Vision Research 51 (2011) 908–924

Contents lists available at ScienceDirect

Vision Research
journal homepage: www.elsevier.com/locate/visres

Grasping the non-conscious: Preserved grip scaling to unseen objects for

immediate but not delayed grasping following a unilateral lesion to
primary visual cortex
Robert L. Whitwell a, Christopher L. Striemer a, David A. Nicolle b, Melvyn A. Goodale a,⇑
a
Department of Psychology, Centre for Brain and Mind, University of Western Ontario, London, Ontario, Canada
b
Department of Ophthalmology, London Health Sciences Centre, University Campus, London, Ontario, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Patients with damage to primary visual cortex can sometimes direct actions towards ‘unseen’ targets
Received 28 June 2010 located in areas of the visual field that are deemed ‘blind’ on the basis of static perimetry tests. Here,
Received in revised form 20 January 2011 we show that a patient with a complete right homonymous hemianopia after a V1 lesion remains
Available online 13 February 2011
sensitive to the width of objects presented in her blind field but only when reaching out to grasp them
in ‘real-time’. A subsequent fMRI experiment revealed spared extra-geniculostriate pathways, which may
Keywords: mediate her preserved abilities. Taken together, the results support the view that visually guided move-
Dorsal stream
ments can be mediated by pathways that do not support visual consciousness.
Grasping
V1
Ó 2011 Elsevier Ltd. All rights reserved.
Blindsight
Perception-action

1. Introduction actor at the precise moment the action is performed. Thus, Goodale
and Milner argue, because the dorsal stream computes visuomotor
The deceptively simple act of reaching out and grasping an ob- transformations de novo each time an action is performed, it has no
ject is an almost perfect example of what the brain has evolved to need for long-term memory. In fact, storing particular visuomotor
do: namely, to control our movements in the world. Vision plays coordinates would be counterproductive since the disposition of
an indispensable role in this skilled behavior, and over the last the goal object with respect to the effector to be used can change
three decades, considerable effort has been spent trying to under- dramatically in just a few seconds. In stark contrast, visual process-
stand the neural substrates of visually-guided grasping. In 1992, ing in the ventral-stream interacts with long-term memory stores
Goodale and Milner proposed that the ventral visual processing of early visual experiences to enable the recognition of objects over
‘stream’ that arises from early visual areas and projects to infero- long periods of time. This means that actions that are performed on
temporal cortex mediates the recognition and identification of remembered objects that are no longer visible must invoke stored
goal objects for action, while the dorsal visual processing stream visual information originally processed by the ventral stream (for a
that arises from early visual areas and projects to the posterior discussion of these issues, see Milner & Goodale (2006)).
parietal cortex (PPC) transforms the particular information about One principal line of evidence for Goodale and Milner’s (1992)
the goal object into the required motor coordinates for action. proposal comes from work with patients with visual form agnosia
According to their account, the division of labour between the and optic ataxia. They have argued that the different patterns of
ventral and the dorsal pathways reflects the different require- spared visual abilities and deficits in these two patient groups
ments of vision-for-perception and vision-for-action respectively. constitute a functional and anatomical double dissociation between
Visual recognition depends on object constancy, the ability to vision-for-perception and vision-for-action. Patient DF, for exam-
identify the same object from different viewpoints, whereas the ple, who has visual form agnosia as a result of damage to the lateral
visual control of action requires that the visuomotor transforma- occipital complex (LOC), a ventral-stream structure (James,
tions reflect the disposition of the goal object with respect to the Culham, Humphrey, Milner, & Goodale, 2003), is still able to reach
out and grasp objects accurately despite the fact that she is unable
to report either manually or verbally their size, shape, or orienta-
⇑ Corresponding author. Address: Canada Research Chair in Visual Neuroscience,
Centre for Brain and Mind, Natural Science Centre, University of Western Ontario,
tion. Thus, DF is unable to use her finger and thumb to reliably indi-
London, Ontario, Canada N6A 5B7. Fax: +1 519 661 3961. cate the width of a target object when asked to give an estimate of
E-mail address: [email protected] (M.A. Goodale). its size, but when she reaches out to grasp that same object her grip

0042-6989/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.visres.2011.02.005
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 909

aperture (the distance between her finger and thumb) smoothly with hemianopia in his left visual field could accurately locate tar-
scales in-flight to the object’s width (Goodale, Milner, Jakobson, & gets presented in his blind field by pointing to them. Weiskrantz
Carey, 1991; see also patient JS: Karnath, Ruter, Mandler, & later showed that DB could localize targets accurately when they
Himmelbach, 2009). The authors argued that her preserved ability were presented in the regions of his blind field outside of his optic
to process visual information about object size, shape, and orienta- disc or ‘blind spot’ (Weiskrantz, 1987). This suggests that light
tion for the purposes of target-directed actions is mediated by a scatter and poor fixation could not account for DB’s accuracy, since
largely functioning and intact dorsal stream. Importantly, if DF is either of these factors would be equally applicable to a stimulus
asked to reach out and grasp an object that she saw only moments presented in his blind spot. Just 1 year prior to Weiskrantz’s find-
before, her performance completely deteriorates and she no longer ings, Poppel, Held, and Frost (1973) showed that four patients with
shows any evidence of grip scaling (Goodale, Jakobson, & Keillor, quadrantanopia (as determined with static perimetry) could scale
1994). This impairment in memory-driven grasping is thought to the amplitude of their saccades to a spot of light flashed briefly
reflect the fact that, because of her ventral-stream damage, she (100 ms) on any one of the three closest target positions that fell
was unable to perceive the dimensions of the object and store that along an oblique meridian within the scotoma. A number of addi-
information in memory for later explicit use. tional studies have shown that some (not all) patients retain resid-
Optic ataxia is a visuomotor disorder that impairs target- ual processing in the blind regions of their visual field and can
directed actions, such as reaching and grasping. It typically arises reliably process stimuli presented in their affected field during
following damage to the dorsal stream, particularly in the superior speeded detection tasks (Corbetta, Marzi, Tassinari, & Aglioti,
parietal lobule (SPL), intra-parietal sulcus (IPS), and parietal– 1990; Marzi, Tassinari, Aglioti, & Lutzemberger, 1986) and forced
occipital cortex (POC) (Karnath & Perenin, 2005; Perenin & Vighetto, response tasks like discrimination (e.g., Weiskrantz, , 1986, 1987)
1988). Nevertheless, because the ventral stream often remains and target localization with either manual pointing (e.g., Corbetta
relatively intact in these patients, they perform reasonably well et al., 1990; Danckert et al., 2003; Perenin & Jeannerod, 1975,
on perceptual tasks that require object localization, identification, 1978), or eye movements (e.g., Barbur, Forsyth, & Findlay, 1988;
or discrimination (Goodale, Meenan et al., 1994; Jeannerod, 1988; Poppel et al., 1973).
Jeannerod, Decety, & Michel, 1994; Jakobson, Archibald, Carey, & Patients with hemianopia show a range of spared visuomotor
Goodale, 1991; Perenin & Vighetto, 1988). Thus, these patients abilities that are not restricted to saccades and target localization.
can accurately estimate the size of target objects (Jeannerod et al., In particular, the ability to act on ‘unseen’ objects has been termed
1994) and yet show profound deficits in grip formation and ‘action blindsight’ (Rossetti & Pisella, 2002; see also Danckert &
coordination when asked to grasp them (Jakobson et al., 1991; Rossetti, 2005). Some patients, for example, can reliably calibrate
Jeannerod et al., 1994; Milner et al., 2001; Perenin & Vighetto, their in-flight grip aperture to the size of novel objects located in
1988). Remarkably, if the task is modified to require the patient to their blind field when reaching out to pick them up (Jackson,
grasp an unseen object viewed 2 s before the response is cued, their 1999; Marcel, 1998; Perenin & Rossetti, 1996) (for recent reviews
performance improves (Milner et al., 2001; Rossetti et al., 2005). As see Cowey, 2010; Danckert & Rossetti, 2005). Perenin and Rossetti
Milner et al. (2001) noted, these results are consistent with the (1996) showed that patient PJG with hemianopia could scale his
observation described earlier that DF’s performance on a similar grip aperture to the width of the objects presented entirely within
task was severely impaired (Goodale, Meenan et al., 1994a). Taken his blind field yet he could neither ‘match’ the object’s size using his
together, these results suggest that skilled target-directed actions thumb and forefinger nor verbally discriminate the objects accord-
depend on intact dorsal-stream structures in which the visual in- ing to their size. Marcel (1998) tested two patients, one of whom
puts are processed at the moment an action is programmed, and was GY, both of whom scaled their grip, categorically at least, to
that actions that follow a delay must be programmed on the basis the diameter of large and small variants of a cylinder and circle.
of remembered visual input which ultimately depends on an intact In contrast to Marcel’s findings, Jackson (1999) later found that
and functioning ventral stream. patient GY could not scale his grip aperture to the size of objects
Although patient DF is able to scale her grip aperture to objects presented entirely within his blind field. Nevertheless, GY did show
that she cannot recognize, it is unclear what remaining visual path- grip scaling to variations in the size of objects when they extended
ways are responsible for this remarkable ability. Retinal inputs to into his sighted field – even though the part of the object that was
primary visual cortex (V1) via the geniculostriate pathway are visible did not change in size from trial to trial. Interestingly,
known to innervate structures in the dorsal stream. It may be the however, with the same display, GY failed to estimate the magni-
case that this pathway is critical for DF’s preserved visuomotor tude of the (unseen) relevant dimension of the objects accurately
abilities, particularly since DF’s V1 appears to be intact (James on a verbal scale of one to six. In addition, recent studies have also
et al., 2003; Milner et al., 1991). In fact, because not all of her ven- shown that patients with blindsight can avoid unseen obstacles
tral stream is damaged, DF has some spared perceptual abilities. (de Gelder et al., 2008; Striemer, Chapman, & Goodale, 2009). In
She can, for example, give explicit (and often detailed) reports of short, several studies have demonstrated that patients with action
the surface properties of objects signaled by their color, visual blindsight can carry out rather complex actions in the absence of
texture, and specular highlights (Humphrey, Goodale, Jakobson, & any conscious vision. Thus, examining patients with action blind-
Servos, 1994). It is important to emphasize once more, however, sight affords the opportunity to investigate what aspects of dorsal
that she cannot give such reports of their shape and geometry. stream function remain in the absence of inputs from V1.
Although inputs from V1 may well be responsible for the pre- The existence of action blindsight suggests that intact visual
served visuomotor abilities in patients with visual form agnosia, pathways outside of the prominent geniculostriate pathway are
there have been reports of similarly preserved visuomotor abilities capable of mediating visually guided actions. One such pathway
in patients with visual field loss who do not have access to any con- from the retina courses through the superior colliculus (SC) and
scious vision following damage to V1. Specifically, Weiskrantz and pulvinar nucleus in the thalamus (i.e., the retino-tectal-pulvinar
colleagues (Sanders, Warrington, Marshall, & Weiskrantz, 1974; pathway) to the motion-sensitive area MT and from there onto
Weiskrantz, Warrington, Sanders, & Marshall, 1974) coined the areas in the dorsal stream. This pathway is known to convey pri-
term ‘‘blindsight’’ to describe patients with visual field loss who marily magnocellular information which has relatively low spatial
deny seeing visual stimuli presented in their affected field but resolution but is extremely sensitive to changes in contrast. Re-
whose voluntary performance can be reliably influenced by visual cently, a direct projection from the eye to the pulvinar, and then
stimulation. This group was the first to show that a patient, DB, to MT has been identified in the marmoset (Warner, Goldshmit,
910 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

& Bourne, 2010). In addition, there are projections from koniocellu- facing her across the table in her blind field and a second experi-
lar cells in the interlaminar regions of the dorsal lateral geniculate menter monitored her eyes. SJ’s description of the phenomenon
nucleus (dLGN) that bypass V1 and project directly to area MT and was not that she could see the hand but that she detected a pertur-
onward to other regions of the dorsal stream (Sincich, Park, bation of a ‘field’ of haze located in her blind field whenever the
Wohlgemuth, & Horton, 2004; for review see Vakalopoulos, hand moved. Her subjective report suggests that she possesses a
2005). The koniocellular cells in the dLGN tend to convey informa- form of Riddoch phenomenon, which refers to a patient’s ability
tion that is similar to that conveyed by the cells in neighboring par- to sense when a visual stimulus is moving within their scotoma
vocellular or magnocellular layers (Hendry & Reid, 2000). A recent but not when the stimulus is stationary (Riddoch, 1917).
study in monkeys with V1 lesions suggests that the projections In addition to patient SJ, we also tested 12 right-handed control
from the interlaminar regions of the dLGN are capable of mediating participants (M = 29 years, SD = 7.4 years) with normal or cor-
some residual visual behavior (including saccades). Specifically, rected-to-normal vision and no prior neurological history. Note
these residual abilities, and the associated BOLD activation in area that the control participants performed the grasping and manual
MT and extrastriate cortex disappeared when the dLGN was matching tasks under the same conditions as patient SJ but with
reversibly inactivated (Schmid et al., 2010). Thus, in the absence fewer trials and in only one test session, rather than two. We
of V1 input, there appears to be a number of extra-geniculostriate did not test the control participants on the preliminary target-
pathways that could continue to provide information about tar- detection and target-localization tasks.
geted objects to the dorsal stream upon which the transformations
required for calibrating grip aperture could be based. 2.2. Experimental setup and design
In the current set of experiments, we replicate and extend
Perenin and Rossetti’s (1996) original demonstration that a patient 2.2.1. Target-localization
with a hemianopia (patient PJG) was capable of scaling a grasping Patient SJ sat 40 cm away from a touch screen monitor (3200 LCD,
movement directed towards an ‘unseen’ object presented in the Mass Multimedia Inc; refresh rate 60 Hz) and fixated on a cross lo-
blind field. We tested a new patient SJ, who has a complete right cated at the center of the touch screen. The touch screen recorded
homonymous hemianopia as a consequence of a stroke in her left the point on the screen where SJ’s finger first landed (end-point)
occipital cortex. Not surprisingly, patient SJ, like patient PJG, was for each trial. The target was a black circle (2.2 cm diameter),
unable to estimate the widths of objects presented in her blind which subtended approximately 3° of visual angle presented for
field. Nevertheless, she did show evidence of blindsight: she could 100 ms at one of four horizontal eccentricities from fixation (10°,
localize ‘unseen’ stimuli presented in her blind field, and could 17°, 24°, or 30°) on a uniform grey background. Each target posi-
scale her in-flight grip to the widths of novel objects located in tion was sampled 20 times, and the trial order was pseudo-
her blind field when reaching out to pick them up. Importantly, randomized for target position. The time from trial onset to target
however, she could do this even when vision was not available onset was randomly varied from 1 to 3 s (in 500 ms increments) to
during the execution of the grasp. But for grip scaling to occur with minimize the predictability of target onset and the incidence antic-
objects in her blind field, the movements had to be programmed ipatory responses. SJ’s eye movements were monitored ‘online’
and initiated while vision was available (i.e., in real-time). When (and recorded for additional evaluation ‘offline’) by an experi-
a brief delay was introduced between last viewing the object and menter using a video camera zoomed-in on one of her eyes, while
programming the movement, SJ showed no evidence of grip scal- a second experimenter advanced the trials. SJ was excellent at
ing. Finally, we report the results of a functional magnetic reso- fixating and very few trials had to be discarded because of eye
nance imaging (fMRI) study with SJ that provides tentative movements.
support for the idea that her spared abilities are mediated by path-
ways to extrastriate cortex that bypass V1. Taken together, these 2.2.2. Explicit target-detection
data contribute to a growing body of evidence that visuomotor net- We also tested patient SJ on the redundant target paradigm.
works in the dorsal stream can operate independently of input This target-detection task requires the participant to respond with
from V1, but only in real-time on a moment-to-moment basis. a button press as soon as a target appears on the screen. Typically,
participants are faster to respond to two targets presented simul-
taneously than only one (Marzi et al., 1986). This phenomenon,
2. Session 1: methods known as redundancy gain, has been linked with the retino-
tecto-pulvinar pathway which is thought to mediate blindsight ob-
2.1. Participants served in some patients with hemianopia or hemispherectomy
(Corbetta et al., 1990; Leh, Johansen-Berg, & Ptito, 2006; Leh,
Patient SJ is a 37 year-old right-handed female who has a com- Mullen, & Ptito, 2006; Leh, Ptito, Schonwiesner, Chakravarty, &
plete right homonymous hemianopia with no macular or temporal Mullen, 2009; Marzi, Mancini, Metitieri, & Savazzi, 2009) and has
crescent sparing following a left posterior cerebral artery stroke revealed residual visual processing in some, though not all, pa-
which occurred 7 years prior to the time of testing. Her lesion, as tients (Leh, Ptito et al., 2006; Marzi et al., 1986). Patient SJ sat
revealed by high-resolution MRI, is restricted to the left occipital 40 cm away from a 3200 LCD touchscreen with her head resting in
lobe and optic radiations with some extension into the parahippo- a chin-rest. The target was a black circle (2.2 cm diameter), which
campal gyrus (Fig. 1a) with a small remaining ‘‘tag’’ of cortex subtended approximately 3° of visual angle presented for 100 ms
visible at the occipital pole. Visual fields were assessed by a neu- on a uniformly grey background. On a given trial, the target could
ro-ophthalmologist (D. Nicolle) using a Goldman Perimeter with appear 10° of visual angle to the left or right of a fixation cross
III-4 sized targets (Fig. 1b). Informed consent was obtained from located at the center of the screen, on both sides, or not at all. Thus,
SJ and all experimental procedures were approved by the Univer- there were four conditions total. Two ‘single target’ conditions (left
sity of Western Ontario Health Sciences Research Ethics Board or right hemi-field), one ‘double target’ condition (two targets
and in full accordance with the Declaration of Helsinki. presented simultaneously, one in each hemi-field), and a blank
Although patient SJ is completely blind in her right visual field condition (‘catch’ trials) on which no target was presented. The
on static perimetry, she reports subjective sensations of motion. task was administered in four separate blocks of 40 trials (four
Informal testing found that SJ was extremely accurate at indicating conditions, 10 trials per condition). Trials were presented in a
whether the experimenter’s hand waved or not while he stood pseudo-random sequence and the time between the onset of the
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 911

Fig. 1. Panel A depicts a saggital and a series of axial slices from a T1-weighted high resolution (192 slice 1 mm iso-voxel) anatomical MRI scan of patient SJ’s left occipital
lesion recently obtained from a 3 Tesla Siemens Tim-Trio MRI scanner at the Robarts Research Institute (London, Ontario, Canada). The lesion is restricted to the left occipital
lobe and optic radiations with a minor extension into the parahippocampal gyrus. The green arrow indicates the spared ‘‘tag’’ of cortex located at the occipital pole. Panel B
depicts the results of the visual field testing with SJ. SJ has a complete right hemianopia with no macular or temporal crescent sparing. Data are presented separately for the
left and right eyes.

trial and the onset of the target(s) was randomized to last any- sively larger object’s width incremented by 1.5 cm to a maximum
where from 1 to 3 s (in 500 ms increments). A tone was not used of 6.5 cm. The objects were painted flat black to reduce intra- and
for this task, since SJ was required to respond only when she saw extra-ocular dispersion and were padded on the bottom to mini-
a target. mize the sound of contact with the surface of the table on place-
ment. The table top was covered with white Bristol board to
2.2.3. Grasping provide a high contrast between the table surface and the black
Targets stimuli were four wooden objects each of which was objects. Patient SJ was seated in front of the table with her head
1.5 cm in height, 5 cm long, and differed from the others only in positioned in a chin-rest. The chin-rest was attached to the table
width. The smallest object width was 2.0 cm with each succes- edge directly in front of the start button, which was 15 cm from
912 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

the table edge. SJ aligned her midline with the start button and the 2.3.2. Explicit target-detection
chin-rest. The fixation target was centered 30 cm directly ahead of Patient SJ fixated centrally on a cross and pressed a button as
the start button and approximately 50 cm from the participant’s soon as she saw a target appear anywhere on the screen. SJ insisted
eyes. The inner edge of the objects was placed 8 cm either to the she did not see any targets in her blind (right) visual field with the
left or to the right of fixation, which corresponded to 9° of visual exception of two discarded single-target trials on which she failed
angle to the left or to the right of fixation. The reach-distance from to maintain fixation.
the start button to the center of the edge of the objects facing SJ
was approximately 32 cm. 2.3.3. Grasping
Patient SJ’s vision was controlled during the grasping session Patient SJ was asked to maintain fixation throughout the test
with a pair of PLATO goggles (Translucent Technologies, Toronto, session. The goggles restricted her vision between trials, which al-
Ontario) worn by the participant. The goggles are equipped with li- lowed the experimenter to place the object for the next trial out of
quid crystal lenses that can switch between opaque and transpar- sight from SJ. This also meant that SJ could not see the fixation tar-
ent states in less than 6 ms. As for the target-localization task, SJ’s get between trials and as a result she was asked to maintain her
eye movements were monitored ‘online’ continuously by one gaze as best she could by visualizing the fixation target. Before
experimenter and recorded for additional evaluation ‘offline’ using testing commenced, the experimenters explained the nature of
a video camera zoomed-in on one of her eyes, while a second video the tasks and administered several practice trials to familiarize
camera recorded her movements. her with the events and the pace from trial to trial. Throughout
Patient SJ’s performance on visually-guided grasping was these practice trials, SJ reported no difficulty maintaining her gaze
tested in blocks of 49 trials (33 trials for the control participants) on the unseen fixation target between trials. Nevertheless, as men-
administered first in the left visual field and then in the right tioned previously, one experimenter monitored her eye move-
visual field. Specifically, one block of trials was administered to ments ‘online’ on the display of a video camera that was
test SJ’s performance with her left hand for objects located in zoomed-in on one of her eyes. A second experimenter concerned
her sighted (left) visual field, and two subsequent blocks of trials himself with placing the objects and observing SJ’s performance.
were administered to test SJ’s performance with her right hand for Trials during which SJ appeared to move her eyes, fumbled the ob-
objects located in her blind (right) visual field. This was done ject, or initiated her movement too early were noted and repeated
because of the expected difference in effect size between the at the end of the block.
two fields. Within each block of 49 trials object size was pseu- Before the onset of a trial and at the end of each trial, patient SJ
do-randomized such that each object appeared at least 12 times depressed the start button with the tips of her thumb and forefin-
and had an equivalent probability of being preceded or followed ger pinched together while keeping her other hand resting on her
by any of the objects. In addition, no object was permitted to lap below the surface of the table. The experimenter initiated each
occur more than twice consecutively. Thus, the trial order pre- trial which began with a 1 s delay followed by a trigger that
vented the experience of grasping any particular object on preced- switched the lenses from their default opaque state to their trans-
ing trials from systematically influencing SJ’s performance on any parent one, which permitted SJ a view of the workspace with the
subset of the objects. This precaution seemed particularly relevant object located in either her sighted (left) or blind (right) visual
following recent studies that have shown that the trial history of field. SJ was instructed to initiate her movement when she could
visual feedback (Whitwell & Goodale, 2009) and indeed that of ob- see the workspace (i.e., when the goggles opened). Patient SJ was
ject size (Dixon & Glover, 2009) can systematically influence grip asked to reach out and pick the object up off the surface of the ta-
aperture. The positions of three infrared emitting diodes (IREDs) ble using only her forefinger and thumb such that contact with the
were recorded in three-dimensional space by Optotrak 3020 (NDI, object meant that her grip opposition axis spanned the width of
Waterloo, Ontario) at 200 Hz (one frame every 5 ms) for 3 s follow- the object. The lenses remained transparent for 1 s following the
ing the response cue. One IRED was attached to the tip of the thumb, release of the start button which permitted SJ a full view of the
a second was attached to the tip of the forefinger, and a third at- workspace that included the time taken to obtain the object (i.e.,
tached to her wrist. The IREDs were positioned such that the pads closed-loop visual feedback). Note that there was no need to em-
of the forefinger and thumb were unobstructed. ploy a tone ‘‘go’’ cue for this task, since SJ could register a change
in the visual input in her sighted (left) visual field as soon as the
2.3. Procedure goggles opened. This was not the case for target-detection and
target-localization tasks, but only the target-localization task
2.3.1. Target-localization required SJ to respond on each trial. Finally, during testing the
First, the touchscreen was calibrated to ensure veridical end- experimenters periodically asked SJ if she could see any of the ob-
point measurements. To calibrate the participant’s eye movements jects in her blind field to which she answered that she could not.
on the camera monitor, one of the experimenters asked the partic-
ipant to fixate on the central cross and then noted the position of 2.4. Data processing and statistical analysis
SJ’s iris on the camera display. The experimenter then asked SJ to
fixate on the target closest to the fixation cross and again noted We used ordinary least squares (OLS) linear regression to model
the position of SJ’s iris on the camera display. Eye-movements of the raw dependent measures as linear functions of the indepen-
2° could be readily detected using this method, which was used dent variables for each field and kinematic task separately using
in all the experiments, including those with the control partici- SPSS version 17.0 (Chicago, Illinois). In addition to the Pearson
pants. For the target-localization task, each trial began with the product-moment correlation (correlation coefficient), r, we report
target stimulus presented in one of four pre-determined locations the unstandardized regression coefficient, b, for two reasons. First,
for 100 ms coinciding temporally with a 1000 Hz tone. The tone the dependent and independent variables share the same units
cued SJ to both look and point to the target’s location. SJ used within each behavioral task administered. In general, b represents
her forefinger to touch the screen where the target had been pre- the change in the dependent measure per incremental increase in
sented. Before and after each pointing movement, SJ kept the tip the independent variable while holding all other modeled variables
of her forefinger resting on a start position. Her performance was constant. In other words, b reflects how sensitive a dependent
tested in her sighted (left) and then in her blind (right) visual field. measure is to changes in the independent variable. In contrast, r re-
SJ was excellent at maintaining fixation. flects the degree to which the z-transformed observations deviate
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 913

from any linear relationship between the two z-transformed vari- linear function of the distance from the fixation cross to the targets
ables. The z-transformed scores, from which the r statistic is de- (also in visual degree angle). The regression coefficient, b, reported
rived, rely on sample standard deviations of the variables, which for target-localization reflects the rate of change in horizontal dis-
will vary from sample to sample. Thus, r removes information tance between the endpoint and the fixation cross (in degrees of vi-
about the original units of the variables and will begin to approach sual angle) as horizontal distance between the target position and
the value of 1 (or 1) if changes in one variable are consistently the fixation cross increases (also in degrees of visual angle).
and reliably matched to a change in the other variable (i.e., the
relationship approaches linearity in z-transformed space). For
2.4.2. Explicit target-detection
these reasons, b has been recommended over r (which for bivariate
For the target-detection task, trials were removed from the
models is equivalent to the standardized regression coefficient,
analysis if patient SJ failed to respond or if her reaction time (RT)
beta) for comparisons between or among samples and for general-
was more than 2 SDs larger then her mean for that condition. As
ization to populations (Cohen, Cohen, West, & Aiken, 2003; Keith,
a result, two trials were discarded. The analysis for the target-
2006). Note that for bivariate cases within a data set, the t-tests
detection task compared patient SJ’s mean RT for single-target tri-
of b and r against zero are equivalent and are assessed with the
als which appeared in her left (sighted) visual field with the mean
same degrees of freedom.
RT for double target trials (targets appear simultaneously in both
For a given participant, either statistic could be computed from
visual fields).
the raw scores or the means of the observations associated with
each level of the independent variable. The former approach would
probably decrease the magnitude of r, but the degrees of freedom 2.4.3. Grasping
used in the t-test of b and r against zero would be computed from In-house software permitted an automated selection of move-
the trial total. In contrast, the latter approach would likely increase ment onset, movement offset, and the principal dependent mea-
the magnitude of r, but the degrees of freedom would be computed sure, peak grip aperture (PGA) for all data from the grasp session.
from the condition total (e.g., the number of objects used). We re- For each sample point, grip aperture was computed as the vector
port both variants of r, with the degrees of freedom differentiating distance between the IREDs located at the tip of the thumb and
the two. The two approaches would not differentially influence b forefinger. The onset of the movement was defined as the first of
provided there were an equivalent number of observations per le- 20 consecutive frames during which the velocity of the thumb
vel of the independent variable. IRED exceeded 30 mm/s. The offset of the movement was defined
To compare patient SJ’s grip scaling performance with the mean as the first of 20 consecutive frames during which the grip aperture
performance of the control participants, we could use b or r. But as velocity fell within ±30 mm/s. The movement onset and movement
discussed above, the use of these statistics has distinct implications. offset frames were used to define a search window within which
We used r to compare the patient’s performance with the mean per- the PGA was selected. Each grip aperture profile was inspected
formance of controls (Crawford, Garthwaite, Howell, & Venneri, visually to ensure that there were no gross errors made in the
2003) rather than b, since a comparison of a single case to a control selection of PGA. For trials where an erroneous PGA was selected,
sample using b requires a more stringent assumption about the dis- the number of consecutive frames criterion for movement offset
tribution from which the slopes were sampled (see Crawford & was adjusted by increments of five.
Garthwaite, 2004). [Although not reported, a comparison of the Patient SJ committed no errors with her left hand for objects lo-
b-statistics yielded identical conclusions.] The r-statistics we used cated in her left (sighted) visual field (control participants M = 3%).
to compare patient SJ’s performance with the mean performance She did, however, commit four errors (4%) with her right hand for
of the controls were computed for each participant separately from objects located in her right (blind) visual field (control participants
the means of the measurements obtained from grasping or M = 3%). These trials were discarded from the analysis and their re-
estimating the objects. peats at the end of the block substituted into the analysis.
Briefly, Crawford, Garthwaite, Howell et al. (2003) recommend Ordinary least squares (OLS) linear regression modeled PGA on
a Fisher-transformation on the r derived from each participant. the width of the object on the current trial. The regression coeffi-
The Crawford and Howell (1998) modified t-test is then conducted cient, b, for this analysis represents the rate of change in the depen-
on the transformed data (Crawford, Garthwaite, Howell et al., dent measure (in mm) per incremental increase in the width of the
2003). This modified t-test treats the control participants as a sam- object (also in mm). Since there were two blocks of trials adminis-
ple, rather than a fixed population. This method allows us to assess tered to test performance in the right (blind) visual field, this
whether SJ’s performance falls within or outside of the normally- model was tested for the full data set and then once for each block
sighted control population without inflating the per-contrast Type separately.
I error rate, a, as traditional techniques involving z- or t-statistics An additional model of Session 1 PGA was tested that included
do (Crawford and Garthwaite, 2005b). We used a per-contrast a the width of the object grasped on the current trial and that of the
criterion of 0.05. Given the grip-scaling abilities of blindsight object grasped on the immediately preceding trial. The additional
patients in the literature in their sighted and blind fields (e.g., model, therefore, assesses whether a significant linear relationship
Jackson, 1999; Perenin & Rossetti, 1996), we adopted a two-tailed between patient SJ’s PGA (1) and the width of the current object (2)
criterion for the comparisons of SJ’s performance in her sighted remains once the contribution of the width of the object grasped
field with that of the normally-sighted controls and a one-tailed on the previous trial (3) has been partialled out of both PGA and
criterion for similar comparisons in her blind field given that one the width of the current object. We employed this model as an
might expect to observe a deficit. additional safeguard against the possibility that SJ may intention-
ally or unintentionally use haptic information about the object
2.4.1. Target-localization grasped on a given trial to guide her performance on the subse-
The principal dependent measure for the target-localization quent trial. Only valid trials that were preceded by a valid trial
task was computed from the recorded forefinger endpoints, which were included in this analysis. Accordingly, the first trial of each
were converted to a horizontal distance from the fixation cross in block and each of the four error trials and their subsequent trials
visual degree angle. No trials were discarded from the target- were discarded (total of ten). For this second model, we report
localization task. An OLS linear regression was employed to model the b-statistics and partial correlations (e.g., r1.23). Given the nature
the endpoints (in visual degree angle) for each field separately as a of the trial order it was not surprising to find that the collinearity
914 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

diagnostic indicated perfect independence between the current As expected, PGA increased linearly as a function of the width of
and preceding object widths (Tolerance = 1.0). the object located in her sighted (left) field which she grasped with
her left hand (r(47) = 0.60, b = 0.25, p < 0.001; r(2) = 0.99). SJ scaled
her grip to the width of the object in her sighted field as reliably as
3. Session 1: results the control participants, t(11) = 0.03, p = 0.98. (Fig. 3 panel C). Crit-
ically, PGA also increased linearly as a function of the width of the
3.1. Target-localization object located in her blind (right) field (r(96) = 0.35, b = 0.17,
p < 0.001; r(2) = 0.89), although SJ’s grip scaling was less reliable
A linear regression of patient SJ’s endpoints on the target loca- than that of the control participants, t(11) = 1.98, p < 0.05
tion indicated that she was able to localize targets in both her (Fig. 3 panel D). Subsequent analyses of SJ’s performance in her
sighted and blind fields (see Fig. 2). blind (right) field indicated a significant linear relationship be-
Patient SJ’s endpoint distance increased with the target posi- tween PGA and object width for each block separately (first block,
tions located in her sighted (left) visual field (r(29) = 0.96, r(47) = 0.40, b = 0.21, p < 0.005; second block, r(47) = 0.30, b = 0.13,
b = 0.76, p < .001; r(2) = 0.99) and again when the targets were lo- p < 0.05).
cated in her blind (right) visual field (r(78) = 0.52, b = 0.31, p < .001; The analysis of PGA (1) as a function of the width of the object
r(2) = 0.86). Qualitatively, SJ performed this task no differently grasped on the current trial (2) and the immediately preceding trial
than sighted individuals would. That is to say SJ did not appear (3) indicated that patient SJ’s PGA increased with the width of the
to adopt an unusual or deliberate strategy to complete this task. current (r(85)12.3 = 0.34, p < 0.001) and previous objects
However, since saccade latencies were not recorded, we cannot (r(85)13.2 = 0.41, p < .001; r(86)23 = 0.01, ns) located in her blind
make a formal quantitative assessment. (right) visual field. This suggests that both the visual information
During the testing, patient SJ insisted that she did not see any of from the current object and experience with the object on the pre-
the targets presented in her blind field. On a few of the trials in vious trial made unique and significant contributions to the pro-
which the target was presented 10° from the fixation cross, she gramming of grip aperture for the object on the current trial. An
remarked that she ‘‘felt’’ like she ‘‘saw a shadow’’. However, when influence of the width of the previous object was not observed
pressed to describe the phenomenon she replied that she could when SJ grasped objects located in her sighted (left) visual field.
derive no shape or form from the impressions. Her description Accordingly, SJ’s PGA increased with the width of the current
suggests a type II blindsight, which reflects an awareness of a (r(46)12.3 = 0.63, p < .001) but not the width of the previous objects
visual stimulus without a concrete description (e.g., a ‘gut’ feeling (r(46)13.2 = 0.28, ns; r(47)23 = 0.10, ns).
see Cowey, 2010). Interestingly, similar, if not identical, subjective
reports like these are reported frequently in the blindsight litera-
ture (see Zeki & Ffytche, 1998). 4. Session 1: discussion

3.2. Target-detection The endpoints of patient SJ’s pointing movements were clearly
guided by the location of the targets in both her sighted (left)
Patient SJ was not reliably faster at responding to two simulta- and blind (right) visual fields. When asked to reach out to pick
neously presented targets (M = 349 ms, SD = 54 ms) than one tar- up novel objects located in her blind (right) visual field with her
get (M = 357, SD = 54 ms) (t(69) = 0.66, p = 0.51). She did not dominant (right) hand, SJ adjusted her grip aperture with the
respond on any of the blank ‘catch’ trials. width of the target object. Interestingly, the results of the second
model that included the width of the object grasped on the previ-
ous trial indicated that SJ (either intentionally or unintentionally)
3.3. Grasping used the experience gleaned from the previous trial to inform
her performance on the following trial when reaching out to pick
Fig. 3 depicts patient SJ’s PGA as a function of the width of the up objects located in her right (blind) but not her left (sighted) vi-
object located in her sighted (left) visual field and grasped with her sual field. Critically, however, visual information about the width
left hand (panel A) and with her right hand in her blind (right) of the current object accounted for a nearly identical amount of
visual field (panel B). variance in her performance on the current trial. Although the

Fig. 2. Session 1 horizontal endpoint distance from the fixation cross (end-point) in visual degree angle from the fixation cross as a function of the target location (also in
horizontal visual degree angle from the fixation cross) for patient SJ when pointing to targets in her left (sighted) visual field (LVF) (panel A) and in her right (blind) visual field
(RVF) (panel B) with her right (dominant) hand while vision remained available throughout the movement (closed-loop feedback).
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 915

Fig. 3. Peak grip aperture (PGA) in Session 1 plotted as a function of object width for patient SJ when she grasped objects located in her sighted (left) visual field (LVF) with
her left (non-dominant) hand (panel A) and in her blind (right) visual field (RVF) with her right (dominant) hand (panel B) with retinal input available during the
programming and execution of the grasp (closed-loop visual feedback). Panels C and D each depict the Fisher-transformed correlation coefficient, rF, that describes the
relationship between PGA and object width for SJ and the normally-sighted control participants, who were tested under identical conditions (LVF: panel C; RVF: panel D).
Solid error bars denote the 95% confidence interval around the mean rF for control participants to compare SJ’s rF. Note that SJ showed sensitivity to the width of objects when
she grasped them in both her sighted and her blind field. The reliability of SJ’s grip scaling in her sighted-field (LVF) did not differ significantly from that of the control
participants (panel C). The reliability of SJ’s grip scaling in her blind field (RVF), however, was significantly less than that of the control participants (panel D).

slopes governing the relationship between object width and PGA found no evidence that SJ was visually conscious of any targets
appear to be similar across hands, only SJ’s performance with her in her blind field even though she showed clear evidence for action
left hand was within the normal range. Had SJ been tested with blindsight.
her dominant (right) hand for both fields, a larger difference would
likely have emerged between her sighted and her blind field. Nev-
5. Session 2: experiments
ertheless, the mean slope for grasps made into SJ’s blind (right)
field across the two blocks of trials (b = 0.17; first block, b = 0.21)
Approximately 4 months following the first session of experi-
was quite similar to the slopes we calculated from the published
ments, we tested patient SJ again for replicability of our initial find-
data on patient PJG (b = 0.22; Perenin & Rossetti, 1996) and patient
ings and to explore the limitations of her grip scaling in her blind
GY (b = 0.20; Jackson, 1999).
field with three additional tasks. Thus, in separate testing blocks
At the same time, patient SJ’s spared performance in these vis-
SJ grasped objects located in either her left or right visual field with
uomotor tasks did not appear to depend on her ability to ‘see’ the
(closed loop) and without (open loop) visual feedback and follow-
targets or, for the purposes of grip scaling, her ability to perceive
ing a 2 s delay without feedback. In addition, as a control task, we
their width, shape, or form; in fact, SJ never reported seeing a tar-
asked SJ to manually estimate the width of the same objects pre-
get when it was presented by itself in her blind field nor was she
sented within her sighted (left) and blind (right) visual fields. In
reliably faster at responding to a target in her good field when an
all cases, SJ used her right hand.
additional target was presented in her blind field during the tar-
get-detection task. It is important to note, however, that patient
SJ’s failure to show a target redundancy effect in our experiment 5.1. Methods
does not mean that she is completely incapable of detecting targets
in her blind field (using a button press). Had we used a forced- 5.1.1. Participants
choice variant of this task she may have very well exhibited In addition to patient SJ we tested 13 right-handed control par-
better-than-chance levels of performance. Furthermore, it is ticipants (M = 30 years, SD = 8 years) 12 of whom were tested on
important to note that there is between- and within-patient vari- the Session 1 grasping tasks. The control participants were tested
ability across the range of residual visual abilities that have been under identical conditions as those used with SJ, but with fewer tri-
tested in blindsight (e.g., Corbetta et al., 1990). Nevertheless, we als to keep the duration of the experiment under 2 h.
916 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

5.1.2. Experiment setup and design principal dependent measure for manual estimation was the man-
The setup required few changes from those used for Session 1 ual estimate aperture (MEA) which was defined as the first of 30
grasping. A new set of four black objects was used. The object consecutive frames following movement onset during which the
height remained the same as those used on Session 1 (1.5 cm), speed at which the thumb and finger closed or opened fell below
but the lengths were increased to 6 cm and the widths were re- 30 mm/s. This threshold reflected a point at which the grip aper-
duced by 0.5 cm each such that the shortest object width was ture remained highly stable and was thought to best represent
1.5 cm and the longest 6 cm. The Session 2 objects were instru- the participant’s estimate of the object’s width given the experi-
mented so that the time at which contact was made with the ob- menter’s instruction to preserve their estimate until vision was
jects could be recorded electronically. Additionally, rather than occluded.
keeping the location of the fixation target constant and varying The data were analyzed using the first model employed for Ses-
the object’s position as was done on Session 1 testing, the location sion 1 grasps, which described PGA as an OLS determined linear
of the fixation target varied while the location of the target object function of the width of the object grasped on the current trial.
remained constant. Thus, all grasps and manual estimates were For grasps executed with closed-loop visual feedback, patient SJ
executed with the right hand while the participant fixated to the committed seven errors in her sighted (left) visual field (11%)
left or to the right of the object. This ensured that the biomechan- and five in her blind (right) visual field (8%); for comparison, the
ical requirements for the action within each field were identical. average error rate for the control participants was 7% for each field.
The objects were located such that center of the edge of the object For grasps executed without visual feedback (open loop), SJ com-
facing the participant was 30 cm from the start button. Patient SJ mitted one error in her sighted (left) field (2%) and two errors in
was seated as on Session 1. The distance from her eyes to the front her blind (right) field (3%); for comparison, the average error rate
edge of the object was 50 cm. The fixation points were located for the control participants was 3% for each field. Finally, for de-
10 cm from the leftmost or rightmost edge of the object. The dis- layed grasps, SJ committed two errors in her sighted field (4%)
tance, therefore, between the fixation target and the closest edge and three errors in her blind field (5%); for comparison, the average
of the object subtended 11° of visual angle. error rate for the control participants was 7% for the left and 3% for
In addition to closed-loop visual feedback, we also asked patient the right visual fields. As on Session 1, all error trials were dis-
SJ to perform grasps without visual feedback (open loop). For this carded from the analysis and the missing values replaced with
condition, the trial events occurred as before, but vision was re- the data from the corresponding trials that were repeated at the
moved once the participant released the start button at the onset end of the block.
of the movement rather than remaining open for an additional sec- In addition to comparing patient SJ’s performance against the
ond. Grasping was also tested in a delay condition in which vision mean performance of the normally-sighted control participants,
was permitted for 1 s and the response cued 2 s thereafter by an we examined whether SJ’s grip-scaling deficits in delayed and
auditory tone. All grasping tasks were administered in separate manual matching tasks dissociated from her performance in the
blocks of 65 trials each for patient SJ (33 trials for the control par- immediate ‘real-time’ grasping task. To this end, we adopted
ticipants), while manual estimates were tested in a block of 33 tri- Crawford, Garthwaite, and Gray’s (2003) criterion for dissociation.
als. The trial order was constructed using the same criterion used We expected that SJ’s performance on real-time grasping in her
for Session 1, except that each object was sampled 16 times for blind field would differ from her performance on delayed grasping
each grasping task and 8 times for manual estimations. or manual estimation more than similar contrasts in the normally-
sighted controls. Additionally, SJ’s performance on at least one of
5.1.3. Procedure these tests should dissociate from the mean performance of the
For all tasks, the sighted (left) visual field was tested before the controls. This was formally tested using Crawford and Garthwaite’s
blind (right) visual field to further familiarize patient SJ with the (2005a) revised standardized difference test (see also Crawford &
requirements, and identical procedures were adopted for the 13 Garthwaite, 2005b). As was the case for the Session 1 results, the
control participants. The starting hand configuration used for Ses- per-contrast a criterion was 0.05, with a one-tailed criterion set
sion 1 grasps was used for grasps and manual estimates for Session for the comparisons between SJ and the controls for performance
2. For the immediate grasping and manual estimation tasks, the in the blind (right) visual field and a two-tailed criterion set for
participants were instructed to initiate their grasps when vision the comparisons between SJ and the controls for performance in
was permitted (i.e., when the goggles opened). For manual esti- the sighted (left) visual field. Finally, although a one-tailed test
mates, SJ used her forefinger and thumb to match the width of would be justified, we adopted the more conservative two-tailed
the target object without reaching towards it. Once satisfied with criterion to assess whether the reliability of SJ’s grip scaling to ob-
her estimate, she was asked to keep her fingers as stable as possi- jects located in her blind field in immediate grasping dissociated
ble until vision was occluded by the goggles. Throughout the man- from her performance in the same field for delayed grasping and
ual estimation task, the bottom of her hand remained rested on the manual estimation.
table which prevented her from inadvertently directing her hand
towards the object. Immediate grasping with closed-loop visual 5.2. Results
feedback was tested first, then manual estimations, followed by
immediate grasping in open loop (i.e., movements without visual 5.2.1. Immediate grasping with visual feedback (closed loop)
feedback), and then delayed grasping. As on Session 1, trials during Similar to the results of Session 1 testing, patient SJ’s PGA in-
which SJ appeared to move her eyes, fumbled the object, or initi- creased with the width of objects located in her sighted (left) field
ated her movement too early were noted and repeated at the (r(63) = 0.72, b = 0.39, p < 0.001; r(2) = 0.99). SJ’s grip scaling in her
end of the block. sighted field was as reliable as that observed in the normally-
sighted controls, t(12) = 0.06, p = 0.95. (see Fig. 4: panels A and C).
5.1.4. Data processing and statistical analysis Critically, SJ’s PGA increased with the width of objects located in
For all tasks, movement onset corresponded to the sample her blind (right) field (r(63) = 0.33, b = 0.14, p < 0.01; r(2) = 0.95).
frame at which the start button was released. For the grasping However, similar to the results of Session 1, SJ’s grip scaling to
tasks, movement offset corresponded to the sample frame at which the width of objects located in her blind field was less reliable than
the contact was made with either the front or back side of the tar- that of the control participants, t(12) = 2.30, p < 0.05 (see Fig. 4:
get object. Grip aperture and PGA were defined as on Session 1. The panels B and D).
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 917

Fig. 4. Peak grip aperture (PGA) in Session 2 plotted as a function of object width for patient SJ when she grasped objects located in her sighted (left) visual field (LVF) (panel
A) and in her blind (right) visual field (RVF) (panel B) with her right (dominant) hand with retinal input available during the programming and execution of the grasp (closed-
loop visual feedback). Panels C and D each depict the Fisher-transformed correlation coefficient, rF, that describes the relationship between PGA and object width for SJ and
the normally-sighted control participants, who were tested under identical conditions (LVF: panel C; RVF: panel D). Solid error bars denote the 95% confidence interval
around the mean rF for control participants to compare SJ’s rF. Similar to what happened in Session 1, SJ showed sensitivity to the width of objects when she grasped them in
both her sighted and her blind field. The reliability of SJ’s grip scaling in her sighted-field (LVF) did not differ significantly from that of the control participants (panel C). The
reliability of SJ’s grip scaling in her blind field (RVF), however, was significantly less than that of the control participants (panel D).

5.2.2. Immediate grasping without visual feedback (open loop) reliability of SJ’s grip scaling in her blind field differed significantly
In general, patient SJ’s performance when grasping without vi- from that observed in the normally-sighted control participants,
sual feedback (i.e., in open loop) resembled her performance with t(12) = 4.53, p < 0.001 (see Fig. 6: panels B and D). The results
closed-loop feedback. Accordingly, her PGA increased with the of the dissociation analysis indicated that the reliability of SJ’s grip
width of objects located in her sighted (left) (r(63) = 0.74, scaling to objects located in her blind field in both the immediate
b = 0.35, p < 0.001; r(2) = 0.98), and she scaled her grip to the width (t(12) = 2.71, p < .01) and the delayed grasping conditions
of the object as reliably as the control participants, t(12) = 0.45, (t(12) = 6.63, p < .001) was significantly poorer than it was in
p = 0.66 (see Fig. 5: panels A and C). the normally-sighted controls. But the difference in the reliability
Critically, SJ’s PGA increased with the width of objects located in of SJ’s grip scaling between the immediate and the delayed
her blind (right) field (r(63) = 0.26, b = 0.12, p < 0.05; r(2) = 0.95). condition in her blind field was significantly greater than the
However, SJ’s grip scaling to the width of objects located in her corresponding difference for the normally-sighted controls,
blind field was less reliable than that of the control participants, t(12) = 3.09, p < .01. The same comparison for SJ’s immediate and
t(12) = 1.73, p = 0.05 (see Fig. 5; panels B and D). delayed grasping in her sighted field yielded no evidence that her
performance declined with delay. That is, although SJ performed
5.2.3. Delayed grasping worse overall than controls when grasping in her blind field, the
Consistent with the results from immediate grasping, when pa- significant decrease in the reliability of grip scaling in the delay
tient SJ grasped objects following a delay, her PGA increased with condition for SJ was much greater than the same decrease in per-
the width of objects presented in her sighted (left) visual field formance observed in the control group. To put it plainly, the intro-
(r(63) = 0.51, b = 0.30, p < .001; r(2) = 0.99), and she scaled her grip duction of a 2 s delay had a much greater impact on SJ’s
to the width of the object as reliably as the control participants, performance compared to controls.
t(12) = 1.86, p = 0.09 (see Fig. 6: panels A and C).
The results were quite different for the blind field. In striking 5.2.4. Immediate manual estimation with visual feedback (closed loop)
contrast to her results from the immediate grasping conditions, The pattern of results observed for manual estimation was sim-
SJ’s PGA in the delay condition showed no sensitivity to the width ilar to that observed with delayed grasping. In her sighted (left)
of objects presented in her blind (right) field (r(63) = 0.11, visual field, patient SJ’s MEA increased with object width
b = 0.04, p = 0.38; r(2) = 0.63). Not surprisingly, therefore, the (r(31) = 0.84, b = 0.99, p < .001) (see Fig. 7: panel A). SJ’s grip scaling
918 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

Fig. 5. Peak grip aperture (PGA) in Session 2 plotted as a function of object width for patient SJ when grasping objects located in her sighted (left) visual field (LVF) (panel A)
and blind (right) visual field (RVF) (panel B) with her right (dominant) hand with retinal input available during the programming but not the execution of the grasp (open-
loop visual feedback). Panels C and D each depict the Fisher-transformed correlation coefficient, rF, that describes the relationship between PGA and object width for SJ and
the normally-sighted control participants, who were tested under identical conditions (LVF: panel C; RVF: panel D). Solid error bars denote the 95% confidence interval
around the mean rF for control participants to compare SJ’s rF. Again, similar to what happened in Session 1, SJ showed sensitivity to the width of objects when she grasped
them in both her sighted and her blind field. The reliability of SJ’s grip scaling in her sighted-field (LVF) did not differ significantly from that of the control participants (panel
C). The reliability of SJ’s grip scaling in her blind field (RVF), however, was significantly less than that of the control participants (panel D).

when manually matching the width of the objects located in her to report the size of an object in her blind (right) field by using a
sighted field was as reliable as that observed in the normally- manual estimate, her grip aperture remained sensitive to the width
sighted controls, t(12) = 1.09, p = 0.30 (see Fig. 7: panel C). of that object when she reached out to pick it up. Furthermore, the
In her blind (left) field, however, her MEA showed no sensitivity results from Session 2 show that her ability to scale her grip to the
to object width (r(31) = 0.08, b = 0.11, p = 0.66; r(2) = 0.53) width of objects in her blind field was not entirely dependent on
(see Fig. 7: panel B). Thus, it was hardly surprising to find that having vision available during the movement; SJ continued to scale
the reliability of SJ’s grip scaling to the widths of objects located her grip even when vision was removed at movement onset (open
in her blind field was significantly poorer than it was for grip scal- loop). Importantly, however, SJ’s sensitivity to object width deteri-
ing in the control participants (t(12) = 5.63, p < 0.001), who had orated dramatically when she was cued to reach out and pick up
no trouble indicating the width of the objects (see Fig. 7: panel the object in her blind field only 2 s after vision of her hand and
D). The results of the dissociation analysis indicated that the the object was removed. Thus, the critical factor appears to be
reliability of SJ’s manual estimation of the width of objects located whether or not vision was available during the programming of
in her blind field was significantly poorer than it was for man- the grasping movement.
ual estimations made by the normally-sighted controls, Even though patient SJ showed evidence for grip scaling to ob-
t(12) = 9.06, p < .001. But the difference in the reliability of SJ’s ject width in her blind (right) field, her scaling was not as sensitive
grip scaling between the immediate grasping and manual esti- as that of our normally-sighted control participants. This suggests
mates in her blind field was significantly greater than the corre- that, in the intact brain, V1 must make some contribution to the
sponding difference in the normally-sighted controls, t(12) = 4.61, visuomotor networks mediating grasping. The fact that SJ per-
p < 0.001. The same comparison with respect to controls for SJ’s formed as well in her sighted (left) visual field as control partici-
immediate grasping and manual estimates in her sighted field pants suggests that she does not possess a global deficit in grip
yielded no significant differences. scaling.
The preserved grip scaling in SJ’s blind field raises the question
5.3. Discussion as to how retinal input from that blind field is reaching the visuo-
motor networks, presumably in the dorsal stream, that are control-
In agreement with Perenin and Rosseti’s (1996) findings, the ling this behavior. In the final Session of testing we used fMRI to
results of Session 2 show that even though patient SJ was unable see if area MT, one of the possible routes into the dorsal stream,
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 919

Fig. 6. Peak grip aperture (PGA) in Session 2 plotted as a function of object width for patient SJ when she was cued to grasp objects located in the left visual field (LVF) (panel
A) and right visual field (RVF) (panel B) with her right (dominant) and 2 s after a 1 s preview of the object. Thus, retinal input was not available for either the programming or
execution of the grasp (delayed grasping). Panels C and D each depict the Fisher-transformed correlation coefficient, rF, that describes the relationship between PGA and
object width for SJ and the normally-sighted control participants, who were tested under identical conditions (LVF: panel C; RVF: panel D). Solid error bars denote the 95%
confidence interval around the mean rF for control participants to compare SJ’s rF. Unlike her performance when retinal input was available during the programming of the
grasp, SJ’s PGA is sensitive to the width of objects only when they are located in her sighted field and not when they are located in her blind field. Critically, the reliability of
SJ’s grip scaling in her sighted-field (LVF) did not differ significantly from that of the control participants (panel C). The reliability of SJ’s grip scaling in her blind field (RVF),
however, was significantly less than that of the control participants (panel D).

was still functioning in SJ’s damaged hemisphere, despite the large aligned to a high resolution (1 mm iso-voxel) 192 slice T1 ana-
lesion in V1. tomical scan.
The stimuli used during the MT+/V5 localizer scans were dis-
played on a 15  20 cm screen using an LCD projector (AVOTEC
6. Session 3: experiment
Silent Vision Model 6011). Patient SJ viewed the stimuli on a screen
through a front surface mirror mounted on top of the head coil and
Approximately 5 months following Session 2 testing, we
was instructed to fixate on a red dot (0.5°) presented at the center
brought patient SJ back for an MRI session to acquire high-
of the screen. An MR-compatible infrared camera was used to
resolution anatomical images and measure blood oxygen-level
monitor her eye movements ‘online’. Each scan began with the first
dependent (BOLD) signaling related to moving stimuli in her blind
of nine blocks of 16 s of baseline stimuli, which were interleaved
and sighted visual fields. Our aim was to test whether or not our
with eight blocks of 16 s of motion stimuli. The baseline stimuli
visual stimulus would invoke BOLD signaling in SJ’s intact extras-
consisted of a pseudo-random pattern of white dots (diameter
triate regions, particularly for moving stimuli presented in her
0.1° visual angle and spaced, on average, 1° of visual angle apart)
blind field – and to see if there was any spared tissue around the
that flickered at 1 Hz against a black background. The motion stim-
V1 lesion that would respond accordingly.
uli consisted of the same pattern of white dots translating coher-
ently across the visual field in a random direction. The direction
6.1. Methods of the moving dots was randomly changed every second. The
sighted and blind fields were tested in separate scans of 136 vol-
6.1.1. Data acquisition and analysis umes (approximately 4 min 30 s) each.
The MRI data were collected on a 3 Tesla Siemens Tim-Trio Data were analyzed using BrainVoyager QX 2.1 (Brain Innova-
MRI scanner at the Robarts Research Institute (London, Ontario) tion, Maastricht, The Netherlands). The data were preprocessed
using a 32-channel head coil. To collect functional data we used prior to the statistical analysis. First, the scans were slice scan time
a T2 weighted single shot EPI imaging sequence (TR = 2 s, corrected using a cubic-spline interpolation. Next, the volumes for
TE = 30 ms, 36 slices, 3 mm iso-voxels, flip angle 90°, in-place res- each functional scan were transformed to coincide spatially with
olution 80  80, FOV = 240  240). The functional data were the first volume of the scan closest to the high-resolution
920 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

Fig. 7. Manual estimate aperture (MEA) in Session 2 plotted as a function of object width for patient SJ when she used her forefinger and thumb to manually match the width
of objects located in her sighted (left) visual field (LVF) (panel A) and right visual field (RVF) (panel B) without reaching towards the object and when retinal input was
available during the programming and execution of the manual estimate (closed-loop visual feedback). Panels C and D each depict the Fisher-transformed correlation
coefficient, rF, that describes the relationship between MEA and object width for SJ and the normally-sighted control participants, who were tested under identical conditions
(LVF: panel C; RVF: panel D). Solid error bars denote the 95% confidence interval around the mean rF for control participants to compare SJ’s rF. Similar to her performance
with delayed grasping, SJ’s MEA is sensitive to the width of objects located in her sighted field but not when they are located in her blind field. Critically, the reliability of SJ’s
manual estimates in her sighted-field (LVF) did not differ significantly from that of the control participants (panel C). The reliability of SJ’s manual estimates in her blind field
(RVF), however, was significantly less than that of the control participants (panel D).

anatomical scan using the trilinear-sync correction method. This 6.2. Results
method uses a trilinear nearest neighbor algorithm to detect spa-
tial discrepancies between temporally adjacent volumes and a As can be seen in Figs. 8a and 9a, moving stimuli presented in
sync-interpolative method to estimate the required transformation SJ’s sighted (left) visual field elicited functional activation in and
with three rotation and three translation parameters. Finally, each around intact cortical tissue forming the calcarine sulcus (i.e., stri-
scan was temporally high-pass filtered by regressing the BOLD ate cortex V1) of SJ’s right hemisphere.
time course using a Fourier basis set composed of two sine and In contrast, moving stimuli presented to SJ’s blind (right) visual
cosine functions and a linear component. No spatial smoothing field failed to invoke activation in any of the grey matter immedi-
kernel was applied to the data. ately surrounding SJ’s lesion in her left hemisphere (see Figs. 8b
Statistical analysis was conducted separately for each scan and 9b), even at an uncorrected threshold of p < .01 (Stoerig,
using multiple regression of the percent transformed signal on Kleinschmidt, & Frahm, 1998). Thus, the vascular response and
the convolved ‘predictor of interest’ for motion (scaled to one) underlying metabolic activity of the remaining portion of cortex
and six ‘predictors of no interest’ corresponding to the time course at patient SJ’s occipital pole is not influenced by moving stimuli.
of the six motion parameter estimates derived from the prepro- This result does not lend support to Campion, Latto, and Smith’s
cessing step mentioned above (see e.g., Johnstone et al., 2006). (1983) suggestion (see also Fendrich, Wessinger, & Gazzaniga,
The dependent measure for statistical analysis is the unstandard- 1992) that spared areas of V1 mediate blindsight.
ized regression coefficient for the motion stimuli (bmotion) and rep- Regions beyond the calcarine sulcus, however, showed robust
resents the average measured rate of change in percent BOLD per activation in response to moving stimuli presented in either the
expected incremental increase in BOLD while all other predictors blind (left) or sighted (right) visual fields (see Fig. 10a and b,
are held constant. The resultant statistical map (activation) reflect respectively).
the t-statistic returned from the test of bmotion against zero and was Bilateral activation was found in motion-sensitive extrastriate
thresholded using a cluster extent determined using an extension regions corresponding to the middle temporal and superior tempo-
of the Forman et al. (1995) Monte Carlo method (see Goebel, ral gyri (MT+/V5 and STS, respectively). Moving stimuli presented
Esposito, & Formisano, 2006) with the cluster-level per-family er- in SJ’s sighted (left) or blind (right) visual field invoked bilateral
ror rate (a) held to 0.01 or less. activation in the motion-sensitive areas MT+/V5 and STS of the
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 921

Fig. 8. Medial view of Session 3 functional imaging MT + localizer data for patient SJ Fig. 10. Lateral view of Session 3 functional imaging MT + localizer data for patient
when the moving visual stimulus was presented in the sighted (left) (panel A) or SJ when the moving visual stimulus was presented in the sighted (left) (top panel A)
blind (right) visual field (panel B). The data are displayed on a segmented and or blind (right) visual field (bottom panel B). The data are displayed on a segmented
smoothed version of SJ’s brain to allow an extensive view of the sulci (dark blue) and smoothed version of SJ’s brain to allow an extensive view of the sulci (dark
and gyri (light blue). The yellow circles depict the site of SJ’s occipital lesion. blue) and gyri (light blue). Abbreviations: LH = left hemisphere, RH = right hemi-
Abbreviations: LH = left hemisphere, RH = right hemisphere, LVF = left visual field, sphere, LVF = left visual field, RVF = right visual field, MT+ = middle-temporal
RVF = right visual field, V1 = primary visual cortex. Note the absence of any cortex, STS = superior temporal sulcus. Note the significant bilateral activation in
activation in V1 in the damaged (left) hemisphere when the motion stimulus was the motion-sensitive complex MT + regardless of whether the motion stimulus
presented within her blind (right) visual field. In comparison, the same stimulus appeared within SJ’s sighted (left) or blind (right) visual field. Also note the lack of
presented within her sighted (left) visual field resulted in robust V1 activation in activation at the occipital pole of SJ’s left hemisphere.
the normal (right) hemisphere.

that things are moving within her blind visual field. Indeed, at
the end of the experiment SJ mentioned that on some occasions
she felt like something was moving in her blind field; however,
she could not ‘‘see’’ it. These results are also consistent with previ-
ous work which demonstrated that BOLD signaling in extrastriate
regions can be modulated by moving stimuli presented in the blind
field of patient GY (Goebel, Muckli, Zanella, Singer, & Stoerig, 2001;
Zeki & Ffytche, 1998).

6.3. Discussion

On Session 3, patient SJ was brought back for an fMRI experi-

ment to measure how her intact and lesioned visual cortices would
respond to moving stimuli presented in her sighted and blind vi-
sual fields. The results were clear. Not surprisingly, moving stimuli
located in SJ’s sighted (left) visual field increased BOLD signaling in
the cortex surrounding the calcarine sulcus and in the expected
extrastriate regions of visual cortex, namely MT+ in both hemi-
spheres. Critically, moving stimuli presented in SJ’s blind (right) vi-
sual field failed to modulate BOLD signaling in the cortical tissue
surrounding SJ’s lesion in the calcarine cortex yet these same stim-
uli elicited a robust response bilaterally in MT+. These findings
raise the possibility that SJ’s spared visuomotor abilities could be
mediated by these intact extrastriate pathways.
In addition, the fact that MT+ is activated by moving stimuli in
Fig. 9. Posterior view of Session 3 functional imaging MT + localizer data for patient
SJ when the moving visual stimulus was presented in the sighted (left) (top panel A)
SJ’s blind field may explain why she is sometimes able to sense that
or blind (right) visual field (bottom panel B). The data are displayed on a segmented things are moving within her blind visual field. Indeed, at the end
and smoothed version of SJ’s brain to allow an extensive view of the sulci (dark of the fMRI experiment SJ mentioned that on some occasions she
blue) and gyri (light blue). Abbreviations: LH = left hemisphere, RH = right hemi- felt like something was moving in her blind field but that she could
sphere, LVF = left visual field, RVF = right visual field, V1 = primary visual cortex.
not ‘‘see’’ it. Overall, these results resonate with previous neuroim-
Note the lack of significant activity in and around the left occipital pole for visual
stimuli presented in SJ’s blind (right) visual field. aging work showing that neurovascular responses in extrastriate
regions including MT+ can be modulated by moving stimuli
restricted to the blind field of patients with hemianopia (Barbur,
right hemisphere (see Fig. 10) (Culham, He, Dukelow, & Verstraten, Watson, Frackowiak, & Zeki, 1993; Baseler, Morland, & Wandell,
2001; Dukelow et al., 2001; Watson et al., 1993). The results are 1999; Goebel et al., 2001; Watson et al., 1993; Zeki & Ffytche,
consistent with SJ’s subjective descriptions of her ability to sense 1998).
922 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

7. Conclusion activation in the tissue immediately adjacent to the lesion in the

calcarine sulcus of her damaged hemisphere, certainly suggests
The fact that SJ could scale her grasp to the width of unseen ob- that the dorsal stream could retain access to the information nec-
jects in her blind field suggests that neither conscious vision nor essary for the visual control of grasping and other visually guided
input from V1 is required for grip scaling. In addition, the fact that actions in the absence of V1.
she could do this even when vision was removed at movement on- Importantly, the visually-guided grasping that survives V1
set shows that the visuomotor networks mediating this spared damage in SJ is limited to real-time. In other words, she can scale
ability can generate reliably scaled grasping without visually- her grasp to unseen objects in her blind field only when they are
driven ‘online’ control. It is important to emphasize, however, that present on her retina. When a delay was introduced between last
grip scaling in SJ’s blind field was still impaired compared to grip viewing the object in her blind field and initiating the grasp, she
scaling in the normally-sighted control participants. But in fact this no longer showed any evidence of scaling. It has been suggested
finding is quite consistent with similar findings in a number of pre- that delayed grasping is driven by visual memories of the target
vious investigations showing that patients with V1 lesions have object that are dependent on previous perceptual processing by
relatively shallow grip-scaling slopes in their blind field (e.g., the ventral stream (Goodale, Jakobson, & Keillor, 1994; Goodale,
Jackson, 1999; Perenin & Rossetti, 1996). Nevertheless, even Meenan et al., 1994). In other words, one has to perceive an object
though this suggests that V1 contributes a good deal of sensitivity in the first place in order to grasp it after a delay when that object
to the programming of grasping, it is clear that grip scaling can is no longer visible. Certainly, there is no evidence that SJ perceived
occur in its absence. the target objects in her blind field. Indeed, throughout both test-
Perhaps the most interesting finding in the current set of exper- ing days SJ repeatedly insisted that she ‘‘saw nothing’’ in her blind
iments is the fact that SJ could not scale her grip aperture to the field. These claims were further validated by the fact that SJ was
width of objects presented in her blind field when a 2 s delay was unable to manually estimate the size of the target objects when
introduced prior to the cue to respond. These findings resonate with they were presented within her blind field. The fact that SJ could
recent observations of patient CB who also developed a profound scale her grasp in real-time but not after a delay is consistent with
hemianopia following a V1 lesion (Striemer et al., 2009). Despite Milner and Goodale’s (2006) contention that the dorsal stream
his lesion, CB was able to avoid unseen obstacles presented in his does not store visuomotor transformations for later use, but rather,
blind field in real-time but failed to do so if vision was removed computes them on demand depending on current task require-
2 s before the cue to reach was presented. Taken together, the cur- ments. According to their account, it is the ventral stream that
rent study and the earlier work with CB provide converging evi- mediates actions based on past visual input. SJ’s failure to perceive
dence that the inputs to the visuomotor networks mediating objects in her blind field and the absence of grip scaling in delayed
these abilities operate optimally in real-time using information grasping suggests that perceptual processing in the ventral stream
immediately available on the retina (Milner & Goodale, 2006). (at least for explicit tasks) depends on V1. Taken together then, SJ’s
There is good reason to believe that the visuomotor networks pattern of spared visuomotor abilities and perceptual deficits fits
mediating spared abilities in patients like SJ and CB with action well with Milner and Goodale’s perception and action model.
blindsight reside in the dorsal visual stream (Milner & Goodale, There are a number of remaining questions that need to be ad-
2006). But how is information conveyed to these networks in the dressed in future investigations. First, to what extent can the dorsal
absence of V1? One candidate pathway involves area MT, which stream stripped of its V1 input extract the necessary information to
is known to project to areas in the posterior parietal cortex includ- grasp more complex objects that have constrained grasp points
ing the intra-parietal sulcus and the parietal–occipital cortex (e.g., the ‘‘Blake shapes’’). Second, how ‘real-time’ is the processing
(Blatt, Andersen, & Stoner, 1990; Galletti et al., 2001; Maunsell & in the dorsal stream without V1? In other words, would delays of
Van Essen, 1983; Ungerleider & Desimone, 1986) that have been less than 2 s allow some grip scaling to take place (e.g., Westwood
implicated in visuomotor control (for review see; Born & Bradley, & Goodale, 2003)? Finally, it will also be important for future
2005; Culham, Cavina-Pratesi, & Singhal, 2006). As we reviewed studies to try and further delineate the brain regions and visual
in the Introduction (Section 1), there is evidence in non-human pri- pathways that subserve action blindsight using functional brain
mates that visual input can reach area MT via projections from the imaging and diffusion tensor imaging, similar to recent efforts that
pulvinar, a nucleus in the thalamus that has been shown to receive have examined the possible neural substrates of spatial summa-
input from the superior colliculus (Berman & Wurtz, 2010, 2011; tion, pupillary responses, and other visual abilities that survive
Kaas & Lyon, 2007; Lyon, Nassi, & Callaway, 2010) or even directly damage to primary visual cortex (Leh, Johansen-Berg et al., 2006;
from the retina (Warner et al., 2010). In addition, there is evidence Leh et al., 2009; Tamietto et al., 2010).
for projections to MT and other extrastriate regions from the inter-
laminar layers of the dLGN (Sincich et al., 2004; Vakalopoulos, Acknowledgments
2005) and these projections have been shown to be critical for at
least some residual visually guided behavior after V1 lesions We would like to thank Dr. Haitao Yang for his help with the
(Schmid et al., 2010). Overall, these findings indicate that the dor- software. In addition, we are grateful to Stefanie Fortunato for
sal stream retains some access to bottom-up retinal input in the her assistance with testing the control participants and to Dr. Lore
absence of V1. [Of course, the ventral stream almost certainly has Thaler for providing the motion stimuli for the fMRI experiment.
access to visual inputs that bypass V1 as well (for a discussion of Special thanks go to SJ for her patience and consistent good humor
this issue, see Milner & Goodale, 2006).] during the many hours of testing. This research was supported by a
The fact that patient SJ shows activation in area MT+ when grant from the Canadian Institutes of Health Research to MAG and
viewing moving visual stimuli presented in her blind field shows through a PhD Post-Graduate Scholarship (PGS from the Natural
that input from the retina could potentially reach dorsal-stream Science and Engineering Research Council of Canada (NSERC) to
structures implicated in visuomotor control without involving RLW and an NSERC Postdoctoral Fellowship to CLS.
V1. One has to be cautious in making this inference, however,
because we have not demonstrated directly that visuomotor areas
References
in the dorsal stream are activated when SJ reaches out to grasp
objects in her blind field. Nevertheless, the robust activation in Barbur, J. L., Forsyth, P. M., & Findlay, J. M. (1988). Human saccadic eye movements
area MT+ from unseen visual stimuli coupled with the absence of in the absence of the geniculocalcarine projection. Brain, 111(1), 63–82.
 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924 923

Barbur, J. L., Watson, J. D. G., Frackowiak, R. S. J., & Zeki, S. (1993). Conscious visual Goodale, M. A., Meenan, J. P., Bulthoff, H. H., Nicolle, D. A., Murphy, K. J., & Racicot, C.
perception without V1. Brain, 116(6), 1293–1302. I. (1994). Current Biology, 4, 604–610.
Baseler, H. A., Morland, A. B., & Wandell, B. A. (1999). Topographic organization of Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and
human visual areas in the absence of input from primary cortex. The Journal of action. Trends in Cognitive Neuroscience, 15(1), 20–25.
Neuroscience, 19(7), 2619–2627. Goodale, M. A., Milner, A. D., Jakobson, L. S., & Carey, D. P. (1991). A neurological
Berman, R. A., & Wurtz, R. H. (2010). Functional identification of a pulvinar path dissociation between perceiving objects and grasping them. Nature, 349(6305),
from superior colliculus to cortical area MT+. The Journal of Neuroscience, 30(18), 154–156.
6342–6354. Hendry, S. H. C., & Reid, R. C. (2000). The koniocellular pathway in primate vision.
Berman, R. A., & Wurtz, R. H. (2011). Signals conveyed in the pulvinar pathway from Annual Review of Neuroscience, 23, 127–153.
superior colliculus to cortical area MT. The Journal of Neuroscience, 31(2), Humphrey, G. K., Goodale, M. A., Jakobson, L. S., & Servos, P. (1994). The role of
373–384. surface information in object recognition: Studies of a visual form agnosic and
Blatt, G. J., Andersen, R. A., & Stoner, G. R. (1990). Visual receptive field organization normal subjects. Perception, 23, 1457–1481.
and cortico-cortical connections of the lateral intraparietal area (area LIP) in the Jackson, S. R. (1999). Pathological perceptual completion in hemianpoia extends to
macaque. The Journal of Comparative Neurology, 299, 421–445. the control of reach-to-grasp movements. NeuroReport, 10(12), 2461–2466.
Born, R. T., & Bradley, D. C. (2005). Structure and function of visual area MT. Annual Jakobson, L. S., Archibald, Y. M., Carey, D. P., & Goodale, M. A. (1991). A kinematic
Review Neuroscience, 28, 157–189. analysis of reaching and grasping movements in a patient recovering from optic
Campion, J., Latto, R., & Smith, Y. M. (1983). Is blindsight an effect of scattered light, ataxia. Neuropsychologia, 29(8), 803–809.
spared cortex, and near-threshold vision? Behavioural Brain Sciences, 6(3), James, T. W., Culham, J., Humphrey, G. K., Milner, A. D., & Goodale, M. A. (2003).
423–447. Ventral occipital lesions impair object recognition but not object-directed
Cohen, J., Cohen, P., West, S., & Aiken, L. (2003). Applied multiple regression grasping: An fMRI study. Brain, 126(11), 2463–2475.
correlational analysis for the behavioural sciences. Mahwah, NJ: Lawrence Jeannerod, M. (1988). The neural and behavioural organization of goal-directed
Earlbaum Associates, Inc.. movements. Oxford: Clarendon Press.
Corbetta, M., Marzi, C. A., Tassinari, G., & Aglioti, S. (1990). Effectiveness of different Jeannerod, M., Decety, J., & Michel, F. (1994). Impairment of grasping movements
task paradigms in revealing blindsight. Brain, 113(3), 603–616. following a bilateral posterior parietal lesion. Neuropsychologia, 32(4), 369–380.
Cowey, A. (2010). The blindsight saga. Experimental Brain Research, 200(1), 3–24. Johnstone, T., Walsh, K. S. O., Greischar, L. L., Alexander, A. L., Fox, A. S., Davidson, R.
Crawford, J. R., & Garthwaite, P. H. (2004). Statistical methods for single-case studies J., et al. (2006). Motion correction and the use of motion covariates in multiple-
in neuropsychology: Comparing the slope of a patient’s regression line with subject fMRI analysis. Human Brain Mapping, 27(10), 779–788.
those of a control sample. Cortex, 40(3), 533–548. Kaas, J. H., & Lyon, D. C. (2007). Pulvinar contributions to the dorsal and ventral
Crawford, J. R., & Garthwaite, P. H. (2005a). Testing for suspected impairments and streams of visual processing in primates. Brain Research Review, 55(2),
dissociations in single-case studies in neuropsychologia: Evaluation of 285–296.
alternatives using monte carlo simulations and revised tests for dissociations. Karnath, H. O., & Perenin, M. T. (2005). Cortical control of visually guided reaching:
Neuropsychologia, 19(3), 318–331. Evidence from patients with optic ataxia. Cerebral Cortex, 15(10), 1561–1569.
Crawford, J. R., & Garthwaite, P. H. (2005b). Evaluation of criteria for classical Karnath, H. O., Ruter, J., Mandler, A., & Himmelbach, M. (2009). The anatomy of
dissociations in single-case studies by monte carlo simulation. object recognition – Visual form agnosia caused by medial occipitotemporal
Neuropsychologia, 19(5), 664–678. stroke. Journal of Neuroscience, 29(18), 5854–5862.
Crawford, J. R., Garthwaite, P. H., & Gray, C. D. (2003). Wanted: Fully operational Keith, T. Z. (2006). Multiple regression and beyond. Boston, MA: Pearson.
definitions of dissociations in single-case studies. Cortex, 39(2), 357–370. Leh, S. E., Johansen-Berg, H., & Ptito, A. (2006). Unconscious vision: New insights
Crawford, J. R., Garthwaite, P. H., Howell, D. C., & Venneri, A. (2003). Intra-individual into the neuronal correlate of blindsight using diffusion tractography. Brain,
measures of association in neuropsychology: Inferential methods for comparing 129(7), 1822–1832.
a single case with a control or normative sample. Journal of International Leh, S. E., Mullen, K. T., & Ptito, A. (2006). Absence of s-cone input in human
Neuropsychological Society, 9(7), 989–1000. blindsight following hemispherectomy. European Journal of Neuroscience,
Crawford, J. R., & Howell, D. C. (1998). Comparing an individual’s test score against 24(10), 2954–2960.
norms derived from small samples. The Clinical Neuropsychologist, 12(4), Leh, S. E., Ptito, A., Schonwiesner, M., Chakravarty, M. M., & Mullen, K. T. (2009).
482–486. Blindsight mediated by an s-cone-independent collicular pathway: An fMRI
Culham, J. C., Cavina-Pratesi, C., & Singhal, A. (2006). The role of parietal cortex in study in hemispherectomized subjects. Journal of Cognitive Neuroscience, 22(4),
visuomotor control: What have we learned from neuroimaging? 670–682.
Neuropsychologia, 44, 2668–2684. Lyon, D. C., Nassi, J. J., & Callaway, E. M. (2010). A disynaptic relay from superior
Culham, J., He, S., Dukelow, S., & Verstraten, F. A. (2001). Visual motion and the colliculus to dorsal stream visual cortex in macaque monkey. Neuron, 65(2),
human brain: What has neuroimaging told us? Acta Psychologica, 107(1–3), 270–279.
69–94. Marcel, A. J. (1998). Blindsight and shape perception: Deficit of visual consciousness
Danckert, J., Revol, P., Pisella, L., Krolak-Salmon, P., Vighetto, A., Goodale, M. A., et al. or of visual function? Brain, 121(8), 1565–1588.
(2003). Measuring unconscious actions in action-blindsight: Exploring the Marzi, C. A., Mancini, F., Metitieri, T., & Savazzi, S. (2009). Blindsight following visual
kinematics of pointing movements to targets in the blind field of two patients cortex deafferentation disappears with purple and red stimuli: A case study.
with cortical hemianopia. Neuropsychologia, 41(8), 1068–1081. Neuropsychologia, 47(5), 1382–1385.
Danckert, J., & Rossetti, Y. (2005). Blindsight in action: What can the different sub- Marzi, C. A., Tassinari, G., Aglioti, S., & Lutzemberger, L. (1986). Spatial summation
types of blindsight tell us about the control of visually guided actions? across the vertical meridian in hemianopics: A test of blindsight.
Neuroscience and Biobehavioural Reviews, 29(7), 1035–1046. Neuropsychologia, 24(6), 749–758.
de Gelder, B., Tamietto, M., van Boxtel, G., Goebel, R., Sahraie, A., van den Stock, J., Maunsell, J. H., & van Essen, D. C. (1983). The connections of the middle temporal
et al. (2008). Intact navigation skills after bilateral loss of striate cortex. Current visual area (MT) and their relationship to a cortical hierarchy in the macaque
Biology, 18(24), R1128–1129. monkey. Journal of Neuroscience, 3, 2563–2586.
Dixon, P., & Glover, S. (2009). Perseveration and contrast effects in grasping. Milner, A. D., Perrett, D. I., Johnston, R. S., Benson, P. J., Jordan, T. R., Heeley, D. W.,
Neuropsychologia, 47(6), 1578–1584. et al. (1991). Perception and action in ‘visual form agnosia’. Brain, 114(1B),
Dukelow, S. P., DeSouza, J. F., Culham, J. C., van den Berg, A. V., Menon, R. S., & Vilis, T. 405–428.
(2001). Distinguishing subregions of the human MT+ complex using visual Milner, A. D., Dijkerman, H. C., Pisella, L., McIntosh, R. D., Tilikete, C., Vighetto, A.,
fields and pursuit eye movements. Journal of Neurophysiology, 86(4), et al. (2001). Grasping the past: Delay can improve visuomotor performance.
1991–2000. Current Biology, 11(23), 1896–1901.
Fendrich, R., Wessinger, C. M., & Gazzaniga, M. S. (1992). Residual vision in a Milner, A. D., & Goodale, M. A. (2006). The visual brain in action (2nd ed.). New York:
scotoma: Implications for blindsight. Science, 258(5087), 1489–1491. Oxford University Press.
Forman, S. D., Cohen, J. D., Fitzgerald, M., Eddy, W. F., Mintun, M. A., & Noll, D. C. Perenin, M. T., & Jeannerod, M. (1975). Residual vision in cortically blind hemifields.
(1995). Improved assessment of significant activation in functional magnetic Neuropsychologia, 13, 1–7.
resonance imaging (fMRI): Use of a cluster size threshold. Magnetic Resonance in Perenin, M. T., & Jeannerod, M. (1978). Visual function within the hemianopic field
Medicine, 33(5), 636–647. following early cerebral hemidecortication in man-I: Spatial localization.
Galletti, C., Gamberini, M., Kutz, D. F., Fattori, P., Luppino, G., & Matelli, M. (2001). Neuropsychologia, 16, 1–13.
The cortical connections of area V6: An occipito-parietal network processing Perenin, M. T., & Rossetti, Y. (1996). Grasping without form discrimination in a
visual information. European Journal of Neuroscience, 13, 1572–1588. hemianopic field. Neuroreport, 7(3), 793–797.
Goebel, R., Esposito, F., & Formisano, E. (2006). Analysis of functional image Perenin, M. T., & Vighetto, A. (1988). Optic ataxia: A specific disruption in
analysis contest (FIAC) data with Brainvoyager QX: From single-subject to visuomotor mechanisms I. Different aspects of the deficit in reaching for
cortically aligned group general linear model analysis and self-organizing objects. Brain, 111(3), 643–674.
group independent component analysis. Human Brain Mapping, 27(5), Poppel, E., Held, R., & Frost, D. (1973). Residual visual function after brain wounds
392–401. involving the central visual pathways in man. Nature, 243, 295–296.
Goebel, R., Muckli, L., Zanella, F. E., Singer, W., & Stoerig, P. (2001). Sustained Riddoch, G. (1917). Dissociation of visual perceptions due to occipital injuries, with
extrastriate cortical activation without visual awareness revealed by fMRI especial reference to appreciation of movement. Brain, 40, 15–57.
studies of hemianopic patients. Vision Research, 41(10–11), 1459–1474. Rossetti, T., & Pisella, L. (2002). Several ‘vision for action’ systems: A guide to
Goodale, M. A., Jakobson, L. S., & Keillor, J. M. (1994). Differences in the visual dissociating and integrating dorsal and ventral functions. In W. Prinz & B.
control of pantomimed and natural grasping movements. Neuropsychologia, Hommel (Eds.), Attention and performance XIX: Common mechanisms in
32(10), 1159–1178. perception and action (pp. 62–119). Oxford: Oxford University Press.
924 R.L. Whitwell et al. / Vision Research 51 (2011) 908–924

Rossetti, Y., Revol, P., McIntosh, R., Pisella, L., Rode, G., & Danckert, J. (2005). Visually Warner, C. E., Goldshmit, Y., & Bourne, J. A. (2010). Retinal afferents synapse with
guided reaching: Bilateral posterior parietal lesions cause a switch from fast relay cells targeting the middle temporal area in the pulvinar and lateral
visuomotor to slow cognitive control. Neuropsychologia, 43(2), 162–177. geniculate nuclei. Frontiers in Neuroanatomy, 4, 8.
Sanders, M. D., Warrington, E. K., Marshall, J., & Weiskrantz, L. (1974). ‘‘Blindsight’’: Watson, J. D. G., Myers, R., Frackowiak, R. S. J., Hajnal, J. V., Woods, R. P., Mazziotta, J.
Vision in a field defect. Lancet, 303(7860), 707–708. C., et al. (1993). Area V5 of the human brain: Evidence from a combined study
Schmid, M. C., Mrowka, S. W., Turchi, J., Saunders, R. C., Wilke, M., Peters, A. J., et al. using positron emission tomography and magnetic resonance imaging. Cerebral
(2010). Blindsight depends on the lateral geniculate nucleus. Nature, 466, Cortex, 3(2), 79–94.
373–377. Weiskrantz, L. (1986). Blindsight: A case study and implications. Toronto: Oxford
Sincich, L. C., Park, K. F., Wohlgemuth, M. J., & Horton, J. C. (2004). Bypassing V1: A University Press.
direct geniculate input to area MT. Nature Neuroscience, 7(10), 1123–1128. Weiskrantz, L. (1987). Residual vision in a scotoma: A follow-up study of ‘form’
Striemer, C. L., Chapman, C., & Goodale, M. A. (2009). ‘‘Real-time’’ obstacle avoidance discrimination. Brain, 110, 77–92.
in the absence of V1. Proceedings of the National Academy of Sciences, USA, Weiskrantz, L., Warrington, E. K., Sanders, M. D., & Marshall, J. (1974). Visual
106(37), 15996–16001. capacity in the hemianopic field following a restricted occipital ablation. Brain,
Stoerig, P., Kleinschmidt, A., & Frahm, J. (1998). No visual responses in denervated 97(4), 709–728.
V1: High-resolution functional magnetic resonance imaging of a blindsight Westwood, D. A., & Goodale, M. A. (2003). Perceptual illusion and the real-time
patient. Neuroreport, 9(1), 21–25. control of action. Spatial Vision, 16(3–4), 243–254.
Tamietto, M., Cauda, F., Corazzini, L. L., Savazzi, S., Marzi, C. A., Goebel, R., et al. Whitwell, R. L., & Goodale, M. A. (2009). Updating the programming of a precision
(2010). Collicular vision guides nonconscious behaviour. Journal of Cognitive grip is a function of recent history of available feedback. Experimental Brain
Neuroscience, 22(5), 888–902. Research, 194(4), 619–629.
Ungerleider, L. G., & Desimone, R. (1986). Cortical connections of visual area MT in Zeki, S., & Ffytche, D. H. (1998). The Riddoch syndrome: Insights into the
the macaque. The Journal of Comparative Neurology, 248(2), 190–222. neurobiology of conscious vision. Brain, 121(1), 25–45.
Vakalopoulos, C. (2005). A theory of blindsight – The anatomy of the unconscious: A
proposal for the koniocellular projections and intralaminar thalamus. Medical
Hypotheses, 65(6), 1183–1190.