Brain (1997), 120, 1163–1171

Neglect in vision and visual imagery: a double
dissociation
H. Branch Coslett
Department of Neurology, Temple University School of
Medicine, Philadelphia, USA

Correspondence to: Dr Coslett, Department of
Neurology, Temple University School of Medicine,
3401 N. Broad Street, Philadelphia, PA 19140, USA

Summary
We report two patients with right hemisphere lesions who
demonstrate a double dissociation on visual imagery and
visual perceptual tasks. One (M.N.) performed normally on
a variety of standard tests for neglect as well as on measures
of visual attention known to be sensitive to the presence of
neglect, yet failed to report items from the left side of an
imagined scene. In contrast, the other (C.I.) performed

normally on tests of visual imagery but demonstrated
substantial neglect on visual perceptual and visual attentional
tasks. These data are not readily accommodated by accounts
which attribute neglect to a single processing deficit, but
suggest that the disorder is a heterogeneous syndrome
attributable to disruptions of different aspects of spatial
cognition.

Keywords: neglect; visual imagery; visual perception
Abbreviations: ANOVA 5 analysis of variance; ISI 5 inter-stimulus interval; RT 5 reaction time

Introduction
Bisiach and Luzzatti (1978) reported two patients who
neglected not only the left side of their environment but also
failed to report the left side of mental images. When asked
to describe a familiar scene, the Piazza del Duomo in Milan,
these patients reported a greater number of landmarks from
the right side of the imagined square; critically, the tendency
to report more landmarks from the right side was apparent
both when the patients imagined themselves standing with
their backs to the cathedral and when facing the cathedral
from the opposite end of the square.
Bisiach and colleagues (Bisiach and Luzzatti, 1978; Bisiach
et al., 1979) attributed this pattern of performance to a failure
to generate or maintain an adequate representation of the left
side of the image. They accounted for the correspondence
between performance on perceptual and mental imagery tasks
in terms of a common cognitive capacity, i.e. the ability to
construct and maintain a representation of the external
environment, that was critical to visual perception as well as
visual imagery.
An alternative account of these and similar data was
proposed by Meador et al. (1987). These investigators
reported a patient with neglect on perceptual tasks who also
failed to report information from the left side of imagined
© Oxford University Press 1997

scenes. The patient’s performance, however, was influenced
by head position; when asked to turn his head toward the
neglected side, the patient reported significantly more items
from the left side of the imagined scene than when instructed
to turn his head to the right. These investigators suggested
that the neglect of the left side of the image was attributable,
at least in part, to an impairment in the ability to direct
attention to this portion of the imagined scene.
More recently, dissociations in performance on tasks
assessing neglect in imagery and visual perception or
exploratory behaviour have been noted. For example,
Guariglia et al. (1993) reported a patient who neglected the
left side of mental images but exhibited no clear neglect
on perceptual tasks (see also Brain, 1941). Additionally,
Bartolomeo et al. (1994) tested 30 patients with right and 30
with left hemisphere lesions on a battery of tests assessing
visuospatial function as well as the ability to generate mental
images. They identified one patient who initially demonstrated
neglect on both visuospatial and imagery tasks, but 8 months
later exhibited neglect only on imagery tasks. Twelve patients
manifested neglect on visuospatial tasks but not imagery
tasks (see also Anderson, 1993).
We report two patients with right hemisphere lesions to

1164

H. B. Coslett

whom sensitive measures of visual attention and visual
imagery were administered. A clear double dissociation was
noted such that one patient, C.I., exhibited profound neglect
on visual perceptual and attentional tasks but performed
normally on a visual imagery task; the second patient, M.N.,
performed normally on tasks assessing visual attention, but
failed to report items from the left side of imagined arrays.
This double dissociation cannot be readily accommodated
by either of the accounts described above; these data are
most consistent with the hypothesis that ‘neglect’ is a
heterogeneous syndrome attributable to disruptions of
different visuo-spatial representations or systems mediating
intention to act in the neglected domain.

Subjects
Subject M.N.
M.N. is a 30-year-old right-handed woman with a highschool education who suffered an extensive right frontoparietal ischaemic infarct at the age of 28 years, probably
due to an embolus from a intra-cardiac tumour. Initially, she
exhibited a left hemiplegia and hemisensory deficit as well
as very mild and transient left-sided neglect. At the time
of the testing reported here (2 years after the infarct),
M.N. was living independently but unemployed. Neurological
examination revealed a moderately severe left hemiparesis
and mild left hemisensory impairment. Visual fields were
full and ocular movements, including smooth pursuit,
saccades and anti-saccades, were normal. There was no
extinction of visual, auditory or tactile stimuli. CAT scan
demonstrated an extensive right fronto-parietal infarction. As
demonstrated in Figs 1 and 2 the right superior parietal lobule
was infarcted, whereas the inferior portions of the angular
and supramarginal gyri appeared to be at least largely
preserved.
Neuropsychological examination revealed no evidence of
neglect on a line-bisection task in which lines ranged in
length from 4 to 32 cm; she performed equally well bisecting
lines in the right and left hemispaces. Similarly, she performed
normally on letter- and line-cancellation tasks presented at
the midline and to the right and left of the midline.
On neuropsychological examination, M.N. performed
normally on Teuber’s visual search test (Teuber et al., 1951)
and scored in the 30th percentile on Raven’s Standard
Progressive Matrices (score of 35). She performed perfectly
on a series of tests requiring that she made judgments about
the physical attributes of named objects or line drawings;
thus, for example, she reported that a ladder was larger than
a shovel and that a lime was green.
M.N. was impaired on a number of visuo-spatial tasks.
For example, she obtained a score of 11 out of 16 on the
Benton Visual Form Discrimination task and 19 out of 30
on the Judgement of Line Orientation. Although drawings
were poorly executed, there was no evidence of neglect of
the left side on these tasks.

Fig. 1 A supra-ventricular CT scan image (M.N.) demonstrating
infarction of the right mesial frontal lobe as well as portions of
the post-central gyrus and, perhaps, the supramarginal gyrus on
the right.

Subject C.I.
C.I. was a 70-year-old right-handed retired labourer, with a
history of focal motor seizures involving the right arm, who
had noted the sudden onset of left-sided weakness and
sensory loss 2 years prior to the investigations reported here.
He initially exhibited profound left-sided neglect but only
mild left-sided weakness.
At the time of the investigations reported here (18 months
after the right hemisphere infarct), C.I. exhibited mild
bilateral pyramidal clumsiness and spasticity which was
more prominent on the left. A left hemisensory deficit was
also noted. Visual fields were full and ocular movements
were normal except for hypometric saccades, noted both
when following an obect moving to the left and when
deviating his eyes to the left to verbal command.
C.I. exhibited evidence of neglect on a variety of standard
clinical measures. He erred to the right a mean of 18 mm
when bisecting 32 lines varying in length from 4 to 32 cm.
Similarly, he cancelled only the rightmost eight of a total of
40 ‘P’s randomly arrayed on a 20.3335.6 cm2 sheet of paper.
He omitted the left sides of drawings.
CAT scan revealed an extensive right hemisphere infarction
involving the temporal, lateral frontal and inferior parietal
lobes as well as small infarcts involving the lateral frontal
lobe and superior occipital lobe on the left (see Fig. 3).

Neglect and visual imagery

Fig. 2 A section CT scan image of M.N., higher than that in
Fig. 1, demonstrating infarction of the superior frontal lobe,
superior portions of the post-central gyrus, the superior parietal
lobule and precuneus on the right.

C.I. was tested on the paradigm employed by Coslett et al.
(1990) in an attempt to distinguish between directional
hypokinesia and hemispatial inattention. In this task, subjects
are asked to bisect a series of lines situated in either the right
or left (body and head) hemispaces. Subjects are not permitted
to gaze directly at the line to be bisected but receive visual
input to guide their performance from a video monitor; as
the monitor and lines to be bisected may be placed in
different hemispaces, the effects of directional hypokinesia
and directional hypokinesia may be dissociated. Consistent
with the hemispatial inattention account, C.I.’s performance
was not significantly influenced by the hemispace in which
he bisected the lines; his performance was significantly
influenced, however, by the location from which he received
visual information.
The testing reported here was approved by the Temple
University Institutional Review Board. Both subjects and
controls gave written, informed consent.

Experimental investigations
The data presented thus far indicate that C.I., but not M.N.,
exhibited substantial neglect on standard visual perceptual
tasks. The following two experiments assessing visual
attention were administered for two reasons. First, experience
in our laboratory has demonstrated that line bisection,

1165

Fig. 3 A CT scan image (C.I.) demonstrating infarction of the
right inferior frontal gyrus, lateral portions of the pre- and
postcentral gyri and portions of the supramarginal, superior and
middle temporal gyri on the right. A small left occipital lobe
infarct is also noted.

cancellation and a number of other clinical measures of
neglect may be insensitive to subtle deficits; i.e. we have
encountered patients who do not exhibit a deficit on linebisection and cancellation tasks yet exhibit a clear and
substantial lateralized impairment on the more demanding
tasks described below (see also Posner et al., 1984). Secondly,
the tasks were included to provide more direct measures of
the integrity of visual attention.

Experiment 1: attentive and pre-attentive
processes in vision
An experiment similar to that described by Treisman and
Souther (1985) assessing ‘pre-attentive’ and ‘attentionrequiring’ visual processes was performed. The experimental
paradigm requires that subjects discriminate between circles
(or ‘O’s) and circles of the same size with an intersecting
vertical line at the ‘six o’clock’ position (‘Q’s). In the preattentive condition the task is to detect a ‘Q’ in an array of
‘O’s; in this task, detection of a ‘Q’ is normally rapid,
effortless and independent of the size of the array. In the
attentive condition, the task is to detect an ‘O’ in an array
of ‘Q’s; normal subjects show an effect of array size in this
condition, presumably reflecting the need to carry out a serial

1166

H. B. Coslett
Table 1 Pre-attentive and attentive processes in vision
M.N.

Simple RT
Pre-attentive (six items)
Pre-attentive (12 items)
Attentive (six items)
Attentive (12 items)

C.I.

Left

Right

Left

Right

522 6 93
(0)
647 6 127
(0)
582 6 74
(0)
750 6 120
(0)
1124 6 291
(0)

549 6 149
(0)
624 6 106
(0)
572 6 48
(0)
776 6 146
(0)
1017 6 256
(0)

948 6 290
(0)
2114 6 678
(25)
2702 6 821
(50)
NR
(100)
NR
(100)

786 6 137
(0)
1425 6 530
(0)
1718 6 652
(0)
2262 6 145
(90)
3989
(95)

The values given are mean RTs in milliseconds 6 SD (and % misses). NR 5 not recorded.

search of the array. Extensive experience with this task with
both normal (Fitzpatrick-DeSalme et al., 1988) and braindamaged subjects (e.g. Stark et al., 1996; Coslett et al., 1995)
has demonstrated it to be a sensitive test of attention-requiring
mechanisms in visual perception.
The experiment was performed in two sessions; there was
a total of five conditions. In both the pre-attentive and
attentive conditions, two set sizes of six and 12 items (Os or
Qs) were used. In addition, a simple reaction-time task was
included as a baseline measure. This task required detection
of a single ‘O’ on an otherwise blank field. For all five
conditions, the target appeared with equal frequency in each
quadrant of the display. Stimuli were presented with a
Gerbrands four-field tachistoscope.
There were 80 trials in the simple reaction time (RT), preattentive (six item), pre-attentive (12 item), attentive (six
item) and attentive (12 item) conditions. The target was
present on half of the trials in each condition. Exposure
duration of the stimuli was controlled by a microcomputer
which provided millisecond accuracy. Trials were initiated
by the experimenter. Each trial consisted of a 1000-ms
fixation point, followed by a stimulus card which was
presented until the subject responded or for a maximum of
4 s. The subject’s task was to depress the telegraph key if
the target item (O or Q) was present and to refrain from
responding if the target was absent. Feedback was provided
only on practice trials. Note that as subjects were to respond
only to the presence of the target, the numbers in parentheses
in Table 1 indicate the percentage of trials on which subjects
failed to detect the target rather than the overall error rate.

Results
As indicated in Table 1, M.N. performed normally on this
task. Like controls, she exhibited a significant effect of
array size on the attentive task for both the left (t 5 5.32,
P , 0.05) and right (t 5 3.66, P , 0.05) hemispaces; no
effect of array size was noted on the pre-attentive task (left:
t 5 1.67; right: t 5 1.03). Additionally, there were no

differences in accuracy for stimuli presented on the left
compared with the right.
C.I., in contrast, demonstrated a profound impairment on
this task. RTs were significantly longer on the left compared
with the righ, on the simple RT and both pre-attentive tasks
(all P , 0.05). He was slower to respond on the pre-attentive
(12 item) compared with the pre-attentive (six item) condition
but this difference was not significant for either the right or
left. Additionally, he made significantly more errors on leftcompared with right-sided stimuli on both the pre-attentive
(six item) (Fisher’s exact probability 5 0.024) and preattentive (12 item) (Fisher’s exact probability 5 0.0002)
tasks. Age-matched controls miss an average of 2% of
stimuli. C.I. performed quite poorly on the attentive tasks,
detecting only three of 40 targets during the 4 s display on
the attentive (six item and 12 item) conditions combined; all
of the detected targets were situated on the far right of
the display. Controls make fewer than 5% errors on both
attentive tasks.
In summary, M.N. performed as well as age-matched
controls on this demanding test of visual search whereas
C.I. exhibited a pattern of performance similar to that
demonstrated by other patients with neglect in our laboratory
(see also Eglin et al., 1989).

Experiment 2: shift of attention
Posner et al. (1984) reported that patients with parietal
lobe lesions are impaired in shifting attention into the
contralesional hemispace when anticipating a target in the
ipsilesional hemispace; this deficit was attributed to an
impairment in disengaging attention from the cued location.
As this ‘disengage deficit’ has been regarded by some to
be a fundamental component of the neglect syndrome, an
experiment similar to that described by Posner et al. (1984)
was performed to assess the mechanisms controlling the shift
of attention. In this experiment, the patient was asked to
respond as quickly as possible to a stimulus presented laterally
in either the right or left visual field. A cue was presented

Neglect and visual imagery

1167

Table 2 Shift of attention
M.N.

C.I.

Left

Right

Left

Right

Interstimulus interval 5 50 ms
Valid
487 6 89
Invalid
613 6 183
Validity effect
126

460 6 121
534 6 121
74

840 6 203
1402 6 456
562

754 6 291
694 6 247
260

Interstimulus interval 5 150 ms
Valid
444 6 92
Invalid
472 6 117
Validity effect
28

461 6 86
527 6 75
66

682 6 177
1299 6 393
617

713 6 180
684 6 163
229

Interstimulus interval 5 500 ms
Valid
392 6 86
Invalid
365 6 53
Validity effect
227

361 6 74
349 6 103
212

566 6 126
870 6 419
304

570 6 148
567 6 192
23

Interstimulus interval 5 1000 ms
Valid
400 6 81
Invalid
301 6 36
Validity effect
299

331 6 64
310 6 95
221

445 6 76
704 6 234
259

430 6 124
518 6 137
88

Shift of attention is given in milliseconds (6SD).

Table 3 Report of visual images (number of items
reported)
M.N.

Living room
Dinging room
Kitchen
Route description

C.I.

Left

Right

Left

Right

3
4
2
3

14
10
7
15

9
6
6
–

8
8
5
–

at a varying inter-stimulus interval (ISI) prior to the stimulus.
On most trials, the cue was presented at the same site at
which the stimulus subsequently appeared (‘valid’ trials); on
some trials, however, the cue appeared at one position but
the stimulus subsequently appeared in the opposite visual
field (‘invalid’ trials). Normal subjects typically respond
more rapidly on valid compared with invalid trials,
presumably because the cue permits them to shift attention
to the probable location of the target; as demonstrated by
Posner et al. (1984), subjects with left-sided neglect typically
exhibit greatly prolonged RTs on invalid trials in which the
cue appears in the right visual field and the stimulus appears
in the left or neglected visual field.
On each trial, three outline boxes were presented, one box
was at the midline and the others were centred 5° to the
right and left of the midline. On valid trials, the box within
which the target subsequently appeared brightened for a
period of 300 ms. On invalid trials, the box on the opposite
side brightened for the same length of time. On neutral trials,
the central box brightened. The target was a filled diamond
which appeared within one of the laterally placed boxes.
There were an equal number of trials on which the ISI were
50, 150, 500 and 1000 ms.

Two-hundred trials were performed in each of two sessions.
Seventy-two percent of the trials were valid, 20% were
invalid and 8% were neutral. There were equal numbers of
valid, invalid and neutral trials for each ISI. Trials were
presented in a random sequence. Subjects were instructed to
maintain fixation on the middle box. The subject’s eye
position was monitored by the investigator; those trials on
which subjects shifted their gaze were repeated at the end of
the session.

Results
M.N. performed normally on this task. She exhibited validity
effects (i.e. RT on invalid trials minus RT on valid trials)
that are well within the range of normal in our laboratory;
more importantly, the validity effects were comparable in the
right and left hemispaces.
The RT data (excluding the neutral trials which were
judged to be too few to analyse) were analysed with an
analysis of variance (ANOVA) in which factors included
target side (right, left), validity (valid, invalid) and ISI
(50, 150, 500 and 1000 ms). M.N. exhibited main effects
of validity (F 5 7.393, P 5 0.007) and ISI (F 5 59.707,
P 5 0.001); she responded significantly faster to valid
(407 6 112 ms) compared with invalid (434 6 151 ms)
trials. Tukey’s test (P , 0.05) demonstrated that M.N. was
significantly slower with ISIs of 50 ms (495 6 122 ms)
and 150 ms (463 6 93 ms) compared with ISIs of 500 (372
6 80 ms) or 1000 ms (353 6 82 ms). Finally, there was a
validity–ISI interaction (F 5 9.455, P 5 0.001) reflecting
the fact that validity effects diminished with longer ISIs. We
note in this context that M.N. was faster (but not significantly)
on invalid trials with long ISIs. Although we have encountered

1168

H. B. Coslett

this pattern of performance in control subjects in our laboratory, we can offer no clear explanation; one possible account
is that on trials with long ISIs subjects assume that the target
will not appear at the cued location and therefore shift their
attention to the uncued location prior to target presentation.
C.I., in contrast, performed abnormally, exhibiting
strikingly asymmetric validity effects. ANOVA revealed
main effects of target (F 5 71.677, P 5 0.0001), validity
(F 5 61.477, P 5 0.0001) and ISI (F 5 43.539, P 5
0.0001) as well as significant target–validity (F 5 62.178,
P 5 0.0001), target–ISI (F 5 5.946, P 5 0.0006) and target–
validity–ISI (F 5 4.508, P 5 0.0041) interactions. From the
perspective of the present investigation, the most relevant
finding is the target–validity interaction; post hoc analysis
with Tukey’s test demonstrated that RTs to left invalid
trials (1069 6 529 ms) were significantly slower than RTs
for right invalid (616 6 151 ms) and left (633 6 231
ms) and right (616 6 214 ms) valid trials (P , 0.05);
the latter three conditions did not differ from each other.
These findings are consistent with those reported in
other patients with neglect (Posner et al., 1984; Coslett
et al., 1988).
Thus, M.N., who demonstrated no evidence of neglect
on routine clinical tasks, also performed normally on two
sensitive tests of attention-requiring visual processing. On
the basis of these data, we conclude that the patient did
not exhibit an impairment in attentional processes involved
in visual perception. C.I., like other patients with neglect
tested in our laboratory, exhibited substantial deficits on
both tasks assessing visual attention.

Experiment 3: report of visual images
Methods
In order to assess the patients’ ability to generate and
analyse visual images of arrays, M.N. and C.I. were asked
to report the contents of rooms in their homes as viewed
from different imagined perspectives. While sitting in our
laboratory, M.N. and C.I. were asked to imagine and list
the contents and features of their living rooms as viewed
from the main entrance to the room; on different occasions,
they were asked to report what they would see in the
living room when gazing into the room from the door at
the opposite end of the room. Additionally, the patients
were asked to describe the contents and layout of their
kitchens and dining rooms in a similar fashion. The
patients were encouraged to ‘look around’ and were given
unlimited time to respond. They were free to move their
head and eyes. The patient’s responses were recorded and
subsequently compared with a detailed floor plan of the
house derived from photographs taken by the investigator
on visits to the patients’ homes.

Results
M.N. consistently reported more objects (e.g. potted plant)
and architectural features (e.g. door to the cellar) on the

right compared with the left, for all of the imagined
rooms. Summing across both of the imagined viewpoints
for the living room (i.e. from the main entrance and from
the dining room door), she reported 14 attributes from the
right side of her image but only three from the left side.
Similarly, she reported 10 attributes from the right side of
the dining room and four from the left and for the kitchen,
she named seven from the right and two from the left. In
all instances, all items reported from the imagined left
side of the room were also reported when they were on
the imagined right side of the room from a different viewpoint; thus, the total number of items reported for the
living room, dining room and kitchen were 14, 10 and
seven. M.N. reported items from the right side of the
imagined array more reliably for the living room (Fisher’s
exact probability 5 0.004), dining room (Fisher’s exact
probability 5 0.01) and kitchen (Fisher’s exact probability 5 0.03).
Data from C.I. were scored in a similar fashion.
Summing across both imagined perspectives, C.I. reported
eight attributes from the right side and nine attributes from
the left for the living room. Similarly, he reported eight
from the right and six from the left side for the dining
room and five from the right compared with six from the
left for the kitchen. The total number of different items
reported for the living room, dining room and kitchen
were 10, eight and six. The difference in report from the
right compared with the left did not approach significance
for any of the rooms (all Fisher’s exact probabilities .2).
M.N. was also asked to describe what she would
encounter in an approximately 15 block walk from her
former home to the home of a close friend; on a separate
occasion she was given the same instructions but asked
to imagine that she was walking from her friend’s house
to her home. As confirmed by the investigator after he
retraced her route, she reported nine landmarks from the
right side of her route compared with only two from
the left side when walking to her friend’s home; she
reported six from the right but only one from the
left when describing her return walk. This pattern of
performance, which has been described by other investigators in patients with overt neglect (Brain, 1941; see also
Bisiach et al., 1993), was also observed when asked to
report the landmarks on her block from different imagined
perspectives.

Discussion
We report data from two patients demonstrating a double
dissociation between neglect on visual imagery and visual
perception tasks. These data are relevant to a number of
competing accounts of neglect and visual imagery more
generally.
Before considering potential explanations for these data, it
may be useful to consider briefly the (relatively uninteresting)
account which concerns differences in difficulty between the

Neglect and visual imagery
visual imagery and visual perception.One might suggest that
visual imagery simply requires more processing resources
than visual perception (see Ogden, 1985). On this account,
the neglect of the left side of imagined arrays as exhibited
by M.N. and other patients (Anderson et al., 1993; Guariglia,
1993) might plausibly be attributed to the increased task
demands associated with visual imagery. This hypothesis
does not, however, provide an adequate explanation for the
double dissociation reported here. If deficits on imagery tasks
reflect the fact that imagery requires greater processing
resources than visual perception, C.I., who exhibits prominent
perceptual deficits, would be expected to perform quite poorly
indeed on the imagery tasks. Thus, the normal performance
on imagery tasks exhibited by C.I. strongly suggests that
differences in task difficulty alone do not account for the
dissociations reported here (see also Bartolomeo, 1994).
How, then, are these data to be explained? These data are
inconsistent with the hypothesis, initially advanced by Bisiach
et al. (1978, 1979), that neglect is attributable to a defect in
an analogue spatial representation system common to both
visual perception and imagery. This account may explain the
performance of patients exhibiting a concordance between
neglect in imagery and perceptual tasks but does not readily
accommodate the dissociation reported here.
Similarly, the hypothesis of Meador et al. (1987), according
to which the deficit in imagery tasks is attributable, at least
in part, to a failure to direct attention to the left side of
the spatial representation, does not predict the dissociation
reported here. If the same intentional systems serve to control
visual attention and motor systems, as implied by Meador
et al. (1987), and M.N. is impaired in directing her attention
to the left side of imagined arrays, how is she able to perform
well on standard tests of visual processing as well as the
demanding tests of visual attention described above which
are assumed to depend critically on the ability to direct
attention to the left?
To explain the double dissociation in performance exhibited
by M.N. and C.I. one must postulate that their deficits
arise from different processing impairments. In addition,
assumptions regarding the cognitive architecture of the visual
and motor systems must be elaborated.
One possible account of M.N.’s performance is that at
least two related but isolatable visuo-spatial representations
are computed from visual input (Milner et al., 1991; Goodale
et al., 1986). The first may be an on-line, real time spatial
representation in which retinotopic information, integrated
with data about head and eye position, is transcoded into a
coordinate system capable of directly accessing motor systems
controlling the hand (Bridgman, 1991) and eyes (Wong and
Mack, 1981). As proposed by other investigators, this system
may correspond at least in part to ‘motor’ (Brigdeman, 1991)
or ‘sensorimotor’ (Paillard, 1987) procedures for reaching.
A second visuo-spatial representation may also be
computed in which retinotopic information, integrated with
data about head and eye position, is represented in an
egocentric coordinate system and maintained across saccades,

1169

perhaps for intervals of a few seconds. We have previously
suggested that a representation of this type may support
judgements about the relative position and size of objects in
the environment as well as providing, at least under certain
circumstances, access to motor systems (Stark et al, 1996;
see also Paillard, 1987; Bridgeman, 1991). This representation
may correspond, at least in part, to the ‘cognitive’ stream
postulated by Bridgman (1991) and Paillard (1987).
Support for these different levels of representation comes
from several sources (see also Goodale et al., 1986). Milner
et al. (1991), for example, reported a patient who exhibited
a striking discrepancy between her good performance in online reaching tasks and her poor performance on tasks
requiring judgments about the size, shape and position of
visual stimuli. For example, although this patient was unable
to match objects reliably on the basis of shape and size,
when reaching to pick up these same objects her hand
assumed a posture appropriate to the object in question
(Goodale et al., 1991).
We (Stark et al., 1996) recently reported a patient, G.W.,
whose performance is also relevant in this context. G.W.
exhibited dramatic differences in her performance on tasks
requiring that she respond to visualized targets compared
with those when she was asked to point to, or make
judgements about, information not currently in view. For
example, G.W. performed well (although not perfectly) when
asked to touch a (currently) visualized target; however, when
asked to touch the same target after closing her eyes, she
groped blindly, often insisting that the target had ‘gone’.
G.W. also performed poorly on a variety of tasks requiring
that she register and maintain spatial information in a format
which supported judgments about the location (but not
identity) of the stimulus. Additionally, G.W. accurately
reported the contents of a room in which she was sitting
with her eyes open, but with her eyes closed she was unable
to describe the layout or contents of the same room or,
indeed, the home in which she had lived for 20 years. G.W.,
who suffered from a gradually progressive disorder, exhibited
focal atrophy in Brodmann’s area 7 bilaterally on MRI scan.
We suggest that M.N.’s performance can be accommodated
on the hypothesis outlined above according to which at least
two visuo-spatial representations are computed. On this
hypothesis, M.N.’s good performance on visual perceptual
and attentional tasks reflects the integrity of the ‘direct’
procedure for accessing output systems. Her impaired
performance on the imagery tasks, in contrast, is attributable
to an impairment of the relatively long-lasting egocentric
visuo-spatial representation which we have termed the ‘master
map of space’ (Stark et al., 1996). Unfortunately, we are
unable to specify the nature of this deficit.
On the account briefly outlined above there are several
possible explanations for the pattern of behaviour
demonstrated by C.I.. One possible account is that he suffers
from an impairment in the ‘direct’, on-line system mediating
motor interactions with visualized stimuli, whereas the master
map of space, which we have suggested is critical for visual

1170

H. B. Coslett

imagery (Stark et al., 1996), is preserved. Furthermore, in
light of his poor performance on Experiments 1 and 2, one
might speculate that C.I.’s deficit is attentional in nature.
A second possible explanation is that C.I. accurately
computes both visuo-spatial representations as described
above but that his impaired performance is attributable to a
‘directional hypokinesia’ (Heilman et al., 1985; Coslett et al.,
1990) or premotor deficit characterized by an impairment in
the control of movement into or towards the neglected
hemispace. As previously noted, Meador et al. (1987)
proposed such an account for their patient who exhibited
neglect of the left in both perceptual and imagery tasks. Note
also that this account requires an additional assumption to
explain the dissociation in performance exhibited by C.I. on
perceptual and imagery tasks. More specifically, one must
assume that the intentional mechanism mediating the
movement of the extremities and, perhaps, eyes into or toward
the neglected hemispace is distinct from the mechanism
controlling the scanning or search of mental images (cf.
Kosslyn, 1994). C.I.’s deficit, on this assumption, would be
attributed to a directional hypokinesia involving the former
but not the latter mechanism. This hypothesis is weakened
by the fact that, as previously noted, C.I. exhibited evidence
of hemispatial inattention but not directional hypokinesia
when tested with the paradigm reported by Coslett et al.
(1990).
A third possibility is that C.I. does indeed suffer from an
impairment in the exploration or representation of stored
images, but that he compensates for this limitation by dividing
the task into multiple, sequential operations. For example,
from this hypothesis one might suggest that when asked to
report the contents of a room, a subject with neglect might
report items from the right side of the array only; if aware
of the deficit, however, the subject might assume a different
imagined perspective from which the originally omitted
objects were on the right side of the image. Thus, by shifting
the imagined perspective, a neglect of the left side of an
image could be masked (see also Bartolomeo et al., 1994).
Although this hypothesis cannot be refuted definitively, we
believe it to be an unlikely explanation for C.I.’s performance
for two reasons. First, the hypothesis assumes that C.I. was
either explicitly or implicitly aware of his deficit and acted
to minimize its consequences. There was no evidence from
perceptual tasks that C.I. was aware of his deficits; indeed,
on many occasions a striking discrepancy was noted between
his profound neglect and his confident assertion that he had
performed well. Secondly, a reliance on a strategy of
switching imagined perspectives would be expected to be
associated with a predictable pattern of report of items; for
example, one would expect items from the same side of an
array to be clustered in the report sequence just as they are
in the array. Examination of the serial order of reported items
demonstrated no consistent clustering of items as a function
of the side of the imagined array on which they were located;
interestingly, informal inspection of the sequence of items
reported suggested that the report sequence was influenced

to a greater degree by semantic category (e.g. type of
furniture) than by location.
Although controversy persists, a number of investigators
have suggested that visual imagery is supported by at least
some of the same neural structures which mediate vision
(see Farah et al., 1992; Roland and Gulyas, 1994; and
following commentaries: Kosslyn, 1994; Behrmann et al.,
1995). It should be noted in this context that the account of
the data from M.N. and C.I. offered here does not contradict
this claim. We have suggested that the impairment in visual
imagery in association with preserved visual perception
exhibited by M.N. is attributable to a disruption of an
egocentric spatial representation which we have previously
suggested provides, at least in part, for the maintenance of
visual information across saccades and time, and which
articulates with cognitive systems. This putative spatial
representation, ‘the master map of space’, is an integral part
of the normal visual system. Thus, the dissociation exhibited
by M.N. and, perhaps, C.I. may be attributed to a disruption
of a subset of the neural structures and corresponding
representations which subserve normal vision.

Acknowledgements
The work described in this paper was presented, in part, at
the International Neuropsychological Society Meeting, 1989.
It was supported by grant RO1 AG08870 from the NIH.

References
Anderson B. Spared awareness for the left side of internal visual
images in patients with left-sided extrapersonal neglect. Neurology
1993; 43: 213–6.
Bartolomeo P, D’Erme P, Gainotti G. The relationship between
visuospatial and representational neglect. Neurology 1994; 44:
1710–4.
Bisiach E. Mental representation in unilateral neglect and related
disorders: the twentieth Bartlett Memorial Lecture. [Review]. Q J
Exp Psychol 1993; 46A: 435–61.
Bisiach E, Luzzatti C. Unilateral negelct of representational space.
Cortex 197814: 129–33.
Bisiach E, Luzzatti C, Perani D. Unilateral neglect, representational
schema and consciousness. Brain 1979; 102: 609–18.
Bisiach E, Brouchon M, Poncet M, Rusconi ML. Unilateral neglect
in route description. Neuropsychologia 1993; 31: 1255–62.
Brain W R. Visual disorientation with special reference to lesions
of the right cerebral hemisphere. Brain 1941; 64: 244–72.
Bridgeman B. Complementary cognitive and motor image
processing. In: Obrecht G, Stark LW, editors, Presbyopia research.
New York: Plenum Press, 1991: 189–98.
Coslett HB, Saffran EM, Fitzpatrick E. Impaired movement of
selective attention in neglect [abstract]. J Clin Exp Neuropsychol
1988; 10: 21.

Neglect and visual imagery
Coslett HB, Bowers D, Fitzpatrick E, Haws B, Heilman KM.
Directional hypokinesia and hemispatial inattention in neglect. Brain
1990; 113: 475–86.
Coslett HB, Stark M, Rajaram S, Saffran EM. Narrowing the
spotlight: a visual attentional disorder in Alzheimer’s disease.
Neurocase 1995; 1: 305–18.
Eglin M, Robertson LC, Knight RT. Visual search performance in
the neglect syndrome. J Cogn Neurosci 1989; 1: 372–85.
Farah MF, Soso MJ, Dasheiff RM. Visual angle of the mind’s eye
before and after unilateral occipital lobectomy. J Exp Psychol Hum
Percept Perform 1992; 18: 241–6.
Fitzpatrick EJ, Coslett HB, Saffran EM. Preattentive and attentive
processing in brain-damaged patients [abstract]. J Clin Exp
Neuropsychol 1988; 10: 29.

1171

Meador KJ, Loring DW, Bowers D, Heilman KM. Remote memory
and neglect syndrome. Neurology 1987; 37: 522–6.
Milner AD, Perrett DI, Johnston RS, Benson PJ, Jordan TR, Heeley
DW, et al. Perception and action in ‘visual form agnosia’. Brain
1991; 114: 405–28.
Ogden JA. Contralesional neglect of constructed visual images in
right and left brain-damaged patients. Neuropsychologia 1985; 23:
273–7.
Posner MI, Walker JA, Friedrich FJ, Rafal RD. Effects of parietal
injury on covert orienting of attention. J Neurosci 1984; 4: 1863–74.
Paillard J. Cognitive versus sensorimotor encoding of spatial
information. In: Ellen P, Thinus-Blanc C, editors. Cognitive
processes and spatial orientation in animal and man, Vol II.
Dordrecht: Martinus Nijhoff, 1987: 43–77.

Goodale MA, Pelisson D, Prablanc C. Large adjustments in visually
guided reaching do not depend on vision of the hand or perception
of target displacement. Nature 1986; 320: 748–50.

Roland PE, Gulyas B. Visual imagery and visual representation [see
comments]. Trends Neurosci 1994; 17: 281–7. Comment in: Trends
Neurosci 1994; 17: 287–94, Comment in: Trends Neurosci 1994;
17: 514–6.

Goodale MA, Milner AD, Jakobson LS, Carey DP. A neurological
dissociation between perceiving objects and grasping them [see
comments]. Nature 1991; 349: 154–6. Comment in: Nature 1991;
349: 102–3, Comment in: Nature 1991; 351: 194–5.

Stark M, Coslett HB, Saffran EM. Impairment of an egocentric
map of locations: implications for perception and action. Cogn
Neuropsychol 1996; 13: 481–523.

Guariglia C, Padovani A, Pantano P, Pizzamiglio L. Unilateral
neglect restricted to visual imagery [see comments]. Nature 1993;
364: 235–7. Comment in: Nature 1993; 364: 193–4.
Heilman KM, Bowers D, Coslett HB, Whelan H, Watson RT.
Directional hypokinesia: prolonged reaction times for leftward
movements in patients with right hemisphere lesions and neglect.
Neurology 1985; 35: 855–9.

Teuber HL, Battersby WS, Bender MB. Performance of complex
visual tasks after cerebral lesions. J Nerv Ment Dis 1951; 114:
413–29.
Treiman A, Souther J. Search asymmetry: a diagnostic for
preattentive processing of separable features. J Exp Psychol Gen
1985; 114: 285–310.
Wong E, Mack A. Saccadic programming and perceived location.
Acta Psychol (Amst) 1981; 48: 123–31.

Kosslyn SM. Image and brain. Cambridge, MA: MIT Press, 1994.
Kosslyn SM, Behrmann M, Jeannerod M. The cognitive science of
mental imagery Neuropsychologia 1995; 33: 1335–44.

Received September 13, 1996. Revised December 20, 1996.
Accepted February 28, 1997