c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7

Available online at www.sciencedirect.com

ScienceDirect
Journal homepage: www.elsevier.com/locate/cortex

Research report

When apperceptive agnosia is explained by a deficit

of primary visual processing

Andrea Serino a,b,1, Roberto Cecere a,b,2, Neil Dundon a,b, Caterina Bertini a,b,
Cristina Sanchez-Castaneda c and Elisabetta Làdavas a,b,*
a
CsrNC, Centro studi e ricerche in Neuroscienze Cognitive, Polo Scientifico-Didattico di Cesena,
ALMA MATER STUDIORUM e Università di Bologna, Italy
b
Dipartimento di Psicologia, ALMA MATER STUDIORUM e Università di Bologna, Italy
c
IRCCS, Fondazione Santa Lucia, Roma, Italy

article info abstract

Article history: Visual agnosia is a deficit in shape perception, affecting figure, object, face and letter
Received 28 January 2013 recognition. Agnosia is usually attributed to lesions to high-order modules of the visual
Reviewed 08 May 2013 system, which combine visual cues to represent the shape of objects. However, most of
Revised 19 July 2013 previously reported agnosia cases presented visual field (VF) defects and poor primary
Accepted 20 December 2013 visual processing. The present case-study aims to verify whether form agnosia could be
Action editor Laurel Buxbaum explained by a deficit in basic visual functions, rather that by a deficit in high-order shape
Published online 31 December 2013 recognition. Patient SDV suffered a bilateral lesion of the occipital cortex due to anoxia.
When tested, he could navigate, interact with others, and was autonomous in daily life
Keywords: activities. However, he could not recognize objects from drawings and figures, read or
Visual agnosia recognize familiar faces. He was able to recognize objects by touch and people from their
Orientation discrimination deficit voice. Assessments of visual functions showed blindness at the centre of the VF, up to
Crowding effect almost 5
, bilaterally, with better stimulus detection in the periphery. Colour and motion
Cortical blindness perception was preserved. Psychophysical experiments showed that SDV’s visual recog-
nition deficits were not explained by poor spatial acuity or by the crowding effect. Rather a
severe deficit in line orientation processing might be a key mechanism explaining SDV’s
agnosia. Line orientation processing is a basic function of primary visual cortex neurons,
necessary for detecting “edges” of visual stimuli to build up a “primal sketch” for object
recognition. We propose, therefore, that some forms of visual agnosia may be explained by
deficits in basic visual functions due to widespread lesions of the primary visual areas,
affecting primary levels of visual processing.
ª 2013 Elsevier Ltd. All rights reserved.

* Corresponding author. Centro studi e ricerche in Neuroscienze Cognitive, Viale Europa 980, 47521 Cesena, Italy.
E-mail address: elisabetta.ladavas@unibo.it (E. Làdavas).
1
Present address: Laboratory of Cognitive Neuroscience & Center for Neuroprosthetics, Ecole Polytechnique Fédérale de Lausanne,
Switzerland.
2
Present address: Centre for Cognitive Neuroimaging, Institute of Neuroscience and Psychology, University of Glasgow, United
Kingdom.
0010-9452/$ e see front matter ª 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.cortex.2013.12.011
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7 13

precede attributing deficits to impairments in high-level

1. Introduction modules of the visual system involved in shape representa-
tion from visual cues (form agnosia), or from integrating vi-
Visual agnosia is a modality-specific perceptual disorder,
sual details into a synthetic whole (integrative agnosia). More
characterized by impaired shape recognition, which affects
specifically, assessments should target primary visual defects,
the recognition of figures, objects, faces and letters. Lissauer
such as spatial acuity, orientation discrimination and crowd-
(Lissauer, 1890) distinguished between apperceptive agnosia, an
ing. Diffuse neural damage to primary visual areas might
impairment of perceptual processing, and associative agnosia,
reduce visual acuity. Low spatial acuity precludes dis-
where visual perception is roughly spared, but a disorder ex-
tinguishing individual features of visual stimuli for recogni-
ists in accessing stored mental representations of concepts
tion, and therefore, a deficit at this level might results in
from vision. Patients with associative agnosia can distinguish
problems in form perception resembling apperceptive
whether two visual stimuli are similar and can produce an
agnosia. Line orientation may be another factor. According to
accurate copy of an unrecognized figure through passive
most popular models of visual perception (e.g., Marr &
reproduction. In contrast, patients with apperceptive agnosia
Hildreth, 1980), detecting the edges of a visual stimulus is
usually fail in copying and in visual discrimination tasks.
the primary level of analysis necessary for shape perception;
However, in less severe cases, they can discriminate and
orientation-selective neurons in the primary visual areas
compare the size of two shapes, distinguish a figure from a
support this function (Hubel & Weisel, 1959; also see: Ferster &
background and distinguish two overlapping figures. Apper-
Miller, 2000, for a review). Line orientation thresholds (Makela,
ceptive agnosia has further been divided into form agnosia
Whitaker, & Rovamo, 1993) increase at higher eccentricities
and integrative agnosia. Form agnosia causes a deficit in
from the fovea. Thus, where defects involve the centre of the
shape and form discrimination, attributed to an inability to
VF, perceptual deficits in apperceptive agnosia patients might
build shape representations from visual cues; integrative
actually be poor orientation perception. Finally, object recog-
agnosia allows access to components or parts of shapes, but
nition in healthy subjects dramatically decreases in the pe-
compromises integration of these component parts into a
riphery because of the so-called “crowding effect” (Pelli &
coherent whole. In general, perceptual deficits associated with
Tillman, 2008). The visual system recognizes objects by
integrative agnosia appear less severe than form agnosia
detecting and then combining their features. Crowding occurs
(Farah, 2004). In either case, one dominant view in the field
when objects are too close together and details from several
arguments that selective lesions to high-order modules of the
objects are combined in a fused, unrecognizable, percept. This
visual system underlie both form and integrative agnosia
effect does not occur in central vision, and increases propor-
(Humphreys & Riddoch, 1987a).
tionally for stimuli presented at increasing distance from the
By definition, agnosia should not be explainable by primary
fovea. Thus, in cases of defects at the centre of the VF,
sensory deficits such as poor visual acuity, visual field (VF)
apparent apperceptive agnosia might be a function of a
defects, or problems in colour, movement or depth perception
crowding effect.
(Farah, 2004). However, from a systematic review of previously
In the present study, we tested these visual functions in a
reported agnosia cases (see Table 1), many patients also
patient (SDV) exhibiting both the typical lesion profile and the
exhibited visual field defects (e.g., H.J.A. Humphreys & Riddoch,
typical symptomatology consistent with apperceptive
1987b; see Bay, 1953; Ettlinger, 1956) either in the periphery, at
agnosia, in order to investigate whether his apparent agnosia
the centre of the VF (e.g., micro-scotomas, see Campion &
could be a function of acuity, orientation sensitivity or
Latto, 1985; Campion, 1987) or both. Also, visual acuity was
crowding difficulties. SDV suffered a heart-attack, and
rarely formally tested, as agnosia patients are not able to
consequent brain anoxia, three years before our examination.
perform common acuity measures (e.g., Snellen, 1862) which
A bilateral lesion of the occipital cortex initially caused
require letter or object recognition. Furthermore, given the
cortical blindness, which progressively recovered. At
absence of standard measures of visual acuity, comparing
screening, SDV was able to navigate, interact with others, and
results across studies is problematic. Thus, sensory deficits
he was autonomous in daily life activities. However, he was
may, at least partially, explain observed perceptual deficits in
unable to recognize objects from drawings and figures. He
shape recognition.
could recognize real objects by touch, but not by vision. He
Given the nature of lesions causing apperceptive agnosia,
could not read and did not recognize familiar faces. He was
comorbid primary visual deficits are likely to be present. Most
able to recognize people from their voice and movements.
lesions are caused by anoxia, due to carbon monoxide intox-
Motion and colour perception appeared preserved. Thus, SDV
ication or cardiac events (Caine & Watson, 2000). Such lesions
presented typical signs of visual agnosia. At formal testing, he
often induce widespread neural loss, involving primary visual
also showed VF defects: he was blind at the centre of the VF,
areas. Typically, patients initially suffer cortical blindness,
up to almost 5
, bilaterally; visual detection improved in the
and after some recovery, they present signs of agnosia. Thus,
periphery.
considering the aetiology and location of the cerebral damage,
patients with apperceptive agnosia might present elementary
visual deficits which can be identified with a proper psycho-
2. Case history
physical assessment.
Thus, prior to diagnosing agnosia, a deeper analysis of vi-
SDV is a 44-year-old, right-handed man, with 8 years of
sual functioning is necessary to identify the mechanism un-
schooling. He suffered an electrocution-induced heart-attack
derlying impaired shape recognition. Such analyses must
3 years before the present examination. Following the event,
 14
Table 1 e Clinical and lesion data from previous studies on agnosic patients. N.A. [ not available.

Patient References Lesion localisation Aetiology Visual field disease Diagnosis

R.C. - Campion & Latto, 1985 Bilateral occipital cortex. Carbon monoxide poisoning Right inferior defect. Multiple scotomas. Apperceptive agnosia
- Campion, 1987
H.C. - Adler, 1944 Bilateral latero-occipital cortex. Carbon monoxide poisoning Normal periphery; homonymous Integrative agnosia
- Adler, 1950 scotoma subtending the midline of the
- Sparr, Jay, Drislane, & lower visual field (Sparr et al. 1991).
Venna, 1991
Schn. - Landis, Graves, Benson, & Occipital cortex. Trauma Concentric restriction to ca. 30
Integrative agnosia
Hebben, 1982

X - Landis et al., 1982
Soldier - Benson & Greenberg, 1969
H.J.A. - Riddoch & Humphreys, 1987
G.E. - Warrington & James, 1988
M.T. - Warrington & James, 1988
F.G. - Warrington & James, 1988
D.F. - Milner et al., 1991
- Steeves et al., 2006
S.Z. - Grossman, Galetta, &
D’ Esposito, 1997
A.P. - Grossman et al., 1997
J.W. - Vecera & Gilds, 1997
S.M. - Konen et al., 2011
c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
N.A. Mercury poisoning Concentric restriction to ca. 03
Apperceptive agnosia
Bilateral occipito-parietal cortex. Carbon monoxide poisoning Bilateral inferior constriction Form agnosia
Bilateral occipito-temporal cortex, Vascular Bilateral superior altitudinal defect Integrative agnosia
involving the lingual and fusiform gyri.
Occipital cortex Haemorrhagic Left homonymous hemianopia with Apperceptive agnosia
macular sparing
Right parietal cortex Tumoral Left homonymous hemianopia Apperceptive agnosia
Right posterior temporal/parietal cortex. Tumoral Partial incongruent left homonymous Apperceptive agnosia
hemianopia
Bilateral latero-occipital cortex; left Carbon monoxide poisoning Relatively normal perimetry for static Form agnosia
parietal cortex. targets in the central visual field up to 30

eccentricity but with some lower visual
field loss (Steeves et al., 2006).
Bilateral occipito-temporal cortex Ischaemia Bilateral inferior altitudinal defect Apperceptive agnosia
Bilateral occipito-temporal cortex Cortical atrophy N.A. Apperceptive agnosia
Both occipital lobes and right parietal area Anoxic encephalopathy Upper left quadrantopia Apperceptive agnosia
Posterior part of the lateral fusiform Closed head injury None Apperceptive agnosia
gyrus in the right hemisphere, within
lateral occipital complex (LOC).
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7 15

Fig. 1 e SDV’s visual field as shown by the static campimetry test performed at the moment of testing. A) Greyscale images
for the left and the right eye. B) Lower value in decibel at which detected stimuli at each location were presented to the left
and right eye. C) Binocular campimetry.
16 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
SDV suffered cortical blindness for almost 10 months, which
progressively ameliorated. At time of the current evaluation, a
campimetry test (Medmont M700 automated perimetry, Mel-
bourne, Australia) showed a complete loss of central vision,
extending up to 5
in the right visual field (RVF) and the left
visual field (LVF) (see Fig. 1).
SDV could not recognize common objects, faces, letters
and words on sight, although tactile and auditory recognition
was intact. Recognition of drawings was worse (0/10 e Boston
Naming test; Kaplan, Goodglass, & Weintraub, 1983), than
pictures (2/10). Recognition of visually presented real objects
was more accurate (7/10). Further, SDV could describe objects’
features and function, pantomime their use and accurately
classify them semantically.
3. Brain lesion
3.1. Image acquisition
MR imaging was performed on a 3T scanner (Allegra, Siemens
Medical Solutions, Erlangen, Germany). SDV underwent an
MRI protocol including axial, coronal and sagittal T2-weighted
turbo spin-echo (TSE) sequences and axial fluid-attenuated
inversion recovery (FLAIR) sequences covering the whole
brain. Twenty-four 5-mm gapless sections and a 230  230
matrix were obtained using all of the MR imaging techniques
available. The axial and the coronal sections ran, respectively,
parallel and perpendicular to a line joining the anterior and
posterior commissures (ACePC line). Whole-brain T1-
weighted images were obtained on the sagittal plane using a
modified driven equilibrium. All of the images were assessed
visually by a neuro-radiologist. Ten healthy males (mean
age ¼ 43.7; range: 39e50 years) underwent the same imaging
protocol, including whole-brain T1-weighted images, to
compare SDV’s gray matter (GM) and white matter (WM)
reduction to a non-lesioned control group.
Fig. 2 e Structural MRI and lesion analysis. A) Axial FLAIR images. B) Axial (top) and sagittal (bottom) views of lesion
reconstruction. In red colour we highlight the region of interest (ROI), covering the area of injury, marked out in Voxel-based
morphometry (VBM) analysis in order to evaluate a possible GM or WM decrease outside the area primarily affected by the
lesion.
3.2. Voxel-based morphometry (VBM) analysis
Image preprocessing was conducted using the VBM toolbox
implemented in SPM8b (Wellcome Department of Imaging
Neuroscience, London, UK; http://www.fil.ion.ucl.ac.uk/spm/)
using Matlab 7.7 (MathWorks, Natick, MA). Images were first
segmented into tissue classes: GM, WM and cerebrospinal
fluid (Ashburner & Friston, 2000, 2005; Good et al., 2001). A
high-dimensional DARTEL normalization (Ashburner, 2007)
was then applied to the tissue-classified GM and WM maps
modulating for nonlinear effects. Finally, the maps were
smoothed using a Gaussian smoothing kernel of 8-mm full
width at half-maximum. The resulting final voxel-size was
1.5 mm3. The resulting GM and WM smoothed and modulated
images were used in the statistical analysis to assess GM and
WM volume changes. Differences in GM and WM between
groups (patient and healthy controls) were assessed by means
of a two-sample t-test design implemented in SPM8b. Because
normalization of nonlinear effects only had been applied
during VBM preprocessing, which corrects for differences in
GM, WM, and intracranial volumes (Buckner et al., 2004), there
was no further need to control for these variables in the sta-
tistical model.
When performing the second level statistics, we masked the
comparison by a region of interest (ROI) covering the area of
injury to evaluate a possible GM or WM decrease outside the
area primarily affected by the lesion. The ROI was manually
traced over the T1-weighted image using MRIcron (http://www.
mccauslandcenter.sc.edu/mricro/mricron/), considering the
area of the lesion in T2-weighted images; the lesion area was
then smoothed and binary coded (0 for the voxels covering the
lesion; 1 for the rest of the image). All voxels with a value 0 were
excluded from the analysis. We used a non-corrected threshold
(p < .001) for exploratory analysis, and then we corrected for
multiple comparisons using a family wise error (FWE) rate
correction set at p < .05. The MNI coordinates of significant
clusters were converted into Talairach coordinates using a
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
nonlinear MNI to Talairach conversion algorithm (http://www.
bioimagesuite.org/Mni2Tal/index.html). VBM methods and
statistical analyses were performed following the rules for
reporting VBM studies, described by Ridgway et al. (2008).
3.3. Results
Neuroradiological MRI examination revealed two encephalo-
malacic areas, hyperintense on T2-weighted and FLAIR im-
ages, involving the GM and WM of both the occipital lobes,
corresponding to the BA 17-18, and to the BA 19 partially. The
lesion extended rostrally to the superior parietal lobes, cor-
responding to the BA 30-31 and partially to the BA 7 (see
Fig. 2A). In addition, a diffuse hyperintensity on T2-weighted
images was present around both the ventricles, which
appeared enlarged, extending through the callosum body. The
posterior part of the callosum body had reduced thickness.
Adjacent sulci were prominent and the occipital horns of both
ventricles were enlarged, reflecting loss of cerebral tissue
volume. A diffuse, moderate enlargement of the subarachnoid
spaces and sulci was also present, more evident in the mesial
superior frontal and superior parietal lobes bilaterally.
3.3.1. VBM
The lesion ROI is shown in Fig. 2B. After masking for the brain
lesion, the patient presented reduced GM in the right pre-
cuneus (BA 7) and in the parietal WM (posterior part of the
superior longitudinal fasciculus) (p < .001 uncorrected results).
However, when correcting for multiple comparisons (FWE),
the two-sample t-test did not show significant differences
between the subject and the control group.
In sum, SDV suffered extensive lesions in the occipital lobe,
involving BA 17, 18 and partially 19. There was no significant
GM or WM reduction in areas outside the lesion.
4. Neuropsychological assessment
4.1. Visual detection
1. “Visual field Test” from the Testbatterie zur Aufmerksam-
keitsprufung (Computerized Battery for the Exam of
Attention; Zimmerman & Fimm, 1992). SDV was presented
with 100 flickering visual stimuli (1
, duration 1000 msec)
while fixating centrally. SDV detected 87% of stimuli in the
upper and lower RVF, 77% in upper and 68% in the lower
LVF.
2. Computerized Visual Field Test: A stimulus array of
52
 45
(horizontally and vertically, respectively) was
projected on the wall. The viewing distance was 120 cm.
Targets were white dots (1
) presented for 100 msec at
different positions on a black background. In total, 96 tar-
gets were presented, i.e., 24 targets for each quadrant of the
VF. Target detections were indicated with a button-press,
under two conditions: (i) eyes centrally fixated (moni-
tored with an eye-tracker) and (ii) eyes free to move. In
condition (i) SDV detected 71% of stimuli in both the upper
and lower RVF (d0 ¼ 1.22; c ¼ .06 for each location) and 58%
(d0 ¼ .88; c ¼ .24) and 33% (d0 ¼ .24; c ¼ .55) of stimuli in the
upper and lower LVF, respectively. In condition (ii) SDV was
17
less accurate; 67% (d0 ¼ .1.22; c ¼ .23) and 42% (d0 ¼ .94;
c ¼ .68) in the upper and lower RVF respectively and 46%
(d0 ¼ 1.05; c ¼ .63) and 29% (d0 ¼ .6; c ¼ .85) in the upper and
lower LVF respectively.
3. Longitudinal and altitudinal visual detection test: A
custom-made semi-circular apparatus presented light
stimuli (red LED) in a dark environment (luminance 90 cd/
m2). Stimuli presented for 100 msec at eight possible
positions: 8
, 24
, 40
, 56
both in RVF and LVF. The test
was repeated in 3 sessions, (i) midline, (ii) lower (20
)
and (iii) upper VF (20
). Eyes were fixated centrally and
monitored by an experimenter behind the apparatus. In
each session, 48 stimuli per position and 96 catch trials
were presented. Results (see: Fig. 3) indicate a gradient of
increasing visual deficit from the periphery to the centre
of VF. SDV was above chance in the periphery of the VF,
but impaired at 8 deg both in RVF and LVF. Collapsed
average detection was above chance for the 3 latitudinal
positions.
4. Equiluminant detection: Gabor patch target stimuli (1
;
spatial frequency (SF) ¼ 2 cycles/deg), were displayed on a
grey homogenous background, in 1 of 6 possible positions
(i.e., at 2.5
, 5
, 7.5
, 10
, 12.5
and 15
) in both RVF and LVF
(see Fig. 4A). Each trial (n ¼ 50) presented either a target or
no target, with equal probability. After each trial, SDV
verbally reported whether a stimulus had been presented.
Results (see Fig. 4) indicate that detection was at chance at
2.5
in both RVF and LVF, and at 5
in LVF. Detection at all
other positions was above chance.
5. Line detection. The task was the same as the previous one,
but horizontal white lines (length ¼ 2
), instead of Gabor
patches, were presented at 5
, 10
and 15
in LVF and RVF.
Results (Table 2) show that SDV’s detection ability was
significantly above chance at any position (all ps < .00001).
In summary, SDV’S detection ability was above chance in
the periphery of the VF. Almost 5
radius around the fovea was
almost blind, while detection progressively improved from the
fovea to 10
of eccentricity; performance was better in the LVF
than RVF.
4.2. Shape perception
Results from the Birmingham Object Recognition Battery
(BORB; Riddoch & Humphreys, 1993) are reported in Table 3. In
the size-matching sub-test, SDV performed poorly (53% cor-
rect responses). In the overlapping shapes sub-tests, he was
severely impaired with letter stimuli and mostly below cut-off
scores with geometrical figures. He was unable to perform the
line-matching test, figure recognition tasks with non-
prototypical figures, nor tasks of figure association and
figure denomination (see Table 3).
4.3. Face perception
Thirteen famous characters (10
 8
) were presented by
computer screen for a maximum of 10 seconds. SDV was able
to name only 2/13 famous faces. Healthy subjects immedi-
ately recognized all faces.
18 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7

Fig. 3 e Visual detection test administered by a semi-circular, custom-made, apparatus. SDV’s percentage of correct
detections are reported for each tested position. d0 and c values are indicated for each latitudinal position of the visual field
(H: high; M: median; L: low). Asterisks indicate a pathological detection performance.

Fig. 4 e A) Visual detection test with equiluminant stimuli (Gabor patches). The figure shows all the tested positions;
numbers indicate the eccentricity from the fixation in visual degrees. B) SDV’s percentage of correct detections with
equiluminant stimuli. Accuracy is reported for all tested eccentricities from the fixation. Chance level is indicated by the
dotted line. C) SDV’s d0 scores in the visual detection test with equiluminant stimuli. d0 values are reported for all tested
eccentricities from the fixation.
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7 19

Table 2 e SDV’s ability to detect horizontal white line (length 2
) at different locations in left and right visual fields (5, 10, 15
visual degrees eccentricity). Percentage of correct detection and d prime scores are reported.

Location 15
10
5
þ5
þ10
þ15

Accuracy 94% 100% 98% 86% 92% 96%
d0 4.33 6.05 4.9 3.47 2.81 4.56

4.4. Reading
SDV was unable to read any letter strings from 55 concrete
Italian words or 55 legal non-words (see Ladavas, Shallice, &
Zanella, 1997). String stimuli were presented one-at-a-time,
in upper-case 18-point Palatino font, at the centre of an A4
piece of paper. String length ranged from 6 to 11 letters. In-
structions were to read the letter string aloud. Omitting or
misreading one or more letters was considered as an error for
the whole letter string.
4.5. Colour perception
SDV accurately named 10 colours (black, red, yellow, blue,
green, orange, purple, pink, grey and brown), presented as
solid-coloured rectangles (12
 5
) on a 1700 computer screen.
We also used a discrimination task to further investigate
colour perception in SDV. On each trails, two solid-coloured
rectangles (8
 9
) were presented with their centre placed
at 5
on the left and on the RVF. In 6 trials, the two rectangles
were of two different colours, in 7 trials they were of the same
colour and of the same hue, and in 14 trials they were of the
same colour, but of different hue. Order of colour and hue
combinations was randomized across trials. On each trail,
SDV was asked to judge whether the two rectangles were
exactly of the same colour and the same hue. Eight healthy
subjects (2 males; mean age 54 years; range 48e58 years) were
also tested as controls. SDV’s performance was not different
than those from healthy controls (error rate SDV: 11%; error
rate, controls: 8%; S.E.M.: 1%; t ¼ 2.043; p ¼ .08). It is worth
noting that neither the present colour naming nor colour
discrimination tasks are optimal measures to deeply assess
colour perception, such as for instance the Ishihara colour
blindness test (Sloan & Habel, 1956) or the Farnsworth-
Munsell 100 hue test (Farnsworth, 1956). Those tasks, how-
ever, rely on reading or on central vision, functions that were
impaired in SDV. Therefore, we acknowledge that SDV’s un-
impaired performance in the tasks used in the present study
cannot be used as an argument to conclude that his colour
vision was perfectly intact. However, it would be sufficient to
exclude severe problems in colour perception, as those found
for shape perception for instance.

Table 3 e SDV’s scores and cut-off values for clinical impairment in sub-tests from the Birmingham Object Recognition
Battery (BORB). Asterisks indicate a pathological performance. N.E. [ Not executed.

Sub-test Correct responses Cut-off

Test 1: Copying of drawings 0%* 100%
Test 2: Matching of line length N.E. >80%
Test 3: Matching of stimulus size 53%* >77%
Test 4: Matching of line orientation N.E. >67%
Test 5: Matching of gaps in circles N.E. >67%
Test 6: Overlapping figures Single letters 36%* >83%
Couples of letters 13%* >83%
(not overlapping)
Couples of letters (overlapping) 6%* >83%
Triplets of letters (not overlapping) 0%* >83%
Triplets of letters (overlapping) 0%* >83%
Single geometrical shapes 78%* >83%
Couples of geometrical shapes (not overlapping) 58%* >83%
Couples of geometrical shapes (overlapping) N.E. >83%
Triplets of geometrical shapes (not overlapping) N.E. >83%
Triplets of geometrical shapes (overlapping) N.E. >83%
Single objects 9%* >83%
Couples of objects N.E. >75%
(not overlapping)
Couples of objects (overlapping) N.E. >90%
Test 7: Recognition across different viewpoints (details) N.E. >76%
Test 8: Recognition across different viewpoints (axes) N.E. >64%
Test 9: Drawing from memory 100% 100%
Test 10: Object decision N.E. >72%
Test 11: Item match N.E. >81%
Test 12: Associative match N.E. >73%
Test 13: Picture naming (living) N.E. >53%
Test 14: Picture naming (non living) N.E. >85%
20 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7

4.6. Movement perception 4.7. Visuo-constructive abilities

Modified Random Dot Kinematograms (Gummel, Ygge,
Benassi, & Bolzani, 2012): A round-shaped patch (ø ¼ 9
) of
150 small moving high-luminance dots was displayed on a
black background on a computer screen, 50 cm from SDV. In
each trial, dots could either move coherently in one of eight
directions (four cardinal, four oblique) at a constant velocity of
6.1
/s, or randomly in a Brownian movement. Signal-to-noise
ratio decreased exponentially across 5 levels, at a rate of 63%
of the previous level. Thus, coherence reduced from level 1 (all
150 dots in coherence) to level 5 (24 dots in coherence). Each
signal-to-noise level was presented eight times and each trial
lasted 200 msec to avoid tracking.
The task was presented at the fovea, 10
in RVF and 10
in
LVF. Eyes were always centrally fixated, monitored with an
eye-tracker (Eye-Track ASL-6000; sampling rate 60 Hz). Eight
age-matched subjects (7 males; mean age 43.1 years; range
29e53 years) were also tested.
A 3  5 two-way ANOVA revealed that Healthy subjects’
accuracy was equivalent at all 3 VF positions (main effect of
Position: p ¼ .76). Performance progressively decreased at
decreasing levels of signal-to-noise ratio [main effect of
Signal-to-Noise ratio F(4,28) ¼ 37.29, p < .00001] (see Fig. 5A).
The Position  Signal-to-Noise interaction was not significant
(p ¼ .56). A separate ANOVA showed that SDV’s accuracy was
also equivalent at the three tested positions (p ¼ .38) and ac-
curacy progressively decreased as a function of decreasing
signal-to-noise ratio (see Fig. 5B for post hoc results).
To compare SDV and controls, we computed the linear
function between accuracy and signal-to-noise. The slope
reflects motion sensitivity, while the intercept reflects a global
index of performance. Slope values were not significantly
different between SDV and controls at any position in VF (SDV:
left ¼ .44, centre ¼ .62, right ¼ .68; Controls: left ¼ .59,
centre ¼ .54, right ¼ .49), or averaged (all ps > .35). However,
intercept values differed significantly between SDV
(left ¼ 21%, centre ¼ 15%, right ¼ 21%) and controls (left ¼ 38,
centre ¼ 44, right ¼ 43, all p < .01), suggesting SDV’s sensitivity
to motion is not different to healthy controls, after correcting
for his poorer global performance.
SDV was relatively good at copying simple geometrical shapes
(5/7 correct reproductions; see Fig. 6A, second panel). How-
ever, when presented with complex stimuli, his copy was
severely impaired (see Fig. 6B). He was also unable to copy
object drawings (0/3; see Table 2). Drawing from memory was
perfect (3/3; See Fig. 6A, third panel).
4.8. Visual imagery
SDV could accurately recall perceptual details from mental
imagery. On verbal presentation (see Policardi et al., 1996 for
stimuli/materials), he could indicate whether an animal has a
long or short tail (17/17 correct) or has upward or downward
ears (13/13), whether a letter has curved or straight lines (11/11),
and whether cities are situated in north or south Italy (20/20).
4.9. Visual search
Control condition: SDV accurately detected 20/20 trials when a
square or circle (red or blue; 2
) randomly appeared in 1 of 30
spatial positions (in the upper or lower LVF and RVF; total area:
20
 15
) on a 1700 monitor. Eyes were fixated centrally and
monitored with an eye-tracker. Twenty stimuli were
randomly mixed with 10 catch trials; SDV reported verbally
upon stimulus detection.
Visual Search condition (shape): SDV failed to discriminate
the target stimulus shape (0/40), when a single square (red or
blue) was presented in an array of 29 circles (red or blue), or
vice-versa. SDV successfully withheld responses in catch tri-
als, i.e., where all stimuli were the same.
Visual Search condition (colour): SDV accurately discrimi-
nated 40/40 target stimuli colours, when a red stimulus (square
or circle) was presented in an array of 29 blue stimuli (squares
or circles), or vice-versa. He also did not respond to catch trials.
4.10. Conclusion from neuropsychological assessment
In summary, SDV presented a central VF deficit, with relative
sparing in the periphery. Colour and motion perception, and

Fig. 5 e Results of the movement perception test. A) Healthy controls’ percentage of correct responses in the left (white
columns), central (grey columns) and right (black columns) visual fields, at decreasing levels of signal-to-noise ratio (from
left to right). Error bars represent the S.E.M. B) SDV’s percentage of correct responses in the left (white columns), central (grey
columns) and right (black columns) visual fields, at decreasing levels of signal-to-noise ratio (from left to right).
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
Fig. 6 e Illustrations of SDV’s performance in visuo-
constructive task. A) Copy and draw from memory of
simple shapes. B) Copy of complex figures.
visual search performance for non-shapes, were normal.
However, shape, face perception and reading were severely
impaired. Copying object drawings was impaired, while
drawing from memory and visual imagery was normal.
SDV’s recognition deficits appear to therefore stem from
impaired perceptual processing of visual information, rather
than impaired access to representations of objects from
vision: indeed, his pattern of abilities (i.e., object drawing from
memory, object judgement from imagery, object recognition
by tactile information, object semantic classification and ob-
ject function description) strongly suggests mental
21
representations of shapes are spared. Thus, SDV’s overall
pattern of impairments and spared functions are compatible
with a diagnosis of visual agnosia. On the other hand, a
diagnosis of pure integrative agnosia cannot be proposed
given the severity of SDV’s perceptual deficits, involving not
only integration, but also basic processing of single elements
of the visual scene (Farah, 2004).
However, a primary sensory impairment may be underly-
ing his symptomatology. In order to test this hypothesis, we
conducted further psychophysical experiments, to assess
SDV’s spatial acuity (contrast sensitivity), orientation
discrimination and crowding, at various eccentricities in the
periphery of his LVF and RVF, relative to controls.
5. Experimental investigation
5.1. Experiment 1. Contrast sensitivity
5.1.1. Method
SDV and 6 age-matched controls (all males; mean age 42.6
years; range 39e52 years) discriminated the orientation of a
square grating (2
; duration 250 msec), which randomly tilted
either 45
left or 45
right. Following each block (n ¼ 40), SF was
increased by a logarithmic factor, starting from .5 cycles per
degree. SF increased block-by-block, until participants
reached chance level (50%) for an entire block. Stimuli were
presented at a distance of 57 cm on a 1500 screen, after lumi-
nance was equated by a luminometer. Eyes were fixated
centrally and monitored with an eye-tracker. The experiment
was run at 3 locations: 10
LVF, 10
RVF and at the fovea.
5.1.2. Results
Accuracy scores for SDV and controls are reported in Fig. 7A.
SDV’s threshold was lower (indicating higher sensitivity) for
stimuli presented in the RVF (16 cycles/deg) than in LVF (8
cycles/deg); at the fovea, SDV was at chance, even at lowest
spatial frequencies. However, although SDV’s accuracy was
significantly lower than controls for all spatial frequencies in
either periphery (all ps < .05), the ensuing contrast sensitivity
function presented a similar, normal, “U” shape profile to
controls. We therefore repeated the experiment in controls,
after setting task difficulty to roughly equated that for SDV. To
this aim, we used a staircase procedure to adjust their base-
lines to SDV’s baseline (w70% for 1 cycle/deg). A 1 cycle/deg
grating was initially presented at 45
, and tilting was reduced
progressively by 50% after each set of three successive correct
responses. Errors reversed increment direction to the
midpoint of the error trial and the previous correct trial.
Threshold was assumed after three changes around the same
tilting values and then confirmed by scores between 6/10 and
8/10 correct for a trial block of 10 stimuli. Otherwise, the
staircase procedure was repeated. Controls then proceeded
through the experiment as above. Results (see Fig. 7B) show
that at this setting, SDV’s discrimination accuracy (with
orientation at 90
) was not significantly different to controls
(orientation at an average of 1.46
; S.E.M. ¼ .180) at any SF (all
ps > .10).
These results suggest contrast sensitivity cannot account
for SDV’s perceptual deficits; after differences in grating
22 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
Fig. 7 e Contrast sensitivity results in the periphery of the right visual field (10
) without (A) and with (B) threshold correction
for healthy controls. Percentage of correct responses is reported for healthy controls (grey line) and SDV (black line) as a
function of spatial frequency. Chance level is indicated by the dotted line.
orientation perception were taken into account, SDV’s
contrast sensitivity profile was comparable to that of controls.
On the back of these results, we next tested SDV’s sensitivity
to orientation.
5.2. Experiment 2. Line orientation
5.2.1. Method
SDV and 8 age-matched controls (5 males; mean age 40 years;
range 30e57 years) verbally indicated the orientation of a
white line (length ¼ 2
), presented either vertically or hori-
zontally. Each stimulus block (n ¼ 50; equal number of
randomly ordered horizontal and vertical lines) was presented
on an equiluminant grey background; 250 msec per stimulus.
Eyes were fixated centrally and monitored with an eye-
tracker. 7 possible positions of VF were tested: at the fovea
and at 5
, 10
and 15
in LVF and RVF.
5.2.2. Results
Results (see Table 4) indicate that SDV’s accuracy was be-
tween 64% and 79% at peripheral positions in RVF, and 58%
and 73% in LVF. He was at chance for stimuli presented at the
centre of the VF. His performance was statistically above
chance at 10
and 15
in RVF and 5
and 10
in LVF (Fisher test,
all ps < .05). Controls were 100% accurate in all tested posi-
tions, significantly higher than SDV (all ps < .0001).
In order to quantify line orientation sensitivity for SDV, we
used a staircase procedure to identify the minimal difference
between two lines orientations SDV was able to perceive with
an accuracy level of 70%. To this aim, we presented SDV with
lines with a different degree of tilting, asking him to verbally
indicate whether the line was vertically presented or titled.
The staircase procedure started with a horizontal line (90
of
tilting) and then tilting was reduced progressively by 50% after
each set of three successive correct responses. Errors reversed
increment direction to the midpoint of the error trial and the
previous correct trial. Threshold was assumed after three
changes around the same tilting values and then confirmed by
scores between 6/10 and 8/10 correct for a trial block of 10
stimuli. Otherwise, the staircase procedure was repeated.
SDV’s thresholds were calculated to be 74
, 21
and 38
of
line tilting, at 5
, 10
and 15
respectively. Orientation
thresholds were assessed also in healthy controls, with their
average thresholds proving to be 1.36
(S.E.M. ¼ .21), 2.18

(.49) and 2.39
(.47) deg at 5
, 10
and 15
; these thresholds
significantly increased with eccentricity [F(2,14) ¼ 9.48;
p < .01]. SDV’s thresholds were significantly higher than
controls in all positions (all ps < .0001).3 Thus, in sum, SDV
could roughly discriminate line orientation at different posi-
tions of his visual periphery, however his performance was
severely impaired, compared to controls. In the next experi-
ment, we investigated whether SDV’s residual perceptual
abilities in the periphery of VF were further affected by the
crowding effect.
5.3. Experiment 3. Crowding effect
5.3.1. Method
In Experiment 3A, as before, SDV and 8 age-matched controls
(5 males; mean age 40 years; range 30e57 years) verbally
discriminated the orientation of a target white line
(length ¼ 2
), presented either vertically or horizontally. Seven
possible positions of VF were tested: at the fovea and at 5
, 10

and 15
in LVF and RVF. In addition, three levels of crowding
were introduced (0 Flankers e stimulus on its own; 0FL, 2
Flankers e a flanker 2.5
east and west of the stimulus; 2FL or 4
Flankers e a flanker 2.5
north, south, east and west of the
stimulus; 4FL). Flanker stimuli were diagonal 2
white lines,
randomly oriented clockwise or anticlockwise. Each block of
stimuli (n ¼ 25) was presented on an equiluminant grey
background; 250 msec/stimulus. Eyes were fixated centrally
and monitored with an eye-tracker. 7 possible positions of VF
were tested: at the fovea and at 5
, 10
and 15
in LVF and RVF.
Two blocks (50 stimuli per block) were performed per flanker
3
In a similar experiment, we measured SDV’s line orientation
threshold with lines centred at the same spot of the visual field
(10
in the right visual field), but tilted anticlockwise. With the
same procedure used for Experiment 4.2, we presented lines at
10
of the right visual field, either vertical or tilted anticlockwise,
on a range between 0 and -90
of rotation. SDV’s line orientation
threshold in that condition was 20
, almost identical to the 21

found with lines presented at 10
, rotated clockwise. Thus, it
seems that the same deficit in line orientation affects different
portions of the visual field. This result rules out the hypothesis
that SDV’s impaired performance in line orientation discrimina-
tion could be explained by the presence of a patchy scotoma,
affecting one portion of SDV’s visual field.
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
Table 4 e SDV’s results in the line orientation task from Experiment 2. Percentage of correct responses, d0 and c values are
reported for stimuli presented in the central and left/right visual fields (5, 10, 15 visual degrees eccentricity).
Left visual field
Eccentricity 15 vd 10 vd
Correct responses (%) 58% 73%
d prime .43 1.33
c .37 .3
condition, per spatial position. Experiment 3B was an identical
design, except that SDV and controls identified 2
geometric
shapes (triangles, squares, rhombus or pentagons); flankers
were parallelograms (at 2
distance).
5.3.2. Results
5.3.2.1. EXPERIMENT 3A: LINE ORIENTATION TASK. At the fovea,
SDV’s accuracy was at chance level for all flanker conditions
(0FL ¼ 51%; 2FL ¼ 54%; 4FL ¼ 55%; Fisher test, all p-
values > .80). In RVF, accuracy in 0FL conditions at 5
, 10
and
15
was 72%, 73% and 80% respectively (all p-values < .05, i.e.,
above chance). In 2FL conditions, performance at 5
, 10
and
15
reduced to chance, i.e., 52%, 52% and 57% respectively (all
p-values > .60; see Fig. 8B). In 4FL conditions, performance
remained at chance, 55%, 46% and 56% respectively for 5
, 10

and 15
(all p-values > .50; see Fig. 8B). At any eccentricity in
RVF, SDV’s performance was significantly higher in 0FL con-
ditions, compared to both 2FL and 4FL (all p-values < .05).
SDV’s accuracy in the RVF therefore appears sensitive to a
crowding effect.
Similar results were found in LVF. Accuracy in 0FL condi-
tions at 5
, 10
and 15
was 75%, 75% and 68% respectively (all
p-values < .05, i.e., above chance). In 2FL conditions, perfor-
mance at 5
, 10
and 15
again reduced to chance, i.e., 62%, 52%
and 61% respectively (all p-values > .30), as in the 4-Flankers
condition: 5
(52%), 10
(53%) and 15
(61%) (all p-values > .40).
Mean accuracy for controls ranged from 100% to 96%, with
no significant differences across conditions. An ANOVA with
the factors Eccentricity (5
, 10
and 15
) and Flankers (0, 2, 4)
revealed no main effects, nor interaction (all p-values > .07).
Controls’ performance was significantly higher than SDV’s for
all conditions (all ps < .0001).
Given that controls were likely to be at ceiling discrimi-
nating vertical and horizontal lines, we again set individual-
ized thresholds, using the calibration methodology in
Experiment 2 (see: Fig. 8B).
An ANOVA using these new accuracy scores, with the
factors Eccentricity (5
, 10
and 15
) and Flankers (0, 2, 4)
showed a significant main effect of Flankers [F(2,16) ¼ 63.84;
p < .0001]. Accuracy was higher when stimuli were presented
in isolation (0FL mean accuracy ¼ 73%, S.E.M. ¼ .002) than
when surrounded by two (2FL ¼ 57%, .017; p < .0001) or by
four flankers (4FL ¼ 50%, .021; p < .0001). The main effect of
Eccentricity was also significant [F(2,16) ¼ 6.156; p < .01]; per-
formance decreased as stimuli were presented more periph-
erally, driven by the significant difference between accuracy
at 5
(62%; S.E.M. ¼ .12) and 15
(57%  .15, p < .01); accuracy
at 10
(61%  .13) was not significantly different from the other
two conditions (both ps > .11). The two-way interaction
Flankers  Eccentricity was also significant [F(4,32) ¼ 4.34;
p < .01]; the effects of two flankers was lower at the lowest
23
Centre Right visual field
5 vd 0 vd 5 vd 10 vd 15 vd
70% 48% 64% 79% 68%
1.09 .1 .72 1.64 .94
.29 .2 .11 .12 .11
eccentricity (5
¼ 64%  .02), compared to higher eccentricities
(10
¼ 55%  .02; p < .01; 15
deg ¼ 53%  .02; p < .01). This
flanker  eccentricity effect was less evident in the 4-Flankers
condition.
In 0FL conditions, controls’ performance after threshold
setting was not different from SDV at any eccentricity (all
ps > .5). We then calculated the accuracy differences between
the 0FL and the 2FL and 4FL conditions, to index crowding
effect. These index scores were only significantly different
between SDV (20%) and controls (8%) in the 2FL condition at 5

(p < .001), presumably due to controls’ lower sensitivity to
crowding for stimuli presented closer to the fovea. In all other
conditions, the crowding index was not significantly different
between SDV and controls. (5
, 4FL: SDV ¼ 17%;
Controls ¼ 21%; 10
, 2FL: SDV ¼ 21%; Controls ¼ 18%; 10
, 4FL:
SDV ¼ 27%; Controls ¼ 18%; 15
, 2FL: SDV ¼ 23%;
Controls ¼ 21%; 15
, 4FL: SDV ¼ 24%; Controls ¼ 29%; all p-
values > .10). These findings suggest that once a general dif-
ference in line orientation perception is controlled, crowding
similarly affects both SDV and controls.
5.3.2.2. EXPERIMENT 3B, SHAPE PERCEPTION. SDV’s shape discrimi-
nation accuracy was above chance (25%) only in the 0FL con-
dition at 10
RVF (47%, p < .05). At all other positions, he was at
chance, at the fovea: (0FL ¼ 27%; 2FL ¼ 30%; 4FL ¼ 23%; all
ps > .64) in RVF: (5
, 0FL ¼ 18%; 2FL ¼ 31%; 4FL ¼ 23%; 15
,
0FL ¼ 34%; 2FL ¼ 23%; 4FL ¼ 19%, all ps > 50) and in LVF: (5
,
0FL ¼ 25%; 2FL ¼ 29%; 4FL ¼ 23%; 10
, 0FL ¼ 30%; 2FL ¼ 29%;
4FL ¼ 8%; 15
, 0FL ¼ 26%; 2FL ¼ 27%; 4FL ¼ 23%; all ps > .81).
Performance also reduced to chance at 10
RVF, for the 2FL and
4 FL condition (29% and 38%, both ps > .27). Thus, SDV could
partially perceive single shapes, at the position of his VF
where he exhibited relatively consistent detection, however
flankers abolished this residual perception.
Controls outperformed SDV in all conditions (all ps > .02).
An ANOVA on control data with factors: Flankers (0, 2, 4) and
Eccentricity (5
, 10
, 15
), revealed a Flanker main effect
[F(2,16) ¼ 58.7; p < .0001]; performance decreased for 2FL (70%)
or 4FL (72%), compared to 0FL (100%). The Eccentricity main
effect was also significant [F(2,16) ¼ 84.4; p < .0001]; perfor-
mance progressively decreased from 5
(98%) to 10
(84%) and
15
(60%) (all ps < .05). The Flanker  Eccentricity interaction
was also significant [F(4,32) ¼ 57.7; p < .00001]; the effect of
Flankers was weak at 5
deg (0FL ¼ 100%, 2FL ¼ 98%;
4FL ¼ 96%), stronger at 10
(0FL ¼ 100%; 2FL ¼ 74%; 4FL ¼ 78%)
and at 15
(0FL ¼ 100%; 2FL ¼ 38%; 4FL ¼ 40%). These results
support previous findings (see Levi & Carney, 2009; Pelli &
Tillman, 2008; for reviews); a significant crowding effect was
apparent for both line and shape perception, with stronger
flanker interference at higher eccentricities. In general, per-
formance was worse in SDV than in controls, however no
24 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7

Fig. 8 e A) Examples of stimuli in the 2-Flanker condition of Experiment 3A. For SDV, the target line could be either vertical
(left panel) or horizontal (right panel) and it was flanked by two diagonal (45
) lines. In the 4-Flanker condition, two more
flankers were added on top and at the bottom of the target. For healthy controls, the target line could be either vertical or
tilted at the individually set threshold (see Methods) and the flankers were the same as for SDV. B) Results of Experiment 3A
e healthy participants. Controls’ percentage of correct responses is reported for the No-Flankers (white columns), 2-Flankers
(grey columns) and 4-Flankers (black columns) conditions. The graph shows the average performance in both visual fields.
C) Results of Experiment 3A e SDV. Percentage of correct responses is reported for the No-Flankers (white columns), 2-
Flankers (grey columns) and 4-Flankers (black columns) conditions, separately for the left and the right visual field. While
the performance of both SDV and controls was above chance in the No-Flankers condition, it dropped at chance level
(indicated by the dotted line) in the 2-Flankers and 4-Flankers conditions.

between group difference was observed when orientation
discrimination was controlled. Because SDV’s perception was
poor, even in no-flanker conditions, increasing the number of
stimuli in the VF affected SDV more than controls. It is note-
worthy that SDV cannot foveate stimuli to the “uncrowded
window” (Pelli & Tillman, 2008), thus he is more exposed to
the crowding effect in everyday life.
6. General discussion
The present case-study describes a neuropsychological and
psychophysical investigation of patient SDV, who initially
developed cortical blindness following anoxia, before partially
recovering. Subsequent impairment in visual shape process-
ing compromised recognition of visually presented objects,
pictures, drawings, faces and letters. Concept representations
remained intact, as did recognition of objects by touch, or
auditory description. Colour and motion perception were
spared, as were abilities to navigate, perform routine everyday
life activities and interact with objects and others. Such a
pattern of impaired and spared abilities would initially sug-
gest that SDV suffers from apperceptive agnosia, and, more
specifically, from form agnosia (Campion, 1987). However,
SDV was almost completely blind at the centre of VF (w5

around the fovea), with visual stimulus detection relatively
preserved in the periphery. Thus, we tested the hypothesis
that SDV’s perceptual deficits dependent on primary
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
components of visual processing: contrast sensitivity (to
index spatial acuity), line orientation and exposure to the
crowding effect.
As with previous agnosia cases (see Table 1), SDV’s visual
acuity could not be tested with optical measures requiring
letter recognition (e.g., Snellen, 1862). We used a contrast
sensitivity task to measure visual stimulus processing at
different spatial frequencies. SDV’s performance in a grating
orientation task was generally lower than normally sighted
controls, however the contrast sensitivity function shape, i.e.,
the effect of increasing SF on orientation perception, was
similar for both groups. SDV’s main underlying deficit
appeared to be orientation perception; after correcting for
baseline orientation accuracy, contrast sensitivity profiles for
SDV and controls were identical.
Therefore, we next administered a simple line orientation
task and observed that SDV is impaired at discriminating be-
tween horizontal and vertical lines. At the fovea, his perfor-
mance was at chance, while at various sites in his periphery, he
was above chance, but significantly impaired, relative to con-
trols. This finding fits SDV’s lesion profile, where the damage is
mainly in primary visual areas with diffuse neural loss.
Sensitivity to orientation is a key feature of primary visual
cortex neurons. Excitatory and inhibitory thalamo-cortical and
intracortical connections cause neuronal populations of the
primary visual cortex to exhibit response preferences for
differently oriented visual stimuli. Orientation sensitivity is
selectively implemented at the level of neural populations
rather than at a single-cell level (Ferster & Miller, 2000).
Computational models of vision posit that line orientation is a
fundamental step for shape perception in the visual system;
orientation channels in V1 are considered “edge detectors”,
necessary to build up the “primal sketch” for object recognition
(Marr, 1976, 1982). If this level of analysis is impaired, pro-
ceeding levels of visual processing are fed with a deteriorated
input. Thus, poor line orientation perception might be the
basic mechanism explaining SDV’s perceptual deficits.
SDV was also tested for crowding. SDV is constantly
exposed to the crowding effect, since he can only process vi-
sual stimuli in the periphery of VF. When stimuli are presented
in the periphery, details of visual images are fused together,
making shape perception difficult. Central vision is uncrowded
(the “uncrowded window”; Pelli & Tillman, 2008), whereas
peripheral vision, beyond 2
, is crowded. Healthy subjects
therefore foveate interesting stimuli to process them in an
uncrowded window. Our results indicate that crowding in the
periphery did not differentially affect SDV relative to controls;
when baseline (no-flanker) performance was controlled,
crowding profiles between groups were identical. Instead SDV
was unable to perform the crowding task at the centre of the
VF. Although performance in the periphery was similar to
controls, it’s important to remember that crowding reduced
his performance to chance level. This effect does not appear in
normal subjects, who can overcome the crowding effect in the
periphery by foveating the stimuli. Thus we can conclude that
in addition to the basic processing (orientation) deficit, shape
perception deficits in SDV are further accentuated by the
crowding effect and the ineffectiveness of foveation.
This case presents a challenge for neuropsychological ac-
counts of visual agnosia. By definition, form agnosia cannot be
25
explained by low-level, sensory, visual defects, and therefore,
following our investigation, SDV should not be classified as
“agnosic”. However, most agnosic patients present VF defects,
which could, at least partially, explain their deficits in form
perception. Clinical and lesion data from a number of previous
cases are summarized in Table 1. Most patients show some VF
deficits, while in other cases, VF evaluation was not reported.
Formal testing of primary visual functions, e.g., visual acuity,
was either not reported or reported incoherently. In many
cases, lesions involved primary visual areas. In addition, most
of these lesions were caused by anoxia, likely resulting in
widespread neural loss. Thus, many of these lesions may have
had impact on primary levels of visual processing, such as
sensitivity to orientation. In addition, these deficits can be
particularly severe in cases of central vision loss, as residual
peripheral vision offers a poor analysis of spatial information
and suffers from the crowing effect.
Previous accounts have attributed agnosia to sensory def-
icits (Bender & Feldman, 1972; Campion & Latto, 1985) or a
series of multiple scotomas (Campion, 1987; Campion & Latto,
1985). Deeper analyses of primary visual functions have been
also conducted to explain a rather specific form of agnosia,
i.e., agnosia for faces e prosopagnosia e in terms of poor
contrast sensitivity at high SF (Barton, Cherkasova, Press,
Intriligator, & O’Connor, 2004) or of an inability of perceiving
curved lines (Kosslyn et al., 1995). Interestingly, in line with
the results of the present paper, Barton et al. (2004) found that
one patient, out of the seven prosopagnosic patients tested,
presented severe deficits in line orientation and curvature
perception, and also had the most difficulty with more basic-
level object recognition.
Here we propose a neurophysiologically plausible model to
explain a more general deficit of visual form perception such
as that suffered by SDV. Namely, diffuse neural damage, as
that induced by anoxia, might have reduced the number of
active neurons within the primary visual cortex, thus
affecting computational capacities necessary to process basic
features of the visual scene, such as line orientation and edge
detection. The primary visual cortex contains orientation-
selective neurons, organized in columns. Cells in neighbour-
ing columns have similar preferred orientations such that the
preferred orientation changes gradually across the surface of
the cortex (Hubel & Weisel, 1959, 1962). A reduction of
computational units in primary visual cortex, therefore, might
impact on such columnar organization, thus affecting orien-
tation discrimination and, as a consequence, the ability of
detecting edges in the visual scene. When such damage af-
fects portion of the VF with small receptive fields and high
spatial resolution, i.e., the fovea and the area immediately
surrounding, fine-grained orientation analysis is lost thus
severely affecting form processing necessary for objects, faces
and letter recognition. Such deficit has less impact on gross
figure-background segregation and global three-dimensional
organization of visual scenes, that can be supported by neu-
rons with much larger receptive fields, at the periphery of the
VF (Livingstone & Hubel, 1988). On the other hand, larger
receptive fields and peripheral vision are necessary to support
other visual functions, such as navigation and movement
perception. SDV showed spared motion perception, relatively
spared peripheral detection and a spared ability to navigate
26 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
and interact with the environment, coupled with a severe
impairment in shape perception, deficits in object, face and
letter recognition and a loss of central vision. Considering the
aetiology of SDV’s lesion and his pattern of spared and
affected visual functions, one proposition is that residual
deficits following anoxia mainly affect the number of active
neurons within the primary visual areas, thus severely
impairing orientation columns in V1, with a stronger impact
at the centre of the VF, whereas residual visual processing and
residual peripheral vision are sufficient to support navigation,
colour and motion perception.
Before concluding, it is important to note that other forms
of visual agnosia exist and they probably depend on damage
to areas other than primary visual cortex. Konen, Behrmann,
Nishimura, and Kastner (2011) recently described patient
SM, who was impaired in shape recognition. His lesion did not
involve the primary visual cortex and an fMRI investigation
showed standard retinotopic activation within the primary
visual areas. However, when objects and shapes were pre-
sented to SM, activations in the lateral occipital cortex and
infero-temporal regions were reduced compared to healthy
controls. In this case, visual agnosia can be explained as a
deficit in post-sensory, perceptual processing of shapes and
objects, due to a lesion of critical areas belonging to the
“ventral stream”, or the “what” system (Milner & Goodale,
1995). That case, however, appears to differ from SDV, and
other cases of “agnosic” patients who present VF defects and
widespread lesions to the primary visual areas. Thus, different
underlying neural mechanisms may have given impetus to
the same apparent shape perception deficit.
In conclusion, we propose that shape perception deficits
defined by the general account of visual agnosia might result
from at least two different mechanisms, one involving wide-
spread lesions of the primary visual areas, thus affecting pri-
mary levels of visual processing, and the other involving
selective lesions to lateral occipital and infero-temporal re-
gions, affecting higher-order visual functions. Therefore a
complete, neurophysiologically grounded, psychophysical
investigation of visual functions is necessary in cases of
apparent agnosia.
Acknowledgements
The authors thank: David Burr and Aurelio Bruno for discus-
sions on the case, David Murphy for his help in programming
the Crowding effect tasks, Maria Concetta Morrone and Marco
Turi for their help in setting the Contrast Sensitivity Task,
Roberto Bolzani and Mariagrazia Benassi for allowing to use
the Motion Perception task, Gaspare Galati for his help in
lesion analysis, Davide Braghittoni for neuropsychological
examination.
references
Adler, A. (1944). Disintegration and restoration of optic
recognition in visual agnosia: analysis of a case. Archives of
Neurology and Psychiatry, 51, 243e259.
Adler, A. (1950). Course and outcome of visual agnosia. Journal of
Nervous and Mental Disease, 111, 41e51.
Ashburner, J. (2007). A fast diffeomorphic image registration
algorithm. NeuroImage, 38, 95e113.
Ashburner, J., & Friston, K. J. (2000). Voxel-based
morphometryethe methods. NeuroImage, 11, 805e821.
Ashburner, J., & Friston, K. J. (2005). Unified segmentation.
NeuroImage, 26, 839e851.
Barton, J. J., Cherkasova, M. V., Press, D. Z., Intriligator, J. M., &
O’Connor, M. (2004). Perceptual functions in prosopagnosia.
Perception, 33, 939e956.
Bay, E. (1953). Disturbances of visual perception and their
examination. Brain, 76, 516e550.
Bender, M. B., & Feldman, M. (1972). The so-called “visual
agnosias”. Brain, 95, 173e186.
Benson, D. F., & Greenberg, J. P. (1969). Visual form agnosia. A
specific defect in visual discrimination. Archives of Neurology,
20(1), 82e89.
Buckner, R. L., Head, D., Parker, J., Fotenos, A. F., Marcus, D.,
Morris, J. C., et al. (2004). A unified approach for morphometric
and functional data analysis in young, old, and demented
adults using automated atlas-based head size normalization:
reliability and validation against manual measurement of
total intracranial volume. NeuroImage, 23, 724e738.
Caine, D., & Watson, J. D. (2000). Neuropsychological and
neuropathological sequelae of cerebral anoxia: a critical review.
Journal of the International Neuropsychological Society, 6, 86e99.
Campion, J. (1987). Apperceptive agnosia: the specification and
description of constructs. In G. W. Humphreys, & M. J. Riddoch
(Eds.), Visual object processing: A cognitive neuropsychological
approach. London: Lawrence Erlbaum Associates.
Campion, J., & Latto, R. (1985). Apperceptive agnosia due to
carbon monoxide poisoning: an interpretation based on
critical band masking from disseminated lesions. Behavioural
Brain Research, 15, 227e240.
Ettlinger, G. (1956). Sensory deficits in visual agnosia. Journal of
Neurology Neurosurgery Psychiatry, 19, 297e307.
Farah, M. J. (2004). Visual agnosia (2nd ed.). Cambridge: MIT Press.
Farnsworth, D. (1956). The FarnswortheMunsell 100-hue test: For the
examination of color discrimination. Macbeth.
Ferster, D., & Miller, K. D. (2000). Neural mechanisms of
orientation selectivity in the visual cortex. Annual Review of
Neuroscience, 23, 441e471.
Good, C. D., Johnsrude, I. S., Ashburner, J., Henson, R. N., Friston, K. J.,
& Frackowiak, R. S. (2001). A voxel-based morphometric study of
ageing in 465 normal adult human brains. NeuroImage, 14, 21e36.
Grossman, M., Galetta, S., & D’ Esposito, M. (1997). Object
recognition difficulty in visual apperceptive agnosia. Brain and
Cognition, 33, 306e342.
Gummel, K., Ygge, J., Benassi, M., & Bolzani, R. (2012). Motion
perception in children with foetal alcohol syndrome. Acta
Paediatrica, 101, e327ee332.
Hubel, D. H., & Weisel, T. N. (1959). Receptive fields of single neurons
in the cat’s striate cortex. Journal of Physiology, 148, 574e591.
Hubel, D. H., & Weisel, T. N. (1962 Jan). Receptive fields, binocular
interaction and functional architecture in the cat’s visual
cortex. Journal of Physiology, 160, 106e154.
Humphreys, G. W., & Riddoch, M. J. (1987a). The fractionation of
visual agnosia. In G. W. Humphreys, & M. J. Riddoch (Eds.),
Visual object processing: A cognitive neuropsychological approach
(pp. 281e306). London: Lawrence Erlbaum Associates.
Humphreys, G. W., & Riddoch, M. J. (1987b). To see but not to see: A case
study of visual agnosia. London: Lawrence Erlbaum Associates.
Kaplan, E., Goodglass, H., & Weintraub, S. (1983). Boston naming
test. Philadelphia: Lea & Febiger.
Konen, C. S., Behrmann, M., Nishimura, M., & Kastner, S. (2011).
The functional neuroanatomy of object agnosia: a case study.
Neuron, 71, 49e60.
 c o r t e x 5 2 ( 2 0 1 4 ) 1 2 e2 7
Kosslyn, S. M., Hamilton, S. E., & Bernstein, J. H. (1995). The
perception of curvature can be selectively disrupted in
prosopagnosia. Brain Cognition, 27(1), 36e58.
Ladavas, E., Shallice, T., & Zanella, M. T. (1997). Preserved
semantic access in neglect dyslexia. Neuropsychologia, 35,
257e270.
Landis, T., Graves, R., Benson, F., & Hebben, N. (1982). Visual
recognition through kinaesthetic mediation. Psychological
Medicine, 12, 515e531.
Levi, D. M., & Carney, T. (2009). Crowding in peripheral vision:
why bigger is better. Current Biology, 19, 1988e1993.
Lissauer, H. (1890). Ein Fall von Seelenblindheit, nebst einem
Beitrag zur Theorie derselben. European Archives of Psychiatry
and Clinical Neuroscience, 21, 222e270.
Livingstone, M., & Hubel, D. (1988). Segregation of form, color,
movement, and depth: anatomy, physiology, and perception.
Science, 240, 740e749.
Makela, P., Whitaker, D., & Rovamo, J. (1993). Modelling of
orientation discrimination across the visual field. Vision
Research, 33, 723e730.
Marr, D. (1976). Early processing of vision information.
Philosophical Transactions of the Royal Society of London B Biological
Sciences, 275, 483e524.
Marr, D. (1982). Vision. San Francisco: Freeman.
Marr, D., & Hildreth, E. (1980). Theory of edge detection.
Proceedings of the Royal Society of London B Biological Sciences, 207,
187e217.
Milner, A. D., & Goodale, M. A. (1995). The visual brain in action.
Oxford: Oxford University Press.
Milner, A. D., Perrett, D. I., Johnston, R. S., Benson, P. J.,
Jordan, T. R., Heeley, D. W., et al. (1991). Perception an action
in visual form agnosia. Brain, 114, 405e428.
27
Pelli, D. G., & Tillman, K. A. (2008). The uncrowded window of
object recognition. Nature Neuroscience, 11, 1129e1135.
Policardi, E., Perani, D., Zago, S., Grassi, F., Fazio, F., & Làdavas, E.
(1996). Failure to evoke visual images in a case of long-lasting
cortical blindness. Neurocase, 2, 381e394.
Riddoch, M. J., & Humphreys, G. W. (1987). A case of integrative
visual agnosia. Brain, 110, 1431e1462.
Riddoch, M. J., & Humphreys, G. W. (1993). BORB: The Birmingham
Object Recognition Battery. Hove (UK): Lawrence Erlbaum
Associates.
Ridgway, G. R., Henley, S. M., Rohrer, J. D., Scahill, R. I.,
Warren, J. D., & Fox, N. C. (2008). Ten simple rules for reporting
voxel-based morphometry studies. NeuroImage, 40, 1429e1435.
Sloan, L. L., & Habel, A. B. (1956). Tests for color deficiency based
on the pseudoisochromatic principle. A.M.A. Archives of
Ophthalmology, 55, 229e239.
Snellen, H. (1862). Optotypi ad Visum Determinandum. Utrecht.
Sparr, S. A., Jay, M., Drislane, F. W., & Venna, N. (1991). A historic
case of visual agnosia revisited after 40 years. Brain, 114,
789e800.
Steeves, J. K. E., Culham, J. C., Duchaine, B. C., Cavina Pratesi, C.,
Valyear, K. F., Schindler, I., et al. (2006). The fusiform face area
is not sufficient for face recognition: evidence from a patient
with dense prosopagnosia and no occipital face area.
Neuropsychologia, 44, 594e609.
Vecera, S. P., & Gilds, K. S. (1997). What is it like to be a patient
with apperceptive agnosia? Consciousness and Cognition, 6,
237e266.
Warrington, E. K., & James, M. (1988). Visual apperceptive agnosia:
a clinico-anatomical study of three cases. Cortex, 24, 13e32.
Zimmerman, P., & Fimm, B. (1992). Testbatterie zur
Aufmerksamkeitsprüfung (TAP). Würselen: Psytest.