Psychiatry Research 178 (2010) 270275

Contents lists available at ScienceDirect

Psychiatry Research
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p s yc h r e s

Visual processing, social cognition and functional outcome in schizophrenia

Philip Brittain a,, Dominic H. ffytche b, Allison McKendrick c, Simon Surguladze a
a
b
c

Affective Neuroscience Group, Section of Cognitive Neuropsychiatry, King's College London, Institute of Psychiatry, 16 De Crespigny Park, London SE5 8AF, UK
Section of Old Age Psychiatry and Centre for Neuroimaging Sciences, King's College London, Institute of Psychiatry, 16 De Crespigny Park, London SE5 8AF, UK
Department of Optometry & Vision Sciences, The University of Melbourne, Parkville, 3010, Australia

a r t i c l e

i n f o

Article history:
Received 16 July 2009
Received in revised form 8 September 2009
Accepted 22 September 2009
Keywords:
Contrast sensitivity
Visual masking
Global motion
Biological motion
Social perception
Functional status

a b s t r a c t
Visual processing decits are well recognised in schizophrenia and have potentially important clinical
implications. First, the pattern of decits for different visual tasks may help understand the underlying
pathophysiology of the visual dysfunction. Second, several studies report decits correlating with functional
outcomes, suggesting that outcome improvement is possible through visual remediation strategies. We
investigated these issues in a group of 64 schizophrenia patients and matched controls with a battery of
visual tasks targeting different points along the visual pathways and by examining direct and indirect
relationships (via a potential mediator) of such decits to functional outcome. The schizophrenia group was
signicantly worse on the visual tasks overall, with the decit constant for low- and high-level processing.
Zero-order correlations suggested minimal association between vision and outcome, however, correlations
between three visual tasks and social perceptual ability were found which in turn correlated with functional
outcome; path analysis conrmed a signicant but small and indirect effect of biological motion processing
ability on functional outcome mediated by social perception. In conclusion, the pathophysiology of visual
dysfunction affects low- and high-level visual areas similarly and the relationship between decits and
outcome is small and indirect.
2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction
A large body of research indicates that individuals with schizophrenia experience visual decits (see Butler et al., 2005). The decits
are found in relation to a variety of tasks, particularly those related to
the magnocellular/dorsal stream, a pathway from retina to visual
cortex and beyond linked to visual motion processing and conveying
signals related to low-spatial frequencies (large scale visual detail),
low contrast, and high temporal frequencies. The parvocellular/ventral system, which conveys color and high spatial frequency information, appears relatively spared. The decits are of potential clinical
interest as they may provide clues as to the underlying pathophysiology of the visual dysfunction; different visual perceptual abilities
are resolved at varying points along the visual pathways so that the
overall pattern of decits may point to particular cortical locations
and processes. Most previous studies have focussed on low-level
processing, i.e., those resolved early-on in the visual pathways,
although higher-level visual decits have also been reported,
revealing problems in cortical-level visual processing. Thus, decits
have been found for abilities such as luminance-icker sensitivity
(Slaghuis and Bishop, 2001) static contrast sensitivity (Keri et al.,
2002) and visual masking (Rassovsky et al., 2005) that involve early,

Corresponding author. Tel.: +44 207848 5228.

E-mail address: Philip.Brittain@kcl.ac.uk (P. Brittain).
0165-1781/$ see front matter 2009 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.psychres.2009.09.013

low-level visual processing and higher-level decits have been

reported for tasks such as global motion (Chen et al., 2003), requiring
the integration of information across the visual eld, and biological
motion, requiring the extraction of walking gure contours (Kim
et al., 2005). No studies, however, have examined a range of low- and
high-level tasks within the same subjects and testing session. If visual
decits increase in magnitude from low- to high-levels of processing,
as has recently been suggested (Butler et al., 2005), this would point
to a mechanism affecting each level of the visual hierarchy or a lowlevel decit amplied by transmission through the increasingly
specialised components of the visual pathways. Alternatively, if the
decit remains constant across each level of the visual hierarchy, this
would suggest a mechanism primarily affecting the earlier stages of
processing.
A second potential area of clinical interest is the relation of visual
decits to functional outcomes, with a number of studies reporting
signicant correlations between these variables in schizophrenia. This
raises the possibility that perceptual training techniques may improve
the poor functional outcomes often seen in schizophrenia. However,
the potential of this approach is unclear, as the few reports available
disagree as to how strong the correlation might be. For example, Kim
et al. (2005) found that 50% of the variance in a functional outcome
measure was accounted for by perceptual ability (r = 0.71), Butler et al.
(2005) reported gures of between 14 and 25% (r = 0.370.50) and
Sergi et al. (2006) reported lower gures still of between <1% and 11%
(r = 0.03 to r = 0.33). Other studies have measured the relationship

P. Brittain et al. / Psychiatry Research 178 (2010) 270275

between functional outcome and visual measures within large-scale

neurocognitive batteries. Some of these studies have reported signicant correlations (Bowen et al., 1994) whilst others have not (Vauth
et al., 2004). Furthermore, such studies tend to use tasks that, although
presented visually, involve other processes such as sustained attention,
e.g., the degraded-stimulus continuous performance task.
The variation in reported correlation strengths between visual
processing and functional outcome might reect the fact that different
studies have probed different levels of the visual system. Tasks
probing low-level visual processing may be less directly linked to
functional outcome than tasks probing higher levels. This might
explain why studies such as Kim et al. (2005) which used a high-level,
biological motion stimulus, found a strong association whilst studies
such as Sergi et al. (2006) using a low-level visual masking paradigm
found lower correlations. Another explanation might be the effect of
certain high-level visual tasks such as biological motion perception
inadvertently tapping into another, unmeasured factor, related to
outcome, social cognition; thereby inating the apparent strength of
relation between vision and function. Social cognition includes varied
domains such as emotion perception, social perception and social
knowledge (Green et al., 2005a), and has been proposed to act as an
intermediary variable between several classes of neurocognition
(including vision) and functional outcome (Brekke et al., 2005). This
theory is based in part on the link between neurocognition and social
cognition (Bozikas et al., 2004) and between social cognition and
functional outcome (Penn et al., 2002) and it seems plausible that if
visual decits produce a relative inability to pick up socially important
visual cues, i.e., micro-expressions or subtle gestures, this could be a
path through which visual decits impact on functional outcome in
schizophrenia. One study thus far has found evidence for this causal
chain (Sergi et al., 2006).
Here, we set out to address the above issues. First, we have studied
the pattern of visual decits in individuals with schizophrenia compared to controls across different levels of visual processing in tasks
which, at least partially, recruit the magnocellular/dorsal stream.
Second, we have studied the relationship between vision and functional outcome in schizophrenia and attempted to address the
question of why previous studies have found varying strengths of
association. A battery of visual tests probing different levels of the
visual hierarchy was undertaken by a cohort of patients and control
subjects together with measures of social perception and, in the
patient group, functional outcome.
2. Methods
2.1. Subjects
Sixty-four persons with a Diagnostic and Statistical Manual of Mental Disorders,
fourth edition (DSM-IV) diagnosis of schizophrenia and 65 control subjects were
recruited. All but ve of the patients were taking antipsychotic medication (50 atypical,
seven typical, two mixed). The patient group were recruited from outpatient and longterm assisted living settings in South London. Diagnosis was conrmed by their treating
clinician, chart review and the psychotic symptoms subsection of the Structured Clinical
Interview for DSM-IV (SCID) Patient Edition (First et al., 2002b). Control subjects with
no self-reported history of psychiatric illness were recruited from the community. The
psychotic screening subsection from the SCID Non-patient edition (First et al., 2002a)
was used to check for lifetime presence of psychotic illness. Potential control subjects
were excluded if any of their rst-degree relatives had a history of psychotic illness.
Exclusion criteria applied to both groups included: current drug or alcohol dependency,
a reading or sensory disability, an identiable neurological condition and having bestcorrected visual acuity below 0.80 decimal, measured using the Freiburg Visual Acuity
test (Bach, 2007). The experimental protocol was approved by the Institute of Psychiatry
ethics committee. Subjects gave written consent and were paid for their participation.

271

85 Hz). The CS task was run from a Mac Tower. The GM and BM tasks were run from a
Centrex PC through a VSG 2/5 graphics card. The MASK and HP tasks were run from a
Dell laptop. Subjects gave their responses by speaking aloud and the test administrator
entered responses on the computer. All tasks were viewed binocularly and were
presented in a random order with the exception of the contrast sensitivity task which
was performed during an initial visual acuity screening session. Viewing distance for
the MASK, GM, BM and HP tasks was 100 cm, and for the visual tasks was kept constant
with a chin rest. The CS task viewing distance was 200 cm and no chin rest was used. All
tasks were forced choice procedures and were preceded by a practise period.
2.2.1. Visual tasks
We focussed on tasks with sensitivity to magnocellular/dorsal stream function,
given the previous evidence of this channels dysfunction in schizophrenia (Butler and
Javitt, 2005). Each section below initially details the aspect of the magnocellular/dorsal
stream targeted by the visual task. Tests of low-level visual processing included contrast detection and visual masking. Intermediate-level visual processing was examined
using a test of global motion perception. The highest level of visual processing was
assessed using a test of biological motion perception.
2.2.1.1. Contrast sensitivity (CS). Magnocellular neurons have lower contrast thresholds
(Livingstone and Hubel, 1988) and preferentially activate to low-spatial frequency
stimuli below about 1.5 cycles/ (Legge, 1978). The Freiburg Contrast Test version 5.6.1
(Bach, 2007) was used to assess contrast detection thresholds. A Landolt C optotype is
used where the gap of the C can have one of four orientations (Fig. 1). Subjects were
required to name the orientation (up, down, left, right). The gure subtended 5.7 of
visual angle and the gap had an equivalent spatial frequency of 0.5 cycles/. Threshold is
estimated with a Best PEST (parameter estimation by sequential testing) algorithm
and an adaptive-staircase procedure. The contrast sensitivity result is the logarithmized inverse of the threshold contrast, dened with the Michelson contrast (Lmax
Lmin)/(Lmax + Lmin). There were 28 trials. Higher values indicated better sensitivity
to contrast.
2.2.1.2. Visual masking (MASK). Magnocellular neurons are sensitive to transient stimuli
(Breitmeyer and Julesz, 1975) and have short latencies (Breitmeyer, 1975). Furthermore, the location masking task used here is thought to further increase magnocellular
involvement (Cadenhead et al., 1998). The masking procedures and stimuli (Fig. 2)
were those used by Green et al. (2002). The target was a square with a small gap in one
of its sides (the gap was irrelevant for the version of the task used here). The square
could appear in any one of four locations on the screen (upper left, upper right, lower
left, lower right). For each trial, the subjects were asked to state the location of the
target. Each target subtended 0.27 of visual angle and was located 1.03 of visual angle
from the xation cross, which was presented 400 ms before each target presentation
for 300 ms. The target's contrast was set for each subject with a thresholding procedure.
This produces a grey scale value called the critical stimulus intensity at which all
subjects can see the unmasked target on approximately 84% of trials. The mask
consisted of a 4 4 array of adjacent boxes (a 2 2 array in each quadrant) that
appeared together in the same spatial locations as the target could appear, i.e., target
and mask were superimposed in one of the quadrants. The duration of the target was
12.5 ms and the duration of the mask was 25 ms. The stimulus onset asynchrony
(SOA) is the time between the onset of the stimulus and the onset of the mask. Twelve
SOAs ranging from 75 (forward masking) to 75 ms (backward masking) were
presented in randomized fashion, 12 trials were presented for each SOA and the four
possible locations of the target were counterbalanced (an SOA of 0 where mask and
target are displayed simultaneously was also presented but was not used in the
analysis). Percentage correct rates for each of the six backward and six forward making
SOAs were averaged and used as the masking performance score, as analysis revealed
there was no differential performance between groups on the forward versus backward
trials. Higher values indicated better performance. A no-mask condition was randomly
interspersed within the masking trials and the observed equal performance between
groups suggested that attentional problems in the patient group would not account for
any decit on this task.
2.2.1.3. Global motion (GM). Global motion is determined in cortical area V5/MT
(Newsome and Par, 1988), a dorsal stream area (Ungerleider and Haxby, 1994). The
stimulus was a 10 of visual angle circular area containing 100 moving dots. Each dot
was 8.5 min arc in diameter. A percentage of the dots moved upwards or downwards
(signal dots) at 2.86/s, whilst the remaining dots (noise dots) moved in random

2.2. Stimuli and procedure

Tasks were presented on a 21 inch gamma corrected ViewSonic G220f monitor in a
dimly lit room (1012 lx). The monitor's mean luminance was 87.7 cd/m2. The
following resolutions and frame rates were applied: contrast sensitivity (CS; 1024 768,
85 Hz), visual masking (MASK; 800 600, 160 Hz), global motion (GM; 800 600,
100 Hz), biological motion (BM; 1024 768, 120 Hz) and Half-PONS (HP; 1024 768,

Fig. 1. Illustration of Landolt C gures used in contrast sensitivity testing. The

orientation of the gap varies on each trial. The luminance contrast of the gure against
the background varies until threshold is approximated.

272

P. Brittain et al. / Psychiatry Research 178 (2010) 270275

Fig. 2. Illustration of the visual masking task with xation cross in the centre. Three
different screen shots are shown from left to right. The target (far left) is a square with
a small gap in one of the sides. The mask (middle) is four composite squares made of
four smaller squares. The two appear separately, although so close together in time that
the presence of the mask disrupts the perception of the target, usually resulting in a
fused image (far right).

directions (Fig. 3). For each trial, the direction of the signal dots (up or down) was
chosen at random. The duration of each trial was 400 ms. Subjects were required to
name the direction of the signal dots. The ratio of signal dots began at 100% and used a
3-down, 1-up staircase procedure which terminated after four reversals. Two staircases
were interleaved and ran simultaneously. Correct identication by the subject resulted
in the signal strength being degraded by replacing signal dots with noise dots. An
incorrect response increased the number of signal dots. Different dots are chosen to be
signal or noise on each frame in order to limit the availability of local motion cues.
Hence, the observer needs to integrate the motion cues across the movie sequence to
determine the signal direction of motion and cannot determine it by tracking individual
dots. The initial step-size was 8% which halved after the rst two reversals. The mean
percentage of signal dots at the last four reversals (two from each staircase) was taken
as the global motion perception threshold. This was transformed into a sensitivity
measure with the formula: 1/threshold, so that like the other visual tasks, higher values
indicated better performance.
2.2.1.4. Biological motion (BM). Biological motion perception is dependent on downstream areas of the dorsal stream, but may also be subserved by ventral stream
components (Vaina et al., 2001). An array of randomly moving dots appeared within a
10 of visual angle square area. Each dot was 6.8 min arc in diameter. The initial number
of dots was 50. Each presentation lasted 3500 ms. After 2000 ms, 12 signal dots moved
in such a way as to represent a human body (Fig. 4) walking, either to the left or to the
right. The gure moved at a velocity of 4.5/s. A 1-down, 1-up staircase procedure was
used and continued for 42 trials. The design was a three-alternative procedure whereby
the subject responded that the gure moved to the left, to the right, or that the subject
had not seen the gure. If the subject responded correctly, the number of randomly
moving noise dots in the next trial increased by 20, reducing to 10 after the rst
reversal. If the response was incorrect or didn't see the number of noise dots in the
next trial decreased by 10. There was no signicant difference in the number of didn't
see responses in the patient and control groups (P > 0.05). The mean number of dots in
trials 3242 was used as the biological motion perception threshold. Pilot work
indicated that this was a sufcient number of trials to reach a performance plateau.
Higher values indicated better biological motion perception.
2.2.2. Social cognitionsocial perception (HP)
The Half-Prole of Nonverbal Sensitivity (Ambady et al., 1995), which consists of
the rst 110 scenes of the Prole of Nonverbal Sensitivity (Rosenthal et al., 1979) was
used to assess social perception, a type of social cognition (Green et al., 2005a). The
Half-PONS consists of a series of 2-s video clips containing the facial expressions, voice
intonations and/or bodily gestures of a Caucasian female acting a variety of social
situations. Each scene contains either one, two or three of these social cues. Before each
clip the subject was presented with a card detailing two possible answers (e.g.,
ordering food in a restaurant or threatening someone), one of which was correct.
Immediately afterwards the subject was required to choose which description best
described the presented clip. Higher scores indicated better social perception.

Fig. 3. Schematic illustration of the global motion dot stimuli. Dots (shown as arrows
indicating their direction of movement) either move in a common direction (signal
direction down or up, down in this example) or move randomly. Three different levels
of signal-to-noise ratio are shown here; 100% coherence (left), 75% coherence (middle),
50% coherence (right).

Fig. 4. Schematic illustration of a point-light walker. Lines joining dots are for
illustration only. When the gure dots move the percept of a human walking is
apparent. Increasing numbers of moving distracter dots (middle and right) eventually
make the human gure imperceptible.

2.2.3. Functional outcome rating scale (RFS)

Functional outcome in the patient group only was assessed with the Role
Functioning Scale (Goodman et al., 1993) which assesses functional status in four
domains (see Table 1) on subscales ranging from 1 (severely impaired) to 7 (optimal)
based on a semi-structured interview. The four scores can be totalled to create a Global
Role Functioning Index (GFI) with scores ranging from 4 to 28. A higher score indicates
better functioning. Raters were trained on the RFS to meet minimum intraclass
correlation coefcients for total ratings of 0.70 (P < 0.05), based on agreements with the
ratings of an Institute of Psychiatry expert diagnostician.
2.3. Statistical analysis
Raw data was transformed where necessary to meet normality assumptions.
To allow comparison between tasks, data was converted into z-scores based on the
means and standard deviations of the control group. All patient z-scores are negative
(indicating a decit away from the control group). A z-score of 1 indicates a decrease
in performance equivalent to one standard deviation of the mean control groups scores
on that task. All analysis was undertaken with these z-scores. The P-value required for
classication of statistical signicance was 0.05 unless stated otherwise. Some subjects
were unable to complete all of the battery for a variety of reasons including fatigue or
the inability to reach a minimum threshold for the visual masking target. This resulted
in 2% of test data missing and for the analysis of variance (ANOVA) and path analysis,
list-wise deletion would have resulted in omitting 16% of subjects, therefore missing
values were replaced using an expectationmaximisation method. Path analysis with
AMOS 7.0 used a Bootstrap approximation with 1000 iterations, producing direct,
indirect (via another variable) and total (direct + indirect) affects of variables in the
model together with their associated P-values and 95% condence intervals.

Table 1
Mean and standard deviation subject characteristics for the patient and control groups.

Age
Gender
Education in years
Visual acuitya
WASI IQ scoreb
Handedness (L/R/Ambi)
PANSSc Positive
PANSS Negative
CPZ equivalentd
Illness duration in years
Role Functioning Scale subscales and total
Working productivity
Independent living, self care
Immediate social network relationships
Extended social network relationships
Global functioning index (GFI)
a

Patient group
(N = 64)

Control group
(N = 65)

41.89 (11.05)
M = 34
14.20 (3.18)
1.31 (0.22)
101.91 (15.24)
(0/60/4)
13.52 (4.51)
14.36 (5.26)
461.95 (381.61)
18.41 (10.79)

41.29 (9.57)
M = 35
14.90 (3.14)
1.33 (0.22)
107.37 (13.49)
(0/62/3)

3.14 (1.76)
4.8 (1.95)
5.11 (1.24)
4.66 (1.36)
17.70 (5.18)

Decimal scale. 1.0 = 6:6 acuity.

Wechsler Abbreviated Scale of Intelligence (2 subtest version).
Positive and Negative Syndrome Scale.
d
Chlorpromazine dose equivalent (mg/day).
P < 0.05.
b
c

P. Brittain et al. / Psychiatry Research 178 (2010) 270275

273

Table 2
Data from the four visual tasks (CS, MASK, GM, BM; putatively ordered from low-level to high-level visual processing) and the social perception task, the half-PONS (HP).
Task

Control group mean raw scores (S.D.)

Patient group mean raw scores (S.D.)

Patient group mean z-scores (S.D.)

P values

Contrast Sensitivity in Log units (CS)

Masking (MASK)
Global motion (GM)
Biological motion (BM)
Half-PONS (HP)

2.24
53.32
0.068
211.95
84.57

2.20
48.04
0.059
186.25
79.74

0.31
0.37
0.32
0.41
0.97

0.086
0.048
0.076
0.020
<0.000

(0.13)
(14.27)
(0.029)
(62.98)
(4.68)

(0.14)
(15.78)
(0.028)
(60.56)
(7.55)

(1.10)
(1.11)
(0.97)
(0.96)
(1.42)

Mean and standard deviation raw scores for the control group and patient group. z-scores are for the patient group only. A negative z-score of 1 indicates a decrease in performance
of one standard deviation of the mean control group scores on that task. P-values are equivalent for raw scores and z-scores.

3. Results
Subject characteristics and the functional outcome measurements
are reported in Table 1. The patient group had a signicantly lower
mean IQ than the control group.
3.1. Task differences between groups
The mean raw scores for each task are presented in Table 2
together with the z-transformed data. The schizophrenia group performed worse on the visual tasks but impairments were greatest on
the social perception task.
To further examine the visual decits in schizophrenia we performed
a mixed model ANOVA with a two-level between group factor (patient
or control) and a four-level within group factor (visual task; CS, MASK,
GM, BM). The main between-group was signicant (F (1,127) = 11.239,
P = 0.001, partial eta2 = 0.081), indicating an overall worse performance by the patient group on the visual tasks. This difference remained
signicant after co-varying for IQ. The task by group interaction was not
signicant, suggesting the decit did not vary across the visual hierarchy
(GreenhouseGeisser corrected F (3,381) = 0.071, P = 0.968, partial
eta2 = 0.001); the interaction effect being equivalent to the main
within-subjects effect of task type due to the control z-scores being
based on their own means and standard deviations. We then compared
the patients z-scores on the CS task and the BM task (the tasks
producing the least and the greatest decit) using a paired t-test. This
was not signicant (t = 0.476, df = 63, P = 0.636) indicating that,
although the z-scores and post hoc t-tests (Table 2) suggest a greater
decit at the highest compared to the lowest levels of the visual system,
the increase in decit is not statistically signicant. These results suggest
the visual decit in schizophrenia is constant across the visual
processing hierarchy for the tests used here.
3.2. Visual performance, social perception and functional outcome
Table 3 shows the zero-order correlation matrix for the visual
tasks, social perception and functional outcome measures in the

schizophrenia group. Of the 20 combinations of visual performance

and functional outcome subscales and totals examined, only 1 showed
a signicant correlation (better performance associated with better
functional outcome).
Three visual tasks (MASK, GM and BM) were signicantly
correlated with the HP (better visual performance associated with
better social perception). The MASK and GM tasks had similar
strength relationships with the HP task but the correlation coefcients
of CS/HP and BM/HP were signicantly different (z = 2.06, N = 64,
P = 0.04) suggesting that lower-levels of visual processing are less
associated than higher levels with social perception. Social perception
was signicantly correlated with three of the four RFS subscales
(better social perception associated with better function) and the GFI.
Neither positive nor negative symptom scores signicantly correlated
with any visual tasks nor the HP. Both symptom scales correlated with
the GFI, however (Negative/GFI R = 0.59, P < 0.000; Positive/GFI =
0.30, P = 0.017).
We explored possible indirect pathways between vision and
functional outcome with path analysis to reveal associations otherwise obscured in zero-order correlations. A composite score from the
three RFS subscales signicantly associated with social perception
was created for the analysis. Only the three visual tasks correlating
with social perception were included (MASK, GM and BM). None of
these visual measures had signicant zero-order associations with the
new composite functional measure. The initial model allowed for all
possible direct and indirect (via social perception) paths between the
three visual tasks and the functional composite score. MASK and GM
were removed from the nal model because of no signicant paths
either to HP or to the RFS composite. There were no signicant direct
effects of BM on outcome. Nor were there any signicant total effects
(indirect + direct, equivalent to the zero-order correlation). Noticeably, there remained a signicant indirect path from BM to the
outcome composite (Fig. 5). When BM z-scores increased by one
standard deviation, the RFS composite scores improved by 0.16 of a
standard deviation respectively. The indirect path remained signicant when positive or negative symptoms were added to the model.
Five commonly used model t indices were used to test the

Table 3
Pearson correlation coefcients between visual tasks, social perception and functional outcome measures for the schizophrenia group P < 0.05, P < 0.01.

Visual tasks
1. Contrast sensitivity (CS)
2. Masking (MASK)
3. Global motion (GM)
4. Biological motion (BM)
Social perception
5. Half-PONS (HP)
Role Functioning Scale (RFS)
6. Working productivity
7. Independent living, self care
8. Immediate social relationships
9. Extended social relationships
10. Global functioning index (GFI)

0.09
0.08
0.21

0.32
0.42

0.47

0.10

0.36

0.36

0.08
0.09
0.24
0.04
0.11

0.11
0.11
0.36
0.11
0.19

0.06
0.01
0.04
0.07
0.05

0.20
0.31
0.32
0.32
0.34

0.58
0.49
0.52
0.81

0.54
0.61
0.86

0.62
0.77

0.82

0.44

0.12
0.13
0.03
0.02
0.09

274

P. Brittain et al. / Psychiatry Research 178 (2010) 270275

Fig. 5. Path model showing signicant direct (solid lines) and indirect (dashed lines)
paths between visual processing, social perception and functional status in schizophrenia. Standardised path coefcients (beta) are noted above each path. The
coefcient for the indirect path refers to both segments, i.e., visionsocial perception
and social perception to function. P < 0.05, P < 0.01.

robustness of the model. Values suggested the model t the data

well: X2 = 0.799; df = 1 P = 0.377; RMSEA = 0.000; AGFI = 0.951;
NFI =0.967, CFI = 1.000.
In the nal model 19.5% (standardised beta = 0.44) of the variance
in social perception is explained by biological motion and 13.5%
(standardised beta = 0.37) of the variance in functional status is
explained by social perception. The only signicant visual processing
predictor found, indirectly accounts for just 3% (standardised
beta = 0.16) of the variance in the functional composite.
4. Discussion
Our study examined visual processing decits in schizophrenia
and their relationship to functional outcome. First, we investigated
whether visual decits in schizophrenia accumulated or remained
constant from low- to high-levels of visual processing. Second, we
investigated the association between vision and outcome and
considered why the strength of any relationship might have varied
widely in previous reports. For this second question we focussed on
two hypotheses, one related to the level of visual processing
measured, the other to the role of a third variable, social cognition.
These issues will be discussed in turn.
With regard to the pattern of decits across the visual hierarchy,
both groups were able to see the stimuli equivalently in terms of
visual acuity. Attentional problems were also unlikely to have caused
the recorded decits. As in previous studies, groups differed in terms
of visual perceptual processing ability, here, in tasks tapping different
levels of the visual hierarchy and involving the magnocellular/dorsal
stream. The z-scores showed that even on the task producing the
largest decit (biological motion), mean performance in the patient
group fell within 0.5 S.D. of the control group's performance and only
two of the tasks individually showed a statistically signicant decit
with post hoc testing. It is not clear why the CS and GM tasks
individually did not produce signicant group differences as found in
many previous studies. However, considering the trend-level performance difference on these two tasks and the consistent overall decit,
it may be that the differing psychometric properties of the tests (e.g.,
the different thresholding techniques used) determined whether the
decit reached signicance or not. As well, with regard to the CS test,
another study (Keri et al., 2002) showed that temporal modulation at
this spatial frequency can be necessary to observe a signicant decit.
The visual decits, equivalent to Cohen's d effect sizes of between
0.30 and 0.42, suggest a modest impairment when individual
visual abilities are considered. However, the main between-group
effect from the ANOVA which encompasses all visual tasks produces
an eta2 value of 0.081, equivalent to a Cohen's d of nearly 0.6, thus, the
visual decit in total should be considered as more substantial than
that suggested by the individual tasks. Taken as a whole, our results
are in general agreement with previous ndings of visual dysfunction
in schizophrenia.
Although the z-scores suggest that decits worsen from the lowest
to the highest level of processing, the interaction of group by task on
performance was not signicant and the difference in patient
performance between the CS and the BM tasks was not signicant.
These results indicate a perceptual dysfunction beginning with early-

visual areas with the faulty signal propagating from one level of the
visual hierarchy to the next, a slight but non-signicant accumulation or amplication occurring between the lowest and highest
levels. In fact, performance was worse on the masking task than on
the global motion task even though the former is most often
described as being an early-visual task (Sergi and Green, 2003)
whereas the latter is thought to be resolved cortically in area V5/MT
(Newsome and Par, 1988). However, recent neuroimaging data
(Green et al., 2005b) suggests that masking may recruit subcortical
thalamic regions and several down-stream cortical visual areas
including the lateral occipital area. As such, masking may recruit a
broader array of processing capacities and be more sensitive to any
underlying decit.
Regarding the relationship between the visual processing measures and the functional outcome measure, the strength of the zeroorder correlations was weak in this patient group. With the exception
of 1 association from 20, they did not replicate the levels of correlation
seen in some earlier studies (Butler et al., 2005; Kim et al., 2005). A
subsequent path analysis revealed that the ability of patients on one of
the visual tasks, biological motion perception, had a signicant and
positive but indirect effect, via social perceptual ability, on a
functional outcome composite measure (better visual performance
associated with better functional outcome).
Our study design allowed us to consider two possible mechanisms
which might underlie the differences found in previous studies
reporting zero-order visual processing and functional outcome
associations. The rst of these was the impact of the visual processing
level studied. The zero-order correlations between vision and
outcome here, whilst mostly non-signicant, did not suggest an
increase in the strength of association for higher- compared to lowerlevels of the visual hierarchy. However, the path analysis suggests that
an increase in association strength for higher levels of the hierarchy
might be found within the indirect pathway. The fact that the highlevel BM task was retained in the path model, whilst the lower-level
MASK and GM tasks were removed, suggests that any signicant
indirect path from vision to outcome runs from the highest level of
visual processing. Also, when the model was re-run with only either
the MASK or GM tasks, indirect path beta weights were 0.13
compared to 0.16 for the BM task. Effectively, the MASK and GM
tasks add no increased predictive ability over the BM task. This was
also reected in the greater association between the BM and HP tasks
than between the CS and HP tasks. For indirect paths, it appears the
level of processing may inuence the strength of association between
vision and outcome and it is possible that this effect contributes to the
variation in reported zero-order correlations found in previous
studies.
The second mechanism was the role of a potential mediating
variable; social cognition. In our study this was assessed with a social
perception task. We postulated that some high-level visual tasks,
such as the biological motion task used by Kim et al. (2005), may tap
in to social cognitive processes and thus inate the apparent
association between vision and functional outcome through the
association between social cognition and the latter. Our study
supports this. The Kim et al. (2005) study, which found a very strong
association between vision and function (r = 0.71), used a biological
motion task which involved potentially socially relevant motor acts
such as kicking, throwing and climbing. Our version of this task
involved just a walking gure. The indirect path revealed in our
analysis suggests that biological motion perception is signicantly
associated with functional outcome via social cognition but that the
zero-order correlation between a more pure measure of biological
motion perception and outcome is non-signicant. These facts suggest
that future studies investigating these associations should ensure any
biological motion measures have as little socially relevant content as
possible to avoid the association with outcome conating with social
cognitive processes.

P. Brittain et al. / Psychiatry Research 178 (2010) 270275

The nding of a signicant indirect path from visual processing

ability, via social perception, to functional outcome, partially supports
the one other previous study specically investigating these variables
in schizophrenia (Sergi et al., 2006). Both studies did not nd many
signicant zero-order associations between vision and outcome.
However, those authors did report a zero-order correlation between
a composite visual factor and a composite functional outcome factor
whereas we report no such zero-order association. In agreement
with that study, we refer to social perception as mediating the effects
of vision on outcome, because, it is important to note, that such
mediating effects may be present even without a zero-order association between two variables (MacKinnon et al., 2000). The indirect
path here is obscured within the zero-order correlation between the
same two variables by the totality of all other unexplored paths
between visual processing and functional outcome.
Our study had several methodological weaknesses. First, although
we choose tasks which involved magnocellular activity, isolating this
channel alone is problematic and the degree to which parvocellular
activity was involved cannot be determined. However, the focus of
this study was on low- versus high-level visual processing and our
aim was not to isolate the magnocellular/dorsal stream, but merely to
partially recruit it to maximise the chances of nding a decit. Second,
the regions critical for determination of each of our visual tasks are
well dened for global motion (Newsome and Par, 1988) and
biological motion (Howard et al., 1996), but less so for contrast
sensitivity and visual masking, further, the nature of potential crosstalk between these areas is poorly understood. Therefore, the ranking
of tasks into low- versus high-level processing is not perfect; actual
performance probably reecting activity in several areas. Third, each
visual task had different psychometric properties (e.g., the different
thresholding techniques used) so that performance across tasks may
not be directly comparable. Fourth, social cognition and functional
outcome were both assessed with single, relatively crude measures. It
is possible that better measurement of these constructs would
identify effects not apparent in our data. Finally, functional status
was not assessed in the control group, thus, we do not know if
associations between vision and outcome are equivalent for both
groups.
In summary, rst, the size of the visual processing decit in
schizophrenia is modest but consistent regardless of where within the
visual pathways it is measured. This indicates a perceptual dysfunction
beginning with early-visual areas which propagates through the visual
hierarchy. Second, the association between visual processing decits
and functional outcomes in schizophrenia is small and indirect. This
suggests that improvements in functional outcomes achievable
though remediation of visual perceptual dysfunction would need to
act via mediating variables, such as social cognition. In a similar regard,
another study (Fisher et al., 2009) recently demonstrated that earlysensory remediation in the auditory domain can improve performance
in higher-order cognitive processes relevant to psychosocial functioning in schizophrenia. However, focussing on the remediation of social
cognitive processes directly, such as social or emotional perception,
may yield more benets in functional outcome.
Acknowledgements
The authors would like to thank Professor Michael Green of UCLA and Dr Stahl from
the Institute of Psychiatry. This work was supported by the Psychiatry Research Trust
and the Institute of Psychiatry Psychological Medicine sub-committee.

References
Ambady, N., Hallahan, M., Rosenthal, R., 1995. On judging and being judged accurately
in zero-acquaintance situations. Journal of Personality and Social Psychology 69,
518529.
Bach, M., 2007. The Freiburg Visual Acuity Testvariability unchanged by post-hoc reanalysis. Graefes Archive for Clinical and Experimental Ophthalmology 24, 965971.

275

Bowen, L., Wallace, C.J., Glynn, S.M., Nuechterlein, K.H., Lutzker, J.R., Kuehnel, T.G., 1994.
Schizophrenic individuals' cognitive functioning and performance in interpersonal
interactions and skills training procedures. Journal of Psychiatric Research 28,
289301.
Bozikas, V.P., Kosmidis, M.H., Anezoulake, D., Giannakou, M., Karavatos, A., 2004.
Relationship of affect recognition with psychopathology and cognitive performance in schizophrenia. Journal of the International Neuropsychological Society
10, 549558.
Breitmeyer, B.G., 1975. Simple reaction time as a measure of the temporal response
properties of transient and sustained channels. Vision Research 15, 14111412.
Breitmeyer, B.G., Julesz, B., 1975. The role of on and off transients in determining the
psychophysical spatial frequency response. Vision Research 16, 411415.
Brekke, J., Kay, D.D., Lee, K.S., Green, M.F., 2005. Biosocial pathways to functional
outcome in schizophrenia. Schizophrenia Research 80, 213225.
Butler, P.D., Javitt, D.C., 2005. Early-stage visual processing decits in schizophrenia.
Current opinion in psychiatry 18, 151157.
Butler, P.D., Zemon, V., Schechter, I., Saperstein, A.M., Hoptman, M.J., Lim, K.O., Revheim,
N., Silipo, G., Javitt, D.C., 2005. Early-stage visual processing and cortical amplication decits in schizophrenia. Archives of General Psychiatry 62, 495504.
Cadenhead, K.S., Serper, Y., Braff, D.L., 1998. Transient versus sustained visual channels
in the visual backward masking decits of schizophrenia patients. Biological
Psychiatry 43, 132138.
Chen, Y.N., Levy, D., Matthysse, S., Holzman, P., 2003. Processing of global, but not local,
motion direction is decient in schizophrenia. Schizophrenia Research 61, 215227.
First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 2002a. Structured Clinical Interview
for DSM-IV-TR Axis I Disorders, Research Version, Non-patient Edition. (SCID-I/NP).
First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 2002b. Structured Clinical Interview
for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition. (SCID-I/P).
Fisher, M., Holland, C., Merzenich, M.M., Vinogradov, S., 2009. Using neuroplasticitybased auditory training to improve verbal memory in schizophrenia. American
Journal of Psychiatry 166, 805811.
Goodman, S.H., Sewell, D.R., Cooley, E.L., Leavitt, N., 1993. Assessing levels of adaptive
functioning: the Role Functioning Scale. Community Mental Health Journal 29,
119131.
Green, M.F., Nuechterlein, K.H., Breitmeyer, B., 2002. Development of a computerized
assessment for visual masking. International Journal of Methods in Psychiatric
Research 11, 8389.
Green, M.F., Olivier, B., Crawley, J.N., Penn, D.L., Silverstein, S., 2005a. Social cognition in
schizophrenia: recommendations from the measurement and treatment research
to improve cognition in schizophrenia new approaches conference. Schizophrenia
Bulletin 31, 882887.
Green, M.F., Glahn, D., Engel, S.A., Nuechterlein, K.H., Sabb, F., Strojwas, M., Cohen, M.S.,
2005b. Regional brain activity associated with visual backward masking. Journal of
Cognitive Neuroscience 17, 1323.
Howard, R.J., Brammer, M., Wright, I., Woodruff, P.W., Bullmore, E.T., Zeki, S., 1996. A
direct demonstration of functional specialization within motion-related visual and
auditory cortex of the human brain. Current Biology 6, 10151019.
Keri, S., Antal, A., Szekeres, G., Benedek, G., Janka, Z., 2002. Spatiotemporal visual processing in schizophrenia. Journal of Neuroscience and Clinical Neurosciences 14,
190196.
Kim, J., Doop, M.L., Blake, R., Park, S., 2005. Impaired visual recognition of biological
motion in schizophrenia. Schizophrenia Research 77, 299307.
Legge, G.E., 1978. Sustained and transient mechanisms in human vision: temporal and
spatial properties. Vision Research 18, 6981.
Livingstone, M., Hubel, D., 1988. Segregation of form, color, movement and depth:
anatomy, physiology and perception. Science 240, 740749.
MacKinnon, D.P., Krull, J.L., Lockwood, C.M., 2000. Equivalence of the mediation,
confounding and suppression effect. Prevention Science 1, 173181.
Newsome, W.T., Par, E.B., 1988. A selective impairment of motion perception following lesions of the middle temporal visual area (MT). Journal of Neuroscience 8,
22012211.
Penn, D.L., Ritchie, M., Francis, J., Combs, D., Martin, J., 2002. Social perception in
schizophrenia: the role of context. Psychiatry Research 109, 149159.
Rassovsky, Y., Green, M.F., Nuechterlein, K.H., Breitmeyer, B., Mintz, J., 2005. Modulation
of attention during visual masking in schizophrenia. American Journal of Psychiatry
162, 15331535.
Rosenthal, R., Hall, J.A., DiMatteo, M.R., Rogers, P.L., Archer, D., 1979. Sensitivity to
nonverbal communication: the PONS test.
Sergi, M.J., Green, M.F., 2003. Social perception and early visual processing in
schizophrenia. Schizophrenia Research 59, 233241.
Sergi, M.J., Rassovsky, Y., Nuechterlein, K.H., Green, M.F., 2006. Social perception as a
mediator of the inuence of early visual processing on functional status in
schizophrenia. American Journal of Psychiatry 163, 448454.
Slaghuis, W.L., Bishop, A.M., 2001. Luminance icker sensitivity in positive- and
negative-symptom schizophrenia. Experimental Brain Research 138, 8899.
Ungerleider, L.G., Haxby, J.V., 1994. What and where in the human brain. Current
Opinion in Neurobiology 4, 157165.
Vaina, L.M., Solomon, J., Chowdhury, S., Sinha, P., Belliveau, J.W., 2001. Functional
neuroanatomy of biological motion perception in humans. Proceedings of the
National Academy of Sciences of the United States of America 98, 1165611661.
Vauth, R., Rsch, N., Wirtz, M., Corrigan, P.W., 2004. Does social cognition inuence the
relation between neurocognitive decits and vocational functioning in schizophrenia? Psychiatry Research 128, 155165.