motion blindness: transitory
akinetopsia induced by transcranial magnetic stimulation of human
area V5. Proc R Soc Lond B Biol Sci 1992; 249: 173-8.
Benton A. Visuoperceptual, visuospatial, and visuoconstructive
disorders. In: Heilman KM, Valenstein E, editors. Clinical
neuropsychology. 2nd ed. New York: Oxford University Press,
1985: 151-85.
Benton AL. Der Benton Test. 6th ed. Bern: 1990.
Boussaoud D, Ungerleider LG, Desimone R. Pathways for motion
analysis: cortical connections of the medial superior temporal and
fundus of the superior temporal visual areas in the macaque. J
Comp Neurol 1990; 296: 462-95.
Braddick OJ. Low-level and high-level processes in apparent motion.
Philos Trans R Soc Lond B Biol Sci 1980; 290: 137-51.
Braddick O. Segmentation versus integration in visual motion
processing. Trends Neurosci 1993; 16: 263-8.
Braunstein ML. The perception of depth through motion. Psychol
Bull 1962; 59: 422-33.
Bravo MJ, Watamaniuk SNJ. Speed segregation and transparency
in random dot displays [abstract]. Invest Ophthalmol Vis Sci 1992:
33 (4 Suppl): 1050.
Breitmeyer BG, Ritter A. The role of visual pattern persistence in
bistable stroboscopic motion. Vision Res 1986; 26: 1801-6.

1126

M. Rizzo et al.

Brodmann K. Vergleichende Lokalisationslehre der Grosshirnrinde
in ihren Prinzipien dargestellt auf Grund des Zellenbaues. Leipzig:
J. A. Barth, 1909.
Clarke S, Miklossy J. Occipital cortex in man: organization of
callosal connections, related myelo- and cytoarchitecture, and
putative boundaries of functional visual areas. J Comp Neurol 1990;
298: 188-214.
Corbetta M, Miezin FM, Dobmeyer S, Shulman GL, Petersen SE.
Selective and divided attention during visual discriminations of
shape, color, and speed: functional anatomy by positron emission
tomography. J Neurosci 1991; 11: 2383-402.
Damasio AR. Disorders of complex visual processing: agnosias,
achromatopsia, Balint's syndrome, and related difficulties of
orientation and construction. In: Mesulam MM, editor. Principles
of behavioral neurology. Philadelphia: F. A. Davis, 1985: 259-88.
Di Lollo V, Woods E. Duration of visible persistence in relation to
range of spatial frequencies. J Exp Psychol Hum Percept Perfom
1981; 7: 754-69.
Dubner R, Zeki SM. Response properties and receptive fields of
cells in an anatomically defined region of the superior temporal
sulcus in the monkey. Brain Res 1971; 35: 528-32.
Ferrera VP, Rudolph KK, Maunsell JHR. Responses of neurons in
the parietal and temporal pathways during a motion task. J Neurosci
1994; 14: 6171-86.
Flechsig P. Developmental (myelogenetic) localisation of the
cerebral cortex in the human subject. Lancet 1901; 2: 1027-9.
Frost BJ, Nakayama K. Single visual neurons code opposing motion
independent of direction. Science 1983; 220: 744-5.
Gibson JJ. The perception of the visual world. Boston: Houghton
Mifflin, 1950.
Goodglass H, Kaplan E. The assessment of aphasia and related
disorders. 2nd ed. Philadelphia: Lea and Febiger, 1983.
Hecaen H, de Ajuriaguerra J. Balint's syndrome (psychic paralysis
of visual fixation) and its minor forms. Brain 1954; 77: 373-400.

Martin KAC. Parallel pathways converge. Curr Biol 1992; 2: 555-7.
Maunsell JHR, Van Essen DC. Functional properties of neurons in
middle temporal visual area of the macaque monkey. I. Selectivity
for stimulus direction, speed, and orientation. J Neurophysiol 1983a;

49: W21-M.
Maunsell JHR, Van Essen DC. Functional properties of neurons in
middle temporal visual area of the macaque monkey. II. Binocular
interactions and sensitivity to binocular disparity. J Neurophysiol
1983b; 49: 1148-67.
McLeod P, Heywood C, Driver J, Zihl J. Selective deficit of visual
search in moving displays after extrastriate damage. Nature 1989;
339: 466-7.
Merigan WH, Maunsell JHR. How parallel are the primate visual
pathways? [Review], Annu Rev Neurosci 1993; 16: 369^02.
Merigan WH, Nealey TA, Maunsell JHR. Visual effects of lesions
of cortical area V2 in macaques. J Neurosci 1993; 13: 3180-91.
Milner B, Teuber HL. Alteration of perception and memory in man:
reflections on methods. In: Weiskrantz L, editor. Analysis of
behavioral change. New York: Harper & Row, 1968: 268-375.
Mooney CM, Ferguson GA. A new closure test. Can J Psychol
1951; 5: 129-33.
Morrow MJ, Sharpe JA. Retinotopic and directional deficits of
smooth pursuit initiation after posterior cerebral hemispheric lesions.
Neurology 1993; 43: 595-603.
Nakayama K. Biological image motion processing: a review.
[Review]. Vision Res 1985; 25: 625-60.
Nawrot M, Blake R. A neural network model of kinetic depth. Vis
Neurosci 1991a; 6: 219-27.
Nawrot M, Blake R. The interplay between stereopsis and structure
from motion. Percept Psychophys 1991b; 49: 230-44.
Nawrot M, Rizzo M. Motion perception deficits from midline
cerebellar lesions in human. Vision Res 1995; 35: 723-31.

Hess RH, Baker CL Jr, Zihl J. The 'motion-blind' patient: lowlevel spatial and temporal filters. J Neurosci 1989; 9: 1628-40.

Nawrot M, Shannon E, Rizzo M. The relative efficacy of cues for
2-D shape perception. Vision Res. 1995. In press.

Holmes G. Disturbances of visual orientation. Br J Ophthalmol
1918; 2: 449-68; 506-16.

Newcombe F. Selective deficits after focal cerebral injury. In:
Dimond SJ, Beaumont JG, editors. Hemisphere function in the
human brain. New York: Halsted Press, 1974: 311-34.

Hooper HE. The Hooper Visual Organization Test: Manual. Beverly
Hills (CA): Western Psychological Services, 1983.
Julesz B. Texton gradients: the texton theory revisited. Biol Cybern
1986; 54: 245-51.
Kulikowski JJ, Tolhurst DJ. Psychophysical evidence for sustained
and transient detectors in human vision. J Physiol (Lond) 1973;
232: 149-62.
Lansdell H. Relation of extent of temporal removals to closure and
visuomotor factors. Percept Mot Skills 1970; 31: 491-8.
Lissauer H. Ein Fall von Seelenblindheit nebst einem Beitrag zur
Theorie derselben. Arch Psychiat NervKrankh 1890; 21: 221-70.
Long ME. The sense,of sight. National Geographic 1992; 182 (Pt
5): 3-41.

Newsome WT, Pare EB. A selective impairment of motion
perception following lesions of the middle temporal area (MT). J
Neurosci 1988; 8: 2201-11.
Newsome WT, Wurtz RH, Dursteler MR. Mikami A. Deficits in
visual motion processing following ibotenic acid lesions of the
middle temporal visual area of the macaque monkey. J Neurosci
1985; 5: 825^0.
Pantle A, Picciano L. A multistable movement display: evidence
for two separate motion systems in human vision. Science 1976;
193: 500-2.
Pasternak T, Merigan WH. Motion perception following lesions of
the superior temporal sulcus in the monkey. Cereb Cortex 1994: 4:
247-59.

Motion and shape perception

1127

Petersik JT. The two-process distinction in apparent motion.
[Review]. Psychol Bull 1989; 106: 107-27.

motion in the ventral part of the medial superior temporal area of
the macaque visual cortex. J Neurophysiol 1993; 69: 128-42.

Petersik JT, Pantle A. Factors controlling the competing sensations
produced by a bistable stroboscopic motion display. Vision Res
1979; 19: 143-54.

Ternus J. The problem of phenomenal identity. In: Ellis WD, editor.
A source book of Gestalt psychology. London: Routledge & Kegan
Paul, 1938: 149-60.

Plant GT, Nakayama K. The characteristics of residual motion
perception in the hemifield contralateral to lateral occipital lesions
in humans. Brain 1993; 116: 1337-53.
Plant GT, Laxer KD, Barbara NM, Schiffman JS, Nakayama K.
Impaired visual motion perception in the contralateral hemifield
following unilateral posterior cerebral lesions in humans. Brain
1993; 116: 1303-35.
Regan D, Beverley KI. Figure-ground segregation by motion contrast
and by luminance contrast. J Opt Soc Am [A] 1984; 1: 433^2.
Regan D, Hong XH. Visual acuity for optotypes made visible by
relative motion. Optom Vis Sci 1990; 67: 49-55.
Regan D, Giaschi D, Sharpe JA, Hong XH. Visual processing of
motion-defined form: selective failure in patients with
parietotemporal lesions. J Neurosci 1992; 12: 2198-210.
Rizzo M. 'Balint's syndrome' and associated visuospatial disorders.
[Review]. Bailliere's Clin Neurol 1993; 2: 415-37.

Thurston SE, Leigh RJ, Crawford T, Thompson A, Kennard C. Two
distinct deficits of visual tracking caused by unilateral lesions of
cerebral cortex in humans. Ann Neurol 1988; 23: 266-73.
Tittle JS, Braunstein ML. Recovery of 3-D shape from binocular
disparity and structure from motion. Percept Psychophys 1993; 54:
157-69.
Treue S, Husain M, Andersen RA. Human perception of structure
from motion. Vision Res 1991; 31: 59-75.
Ungerleider LG, Desimone R. Cortical connections of visual area
MT in the macaque. J Comp Neurol 1986; 248: 190-222.
Ungerleider LG, Mishkin M. Two cortical visual systems. In: Ingle
DJ, Goodale MA, Mansfield RJW, editors. Analysis of visual
behavior. Cambridge (MA): MIT Press, 1982: 549-86.
Vaina LM. Selective impairment of visual motion interpretation
following lesions of the right occipito-parietal area in humans. Biol
Cybern 1989; 61: 347-59.

Rizzo M. The role of striate coretx: evidence from human lesions
studies. Cereb Cortex 1994; 10: 505-40.

Vaina LM, Lemay M, Bienfang DC, Choi AY, Nakayama K. Intact
'biological motion' and 'structure from motion' perception in a
patient with impaired motion mechanisms. A case study. Vis
Neurosci 1990; 5: 353-69.

Rizzo M, Rotella D, Darling W. Trouble reaching after right
occipito-temporal damage. Neuropscyhologia 1992; 30: 711-22.

Wallach H, O'Connell DN. The kinetic depth effect. J Exp Psychol
1953; 45: 205-17.

Rouse MW, Tittle JS, Braunstein ML. Stereoscopic depth perception
by static stereo-deficient observers in dynamic displays with constant
and changing disparity. Optom Vis Sci 1989; 66: 355-62.

Watson JDG, Myers R, Frackowiak RSJ, Hajnal JV, Woods RP,
Mazziotta JC, et al. Area V5 of the human brain: evidence from a
combined study using positron emission tomography and magnetic
resonance imaging. Cereb Cortex 1993; 3: 79-94.

Salzman CD, Newsome WT. Neural mechanisms for forming a
perceptual decision. Science 1994; 264: 231-7.
Schiller PH. The effects of V4 and middle temporal (MT) area
lesions on visual performance in the rhesus monkey. Vis Neurosci
1993; 10: 717-46.

Williams D, Phillips G. Structure from motion in a stochastic
display [abstract]. J Opt Soc Am [A] 1986; 3 Suppl: P12-13.
Williams D, Phillips G. Rigid 3-D percept from stochastic 1-D
motion [abstract]. J Opt Soc Am [A] 1987; 4 Suppl: P35.
Williams D, Sekuler R. Coherent global motion percepts from
stochastic local motions. Vision Res 1984; 24: 55-62.

Schiller PH, Logothetis NK. The color-opponent and broad-band
channels of the primate visual system. [Review], Trends Neurosci
1990; 13: 392-8.

Wolpert T. Die Simultanagnosie. Stoning der Gesamtauffassung.
Zschr ges Neur Psychiat 1924; 93: 397-415.

Shipp S, de Jong BM, Zihl J, Frackowiak RSJ, Zeki S. The brain
activity related to residual motion vision in a patient with bilateral
lesions of V5. Brain 1994; 117: 1023-38.

Zeki SM. Functional organization of a visual area in the posterior
bank of the superior temporal sulcus of the rhesus monkey. J Physiol
(Lond) 1974; 236: 549-73.

Siegel RM, Andersen RA. Perception of three-dimensional structure
from motion in monkey and man. Nature 1988; 331: 259-61.

Zeki S. A century of cerebral achromatopsia. [Review]. Brain 1990;
113: 1721-77

Sivan AB. Benton Visual Retention Test, Manual 5th ed. San
Antonio: Psychological Corporation, 1992.

Zeki S. Cerebral akinetopsia (visual motion blindness): a review.
[Review]. Brain 1991; 114: 811-24.

Spreen O, Strauss E. A compendium of neuropsychological tests:
administration, norms, and commentary. New York: Oxford
University Press, 1991: 157-67.
Stoner GR, Albright TD. Image segmentation cues in motion
processing: implications for modularity in vision. J Cogn Neurosci
1993; 5: 129-49.
Tanaka T, Sugita Y, Moriya M, Saito H-A. Analysis of object

ZihL J, von Cramon D, Mai N. Selective disturbance of movement
vision after bilateral brain damage. Brain 1983; 106: 313—40.
Zihl J, von Cramon D, Mai N, Schmid C. Disturbance of movement
vision after bilateral posterior brain damage. Brain 1991; 114:
2235-52.
Received December 8, 1994. Revised April 3, 1995.
Accepted April 14, 1995