Motivation, Emotion, Feeding, Drinking, Sexual Behaviour
1111
2
3
4
5
6
7
8
9
10111
1
2
3
4
5
6
7
8
9
20111
1
2
3
4
5
6
7
8
9
30111
1
2
3
4
5
6
7
8
9
40111
1
2
3
4
5
6
7
8
9
50111
1
2
3
4
5
6111p

Website publication 22 September 1998

NeuroReport 9, 2919–2925 (1998)

PATIENTS with striate cortical damage causing a hemianopic field defect can nevertheless demonstrate residual
visual capacities in their blind field. Previous research
investigating blindsight required patients to respond
explicitly to stimuli appearing in the blind field by
making forced choice judgements. We present data from
a patient with a left occipital lesion resulting in a
homonymous hemianopia, using the flanker task. This
patient displayed a significant flanker congruency effect
(FCE) for colour and letter stimuli even when they
appeared in the blind field. A control patient with a lesion
of the right thalamus showed no FCE in the blind field.
This suggests that thalamo-extrastriate neural pathways
are necessary for residual functioning in blindsight.
NeuroReport 9: 2919–2925 © 1998 Lippincott Williams &
Wilkins.

Investigating form and
colour perception in
blindsight using an
interference task

Key words: Blindsight; Extrastriate cortex; Flanker task;
Thalamus

Introduction
Lesions of the striate visual cortex lead to visual field
defects in the contralateral hemifield corresponding
topographically to the damaged region of cortex.1
Despite their visual field defects, patients with lesions
of striate cortex may retain some residual visual
capacities in their blind field.2 This ‘blindsight’ has
been demonstrated on tasks which require the localization of stimuli in the blind field by saccades or
finger pointing, and the discrimination of motion,
wavelengths and simple forms.1,3–11 These tasks typically require patients to make forced choice guesses
regarding stimuli presented in the blind field.2,4 That
is, patients are asked to respond to stimuli they confidently assert they cannot see! However, guessing
paradigms have been criticized, as factors such as
changes in decision criteria and uncontrolled eye
movements that bring stimuli into the sighted field
may produce a pattern of results suggestive of blindsight when, in reality, blindsight is not occurring.12,13
0959-4965 © 1998 Lippincott Williams & Wilkins

James Danckert,1,2,CA Paul Maruff,2,3
Glynda Kinsella,1,4 Steven de Graaff4
and Jon Currie2,5
1

School of Psychological Science, Faculty of
Science and Technology, La Trobe University,
Bundoora, Victoria, Australia 3083;
2
Neurophysiology and Neurovisual Research
Unit, Mental Health Research Institute of
Victoria, Locked Bag 11, Parkville, Victoria,
Australia 3052; 3School of Biophysics and
Electrical Engineering, Swinburne University of
Technology, Hawthorn, Victoria, Australia 3122;
4
Caulfield General Medical Centre, Caulfield,
Victoria, Australia 3162; 5Brain Research Unit,
Drug and Alcohol Services, Westmead
Hospital, Sydney, New South Wales,
Australia 2145
CA,2

Corresponding Author and Address

Several recent papers have addressed these criticisms
in well controlled experimental conditions with findings generally suggesting that near threshold vision,
changes in decision criteria and light scatter are
unlikely to be responsible for the performances of
patients with blindsight.14
A more elegant method for assessing the processing of information in the blind field, has been to
measure the interference caused by stimuli presented
in the blind field on responses to targets presented
in the sighted field.5,15–17 For example, both saccades
and attentional shifts towards targets in the sighted
field are slowed by prior presentation of stimuli
to the blind field.5,15 In addition, normal spatial
summation effects are observed when targets are
presented simultaneously in the sighted and blind
visual fields.16,17 The advantage of these paradigms is
that patients are never asked to respond explicitly to
stimuli appearing in their blind field, avoiding some
of the problems of the forced choice guessing paradigms mentioned above.12,17
Vol 9 No 13 14 September 1998

2919

J. Danckert et al.
1111
2
3
4
5
6
7
8
9
10111
1
2
3
4
5
6
7
8
9
20111
1
2
3
4
5
6
7
8
9
30111
1
2
3
4
5
6
7
8
9
40111
1
2
3
4
5
6
7
8
9
50111
1
2
3
4
5
6111p

The flanker task is an interference paradigm that
may be useful for assessing blindsight. The task
requires participants to respond to central targets
while ignoring flanking stimuli that can be either
congruent or incongruent with the target.18 Reaction
times (RT) to identify targets are faster when flankers
and targets are congruent (e.g. both red) than
when targets and flankers are incongruent (e.g. red
target/green flanker). This RT difference has been
termed the flanker congruency effect (FCE) and
has been demonstrated for a wide range of stimuli,
including letters, colours and line orientation.19,20
With targets presented in the sighted field and
flankers presented in the blind field, the FCE may
provide a reliable index of processing in the blind
field.
We used two flanker tasks, one with colour stimuli
and the other with letter stimuli, to investigate the
residual visual capacities of a patient with a left occipital lobe infarct (patient AG) and a patient with a
lesion of the right thalamus (patient CP). CP’s lesion
gave rise to a visual field defect that was qualitatively
similar to that of AG’s, however the site of the lesion
was such that visual processing in both the thalamus and thalamo-extrastriate neural pathways were
disrupted. It was hypothesized that AG would show

a significant FCE when flankers appeared in the blind
field, whereas CP would not show any FCE when
flankers appeared in the blind field.

Materials and Methods
Patient AG is a 66-year-old male, admitted to
hospital for abdominal surgery. After surgery he
experienced hypertensive episodes and an acute
confusional state. Upon recovery from these episodes
he exhibited mild aphasia, cerebellar ataxia and a mild
left hemiplegia, which has since resolved. CT scans
indicated an area of hypodensity with a well defined
border in the left occipital lobe involving both grey
and white matter, suggestive of recent infarction. CT
showed no evidence of any intracerebral haemorrhages and sulcal and ventricular sizes were within
normal limits (Fig. 1A). Goldman Perimetry testing
of AG’s visual fields indicated a dense, macula splitting right visual field (RVF) homonymous hemianopia. Experimental testing with patient AG was
conducted approximately 6 weeks after the onset of
his stroke.
CP is an 80-year-old male who suffered a right
hemisphere stroke affecting deep grey matter structures in 1995. An MRI scan indicated substantial

FIG. 1. Outline of area of infarction as traced from CT scan results for patient AG (A) and patient CP (B). Lesion areas are depicted in
cross-hatched regions and ventricles are depicted in black.

2920 Vol 9 No 13 14 September 1998

Form and colour perception in blindsight
1111
2
3
4
5
6
7
8
9
10111
1
2
3
4
5
6
7
8
9
20111
1
2
3
4
5
6
7
8
9
30111
1
2
3
4
5
6
7
8
9
40111
1
2
3
4
5
6
7
8
9
50111
1
2
3
4
5
6111p

damage to the right thalamic nuclei with destruction
of the lateral geniculate nucleus and surrounding
white matter (Fig. 1B). No other mass affects were
observed and ventricular and sulcal dimensions were
within normal limits. Goldman Perimetry testing
indicated a dense, macular splitting left visual field
(LVF) homonymous hemianopia.
Ten healthy, older controls completed each experiment (eight female, two male; mean (± s.d.) age 65.9
± 5.1 for the colour flanker, and six female, four male;
age 69.4 ± 7.1 for the letter flanker). Controls were
recruited from the community and exclusion criteria
included a past history of neurological or psychiatric
disorders, drug abuse, vascular pathology or
traumatic brain injury. The protocol was approved
by the institutional ethics committee and informed
consent was obtained prior to commencement.
All tasks were created using Micro-Electronic
Laboratory software (MEL-2)21 and were presented
on a standard PC. Participants were seated 50 cm
from the screen in a dimly lit, sound attenuated room.
Background luminance of all displays was 1 cd/m2,
and all stimuli were presented at 1.2 log units above
background luminance. Prior to completing the
flanker tasks, each patient performed simple detection tasks using both letter and colour stimuli to
ensure they were unaware of flanker task stimuli
when it appeared in their blind field. Targets in these
tasks appeared 6.5° and 13° to the left and right of a
central fixation cross. Stimuli remained present for
between 3000 and 7000 ms or until a response was
made. Each task consisted of 50 trials and lasted
~5 min. Neither patient was able to detect any targets
appearing in their blind field at either eccentricity.
Two versions of the flanker task were used, one
using colour targets and a second using letter targets.
Patients had to verbally identify targets in each task.
Colour targets consisted of a red or green colour
patch subtending 1.7° of visual angle. Flankers were
also either red or green in colour as previous research
has shown that patients with blindsight can make
wavelength discriminations corresponding to these
colours in the blind field.1,3 Colour flanker stimuli
subtended 4.6° of visual angle. Targets for the letter
flanker task consisted of the upper case letters E and
O subtending 1.4° of visual angle. Flankers were
also upper case E and O subtending 2° of visual
angle. Flankers and targets were monochromatic. All
other aspects of the flanker tasks were identical.
Flankers could appear to the left or right of the
target with equal probability. Targets also appeared
without flankers (no-flanker condition) and equivalent numbers of trials with and without flankers were
included in each task. The latency of verbal responses
was recorded using a microphone and vocal reaction
times (VRT) were defined as the time between

stimulus onset and voice onset as detected by the
microphone. Following each trial the experimenter
entered the participant’s response via a button box
to measure accuracy. The factors of visual field (i.e.
LVF vs RVF) and congruency of flankers (ie. congruent, incongruent or no flanker) were randomized
within tasks. Each participant completed 20 practice
trials for calibration of the microphone and to allow
participants to become familiar with the task. On the
basis of the results from the simple detection tasks,
flankers appeared 6.5° to the left or right of the target.
For both tasks, a single trial consisted of the
appearance of a central fixation stimulus which signalled the beginning of the trial and informed the
participant to fix their eyes on the centre of the screen
where targets appeared. After a variable period of
time the target and flanker appeared simultaneously
and remained present until the participant responded.
The next trial began after the experimenter had
entered the participant’s response. Participants
completed three blocks of 60 trials with rest periods
in between. At the end of each block both patients
were asked if they had seen any stimuli appearing in
their blind field to obtain a subjective verbal report
concerning conscious perception of stimuli in the
blind field. Neither patient reported seeing any
stimuli in their blind field.
Eye movements were monitored by the experimenter and trials in which eye movements occurred
were abandoned and rescheduled to the end of the
block.15 Patient AG made eye movements away from
fixation on < 3% of trials in both flanker tasks, with
some indication that he was more likely to make eye
movements towards flankers appearing in the sighted
field. Similarly, patient CP made very few eye
movements in either flanker task (< 2%) with no
suggestion of any trend or pattern in the trials on
which CP made eye movements. For each participant
in each task, individual VRT distributions were
inspected for anticipatory responses (VRT < 2nd
percentile) and abnormally slow responses (VRT >
98th percentile).15 These data were excluded from
further analysis. For controls, performance in each
flanker task was compared by submitting mean
VRT for each congruency condition (congruent,
incongruent and no flanker) in each visual field
(LVF, RVF) to repeated measures ANOVA. For the
patients, median RT from the congruent, incongruent
and no flanker conditions in both the sighted and
blind fields were compared using Wilcoxin matchedpairs signed ranks test. For all participants a FCE
was also calculated by subtracting VRT to the
congruent condition from VRT to the incongruent
condition.20 The magnitude of this effect was
compared between visual fields using paired samples
t-tests.
Vol 9 No 13 14 September 1998

2921

J. Danckert et al.
1111
2
3
4
5
6
7
8
9
10111
1
2
3
4
5
6
7
8
9
20111
1
2
3
4
5
6
7
8
9
30111
1
2
3
4
5
6
7
8
9
40111
1
2
3
4
5
6
7
8
9
50111
1
2
3
4
5
6111p

Results
For controls performance on the colour flanker
task, ANOVA indicated a significant main effect of
congruency (F(1,9) = 16.71, p < 0.005), but no effect
of visual field and no two-way interaction (Fig. 2A).
Similarly, for the letter flanker task, ANOVA indicated a significant main effect of congruency (F(1,9)
= 10, p < 0.05), but no effect of visual field and no
two-way interaction (Fig. 2B).
With data collapsed across visual fields, there was
no significant difference between VRT to the congruent and no flanker conditions. In contrast, VRTs
in the incongruent condition were significantly slower
than those in the congruent condition for both tasks
(colour flanker: t(9) = –4.09, p < 0.005; letter flanker:
t(9) = –3.67, p < 0.01). This pattern of performance
yielded a positive FCE in both tasks (Table 1).
On the colour flanker task patient AG showed a
significant difference between median VRTs in the
congruent and incongruent conditions when flankers
appeared in both the sighted (FCE = 118 ms; Z =
–1.99, p < 0.05) and blind fields (FCE = 74; Z = –2.25,
p < 0.05; Fig. 3A), which gave rise to a positive
FCE in both AG’s sighted and blind visual fields
(Table 1). Patient CP showed a significant difference
between median VRT in the congruent and incongruent conditions when flankers appeared in the
sighted field (FCE = 144 ms; Z = –3.376, p < 0.001;
Fig. 3B) but not when flankers appeared in the
blind field (FCE = –9 ms). Thus, CP demonstrated a
positive FCE only when colour flankers appeared
in his sighted visual field (Table 1). For the letter
flanker task patient AG showed a significant difference between median VRT in the congruent and
incongruent conditions when flankers appeared in

both the sighted (FCE = 114 ms; Z = –2.2, p < 0.05)
and blind fields (FCE = 53 ms; Z = –2.94, p < 0.005;
Fig. 4A), which gave rise to a positive FCE in both
AG’s sighted and blind visual fields (Table 1). Patient
CP showed a significant difference between median
VRT in the congruent and incongruent conditions
when flankers appeared in his sighted field (FCE =
64 ms; Z = –1.66, p < 0.05; Fig. 4B) but not when
flankers appeared in the blind field (FCE = 2 ms).
Therefore, CP only demonstrated a positive FCE
when flankers appeared in his sighted visual field
(Table 1).

Discussion
These experiments show that AG could process
colour and form information in his blind field. This
blindsight was inferred from the pattern of interference from flankers presented in the blind field on
VRT to targets presented in the sighted field. Thus,
AG exhibited a normal FCE when flankers appeared
in his blind field even though he was unaware of
their presence (Table 1; Figs 3A,4A). In contrast, CP
showed no FCE when flankers appeared in his blind
field (Table 1; Figs 3B,4B). These results suggest that
residual visual processing in the blind field may
depend upon the integrity of thalamo-extrastriate
pathways in the damaged hemisphere.1,4,5
AG’s performances in these tasks are unlikely to
be due to eye movements for three reasons. First, in
the detection tasks, AG failed to detect any stimuli
appearing in his blind field for periods of ⭓ 3 s, which
would have allowed many eye movements to be
made. This suggests that AG had no difficulty maintaining central fixation. Second, in the flanker tasks,
the goal was to identify a central target, making it

FIG. 2. Performance of healthy older control participants in both the colour (A) and letter (B) flanker tasks. c, congruent condition; n, no
flanker condition; i, incongruent condition. All measurements are in ms and figures represented are means ± s.e.

2922 Vol 9 No 13 14 September 1998

Form and colour perception in blindsight
1111
2
3
4
5
6
7
8
9
10111
1
2
3
4
5
6
7
8
9
20111
1
2
3
4
5
6
7
8
9
30111
1
2
3
4
5
6
7
8
9
40111
1
2
3
4
5
6
7
8
9
50111
1
2
3
4
5
6111p

Table 1. FCE for both controls and patients on the colour and letter flanker tasks. All figures are
in milliseconds. For the controls figures represent means (±s.e.) while for the patients median
figures are given.

colour flanker task
letter flanker task

Controls

AG sighted

AG blind

CP sighted

CP blind

38.4 ± 35.2*
39.2 ± 32.5*

118*
114*

74*
53*

144*
64*

–9
2

* indicates a significant FCE

unnecessary for AG to make eye movements in order
to prepare a response. Third, the minimal number of
trials in which AG made eye movements occurred
mainly on trials in which flankers appeared in his
sighted field, suggesting he was distracted by flankers
he had conscious awareness of.
To date there have been three major criticisms of
theories of blindsight that invoke a second visual
pathway to explain the residual visual functions of
patients with striate lesions.12,13 First, it has been
suggested that patients asked to make guesses
regarding stimuli in their blind field may adopt lax
decision criteria.12 However, AG was not required
to make decisions about stimuli appearing in his
blind field. Instead, he had to identify central
targets. Therefore, the differential effects of flankers
presented to the blind field could not be due to
changes in decision criteria.14,16,17 A second criticism
is that light from stimuli in the blind field may scatter
to regions of the retina projecting to intact regions
of striate cortex.12 While we cannot conclusively
eliminate this explanation in the current set of experiments, it would seem unlikely to completely account
for AG’s blindsight. If the light from flankers
presented in the blind field had scattered into the
sighted field, then the same interference effects from
flankers should have been observed in both patients.

This was not the case. In addition, flankers were
presented 6.5° away from central fixation making it
unlikely that light scattered into the intact visual
field. A final criticism is that the residual capacities
in patients with blindsight may arise as a result of
processing in spared ‘islands’ of striate cortex.13 Even
though the results of AG’s CT scans (Fig. 1A) suggest
that the possibility of spared islands of striate cortex
is remote, this is by no means conclusive. However,
the results of this previous research implying that
spared islands of cortex mediate blindsight, have
proven difficult to replicate.22 Furthermore, AG was
never aware of stimuli presented in his blind field.
Therefore, while further research placing flankers in
different positions within the blind field would be
necessary to comprehensively address this issue, it
would seem unlikely that spared islands of cortex
could account for AG’s blindsight.13,22
Although it is not possible from the present data
to determine precisely the neural pathways responsible for AG’s residual visual capacities, it is possible
that processing in geniculo-extrastriate neural pathways was involved. Patient CP did not show any
effects of flanking stimuli appearing in his blind field.
CP’s thalamic lesion suggests that the integrity of
visual pathways involving the thalamus are necessary
for effective processing of visual information to

FIG. 3. Median VRT for AG (A) and CP (B) on the colour flanker task. All measurements are in ms. c, congruent condition; n, no flanker
condition; i, incongruent condition.

Vol 9 No 13 14 September 1998

2923

J. Danckert et al.
1111
2
3
4
5
6
7
8
9
10111
1
2
3
4
5
6
7
8
9
20111
1
2
3
4
5
6
7
8
9
30111
1
2
3
4
5
6
7
8
9
40111
1
2
3
4
5
6
7
8
9
50111
1
2
3
4
5
6111p

FIG. 4. Median VRT for AG (A) and CP (B) on the letter flanker task. All measurements are in ms. c, congruent condition; n, no flanker
condition; i, incongruent condition.

occur. In contrast, striate visual cortex does not
appear to be essential for visual processing to occur,
although it may be necessary for visual stimuli to
reach conscious awareness.3,4,23 As previous neurophysiological research suggests that the superior
colliculus does not contain colour opponent cells3 it
is unlikely that AG’s residual visual processing was
mediated by retino-tectal pathways.5 Therefore, the
implicit processing of colour and form stimuli exhibited by AG in his blind field suggests that this
residual function is mediated by thalamic and extrastriate neural pathways.24,25 Neurophysiological
research in monkeys has shown that the thalamus
has rich connections with extrastriate regions such as
V3 and V4, that have been shown to be necessary
for processing of colour and form.6 Similarly, neuroimaging studies in humans suggest that the thalamus,
and in particular the pulvinar nucleus, is important
in tasks requiring attention to both colour and form.25
The results of the current experiments suggest that
while striate visual cortex is not essential for normal
interference effects to occur, the integrity of neural
pathways between the thalamus and extrastriate
cortical regions may mediate processing of visual
properties such as colour and form. An interesting
question arising from the processing of form in the
blind field is whether or not it was the congruence
of form (ie. the outline or orientation of letters) or
the congruence of identity (i.e. semantic meaning)
that determined the interference effects observed in
the blind field.
The results of this study suggest that the flanker
task represents a reliable new methodology for
assessing the residual visual capacities of patients with
2924 Vol 9 No 13 14 September 1998

blindsight. As with several other paradigms, the
flanker task avoids the need for patients to respond
explicitly to stimuli appearing in their blind field,
eliminates issues of changes in decision criteria and
avoids any confounding effects of eye movements
that may bring stimuli in the blind field into the
sighted field.5,15–17 Finally, the flanker task explicitly
addresses the link between perception and response
and has potential utility for presenting a wide
variety of stimuli to the blind field of hemianopic
patients.18,19

Conclusion
The flanker task has proven to be a reliable means
for assessing the residual visual capacities of patients
with lesions of the striate visual cortex leading to
specific visual field defects. One patient with a lesion
of the left occipital lobe showed a reliable interference effect of flanking stimuli appearing in his blind
field on responses made to targets appearing in his
sighted field. This was true for both colour and letter
stimuli. This task can now be used to investigate in
greater detail the nature and extent of implicit
processing of stimuli presented to the blind field of
patients with lesions of striate cortex. A challenge
that arises from this research is the problem of delineating the nature of the implicit processing. That is,
do patients with lesions of the striate cortex simply
respond to primitive aspects of stimuli appearing in
their blind field (i.e. form outline, wavelength, orientation) or does the information presented in the blind
visual field receive more detailed analysis (i.e. identification of semantic or categorical meaning)?

Form and colour perception in blindsight
1111
2
3
4
5
6
7
8
9
10111
1
2
3
4
5
6
7
8
9
20111
1
2
3
4
5
6
7
8
9
30111
1
2
3
4
5
6
7
8
9
40111
1
2
3
4
5
6
7
8
9
50111
1
2
3
4
5
6111p

References
1. Stoerig P. Trends Neurosci 19, 401–406 (1996).
2. Weiskrantz L, Warrington EK, Sanders MD et al. Brain 97, 709–728 (1974).
3. Cowey A and Stoerig P. Trends Neurosci 14 140–145 (1991).
4. Milner AD and Goodale MA. The Visual Brain in Action. New York: Oxford
University Press (1995).
5. Rafal RD, Smith J, Krantz J et al. Science 250, 118–121 (1990).
6. Stoerig P and Cowey A. Behav Brain Res 71, 147–156 (1995).
7. Perenin MT and Jeannerod M. Neuropsychologia 16, 1–13 (1978).
8. Stoerig P and Cowey A. Nature 342, 916–918 (1989).
9. Perenin MT. NeuroReport 2, 397–400 (1991).
10. Castiello U, Scarpa M and Bennett K. Nature 374, 805–807 (1995).
11. Weiskrantz L. Q J Exp Psychol 32, 365–386 (1990).
12. Campion J, Latto R and Smith YM. Behav Brain Sci 6, 423–486 (1983).
13. Gazzaniga MS, Fendrich R and Wessinger CM. Curr Dir Psych Sci 3, 93–96
(1994).
14. Azzopardi P and Cowey A. Proc Natl Acad Sci USA 94, 14190–14194 (1997).
15. Maruff P, Sanbach J, Danckert J et al. Proceedings of the First Perception
for Action Conference, Monash University Press, Gippsland, Victoria, 1996;
115–121.

16. Marzi CA, Tassinari G, Aglioti S et al. Neuropsychologia 24, 749–758
(1986).
17. Tomaiuolo F, Ptito M, Marzi CA et al. Brain 120, 795–803 (1997).
18. Eriksen BA and Eriksen CW. Percept Psychophys 16, 143–149 (1974).
19. Miller J. Percept Psychophys 49, 270–288 (1991).
20. Rafal R, Gershberg F, Egly R et al. Neuropsychologia 34, 1197–1202
(1996).
21. Rodgers AK, Schneider W, Pitcher E et al. Mel Professional: Language
Reference Guide, (Vol 1). University of Pittsburgh: Psychology Software
Tools, Inc. Learning Research and Development Centre, 1995.
22. Kentridge RW, Heywood CA and Weiskrantz L. J Cogn Neurosci 9, 191–202
(1997).
23. Block N. Behav Brain Sci 18, 227–287 (1995).
24. Zeki S. Sci Am Spec Iss. 43–50 (Sept 1992)
25. Corbetta M, Miezin FM, Dobmeyer S et al. J Neurosci 11, 2383–2402
(1991).

Received 1 July 1998;
accepted 7 July 1998

Vol 9 No 13 14 September 1998

2925