Abnormalities in Visually Guided Saccades Suggest
Corticofugal Dysregulation in Never-Treated
James L. Reilly, Margret S.H. Harris, Matcheri S. Keshavan, and John A. Sweeney
Background: Previous studies have reported intact visually guided saccades in schizophrenia, but these are limited by potential acute
and long-term pharmacological treatment effects, small sample sizes, and a failure to follow patients over time.
Methods: Visually guided saccades were examined in 44 antipsychotic-naive patients experiencing their first episode of schizophrenia
prior to treatment and again after 6, 26, and 52 weeks of antipsychotic treatment. Thirty-nine matched healthy individuals were
followed over the same period.
Results: Before treatment, patients showed faster saccade latencies to unpredictable visual targets, suggesting reduced inhibitory
regulation of brainstem saccade generators by neocortical attentional systems. Risperidone treatment reduced this deficit, suggesting
a facilitation of attentional function, but haloperidol treatment did not. However, there was also a modest decline in saccade accuracy
after risperidone treatment. The ability to sustain fixation of static central and peripheral targets was unimpaired before and after
Conclusions: These findings provide evidence for impairments in neocortical attentional systems that cause reduced corticofugal
regulation of brainstem systems in schizophrenia. This dysfunction appears to be minimized by the atypical antipsychotic risperidone
but at the cost of a subtle reduction in saccade accuracy, possibly mediated via adverse effects on cerebellar vermis function.
Key Words: Schizophrenia, first-episode, eye movements, dorso-
lateral prefrontal cortex, spatial attention, treatment effects
tor studies are a promising approach for investigating regional
neurophysiological dysfunction in schizophrenia and the effects
of treatment on widely distributed brain systems. Based on single
cell recording and lesion studies of nonhuman primates, the
neural circuitry for sensorimotor and cognitive control of eye
movements is one of the better delineated systems in the primate
brain (Goldberg and Colby 1992; Goldman-Rakic 1988). Further-
more, evidence from functional imaging and postmortem studies
has specified homologous regions involved in oculomotor con-
trol in humans (O’Driscoll et al 2000; Rosano et al 2002, 2003;
Sweeney et al 1996).
Early oculomotor studies of schizophrenia focused primarily
on pursuit eye movements (Clementz and Sweeney 1990; Levy
et al 1994). More recently, studies of saccadic eye movements
have been conducted in patients with schizophrenia (see Broerse
et al 2001a). In the most basic saccade paradigm evaluating
reflexive or visually guided saccades, subjects execute saccadic
eye movements to unpredictable visual stimuli. Reflexive sac-
cades require a shift of spatial attention to a new location and the
preparation of a precise motor program to shift the eyes to that
location. The neural circuitry for reflexive saccadic eye move-
ments involves cortical regions including parietal, frontal, and
ognitive deficits are defining features of schizophrenia,
yet the underlying neurophysiologic disturbances causing
these impairments are not well characterized. Oculomo-
supplementary eye fields, as well as the striatum, brainstem, and
cerebellum. The peak velocity of saccades is determined primar-
ily by brainstem mechanisms (Van Gisbergen et al 1981), while
saccade accuracy is controlled primarily by the cerebellum
(Robinson et al 1993). The latency of saccades depends more
heavily on projections from cortical eye fields and basal ganglia
to the superior colliculus (Munoz and Fecteau 2002; Schall and
Previous studies of visually guided saccades in schizophrenia
have reported normal basic control of saccadic eye movements
(Broerse et al 2001b; Fukushima et al 1990a, 1990b, 1994; Iacono
et al 1981; Karoumi et al 1998; Levin et al 1981; Maruff et al 1998;
Mather and Putchat 1982). Interpretation of findings from these
studies is complicated by acute and long-term psychotropic
medication and disease effects, small sample sizes, and exami-
nation of only smaller saccades rather than the full dynamic
range of saccades and attention shifts across the visual field.
Few studies have examined visually guided saccades among
patients who are treatment naive or unmedicated at the time of
testing. In investigations of patients in their first episode of
psychosis, Hutton et al (1998, 2001, 2002) did not observe any
abnormality in reflexive saccade latency or accuracy among
neuroleptically treated or untreated patients, but the appearance
of peripheral targets was paired with an auditory tone, which
could have affected responses to the visual targets. Broerse et al
(2002) compared first-episode patients after treatment with either
olanzapine or risperidone and failed to detect any difference in
reflexive saccades between the two treatment groups or from
healthy control subjects, but these investigators used an auditory
tone to signal target onset similar to Hutton et al (1998, 2001,
2002). In a comparison of the effects of a typical (haloperidol)
and an atypical (risperidone) neuroleptic in a small group of
never-treated patients, Sweeney et al (1997) reported normal
reflexive saccades prior to treatment but observed prolonged
latency and decreased peak velocity and accuracy of saccades
after 4 to 6 weeks of treatment with risperidone. Straube et al
(1999) and Muller et al (1999) examined saccade parameters
using a gap paradigm among a sample of first-episode treatment-
naive patients and in a smaller subgroup after treatment with
From the Center for Cognitive Medicine (JLR, MSHH, JAS), University of
Illinois at Chicago, Chicago, Illinois; and Department of Psychiatry and
Behavoiral Neurosciences (MSK), Wayne State University, Detroit, Mich-
icine, 912 S Wood Street, MC 913, University of Illinois at Chicago,
Chicago, IL 60612; E-mail: email@example.com.
Received April 16, 2004; revised September 24, 2004; accepted October 27,
BIOL PSYCHIATRY 2005;57:145–154
© 2005 Society of Biological Psychiatry
atypical antipsychotics. They found no differences between
patients and control subjects in latency or accuracy of reflexive
saccades at baseline or after treatment (Muller et al 1999) but did
report a substantial reduction in peak saccade velocity after
treatment (Straube et al 1999).
Given the well-documented attention and sensory informa-
tion processing problems in schizophrenia, including abnormal-
ities in early processing of visual information (Braus et al 2002;
Butler et al 2001; Foxe et al 2001), visual orienting (Huey and
Wexler 1994; Posner et al 1988), sustaining and shifting visual
attention (Cornblatt and Keilp 1994; Nuechterlein et al 1994), and
differential responsiveness to information presented across the
visual field (Cegalis et al 1977), the finding of intact visually
guided saccades is somewhat surprising. Furthermore, normal
visually guided saccades among patients suggests preservation of
a major sensorimotor pathway involving widely distributed neo-
cortical (parietal and frontal) and subcortical brain regions that
would be important to establish in a disorder believed to be
characterized by widespread disruptions in neural circuitry.
The present study examined visual attentional and sensori-
motor systems in antipsychotic-naive first-episode schizophrenia
and the impact of two antipsychotic medications on these
systems. Visually guided saccades to temporally and spatially
unpredictable targets were elicited from ?30° across the hori-
zontal visual field. Patients were examined prior to treatment and
then re-examined three more times over the course of a 1-year
follow-up. Performance of patients was compared with that of
matched healthy individuals studied over a similar time period.
Methods and Materials
Patients and healthy subjects participated in the University of
Pittsburgh First-Episode Project, a prospective study of the early
course of schizophrenia. The patient group consisted of 44
antipsychotic-naive individuals (18 female patients, 26 male
patients) with a DSM-IV diagnosis of schizophrenia (n ? 39),
schizoaffective disorder (n ? 4), or schizophreniform disorder
(n ? 1). Diagnoses were established using structured clinical
interviews (SCID; Spitzer et al 1987) reviewed with other clinical
data at consensus diagnosis meetings. Thirty-nine healthy indi-
viduals (13 female subjects, 26 male subjects) recruited from the
community were selected to match the patient group on age,
gender, and parental socioeconomic status (SES) (see Table 1)
and did not meet criteria for any current or past Axis I disorder
according to SCID interviews. All subjects met the following
criteria: 1) age between 18 and 49 years; 2) no known systemic
or neurologic disease; 3) no prior treatment with electroconvul-
sive therapy; 4) no history of head trauma with loss of conscious-
ness; 5) no lifetime history of substance dependence or of
substance abuse for at least 3 months prior to entering study;
6) no anticonvulsants or benzodiazepines for at least 1 month
prior to entering study; 7) no coffee, tea, or cigarettes 1 hour
prior to eye movement testing; and 8) at least 20/40 uncorrected
or corrected far acuity. All subjects provided verbal and written
informed consent, and the study was approved by the University
of Pittsburgh Institutional Review Board (IRB).
Patients’ baseline eye movement studies were conducted
within 7 days before treatment initiation, and follow-up testing
was conducted approximately 6, 26, and 52 weeks after treat-
ment initiation. Clinicians, blind to performance on eye move-
ment tasks, completed clinical ratings in parallel with eye
movement testing using the Brief Psychiatric Rating Scale (BPRS)
(Overall and Gorman 1962), the Schedule for the Assessment of
Positive Symptoms (SAPS) (Andreasen 1984b) and Schedule for the
Assessment of Negative Symptoms (SANS) (Andreasen 1984a), and
the 24-item Hamilton Depression Rating Scale (Hamilton 1960).
Thirteen patients were initially treated with haloperidol and 24
patients were initially treated with risperidone (Table 2). Seven
remaining patients were treated with other psychotropic medi-
cations. Patients were not randomly assigned to receive a
particular antipsychotic medication but were treated in accor-
dance with the standard practice of our clinical service at the time
of their enrollment. Patients recruited at the outset of the study
were typically prescribed haloperidol, while those patients en-
rolled in later years were more often prescribed risperidone.
Patients receiving haloperidol and risperidone did not differ in
clinical ratings at baseline and showed similar clinical improve-
ment with treatment (all p’s ? .05) (Table 3). Findings from a
subset of subjects in the present study (n ? 10 per treatment
group) were reported in a prior study (Sweeney et al 1997) that
examined visually guided saccades at baseline and at 6 weeks
after treatment initiation.
Subjects were tested alone in a darkened black room and
were positioned facing a circular black arc with a 1-m radius and
Table 1. Demographic Characteristics of Schizophrenia Patients and
Healthy Individuals at Baseline
(n ? 44)
(n ? 39)p Value
p value for gender reflects significance level for ?2test. Mean (SD) re-
ported for age, IQ (Ammons Quick Test), and parental SES (Hollingshead),
with p value reflecting significance level for independent sample t tests.
IQ, intelligence quotient; SES, socioeconomic status.
Table 2. Medication Dosages (mg) at 6-Week, 26-Week, and 52-Week Follow-Up
6-Week Follow-Up26-Week Follow-Up 52-Week Follow-Up
n M (SD)nb
Other Atypical Antipsychotica
8 4.6 (4.7)
Values are daily dose of medication (mg). M, mean; SD, standard deviation.
aDoses for other antipsychotics are presented in approximate haloperidol equivalents. These patients were not
included in drug effect comparisons.
bn’s do not include patients who changed antipsychotic medication over the course of the study.
146 BIOL PSYCHIATRY 2005;57:145–154
J.L. Reilly et al
red light-emitting diodes (LED) embedded in the horizontal
plane at eye level. A chin and forehead rest minimized head
movement. A technician in an adjacent room provided instruc-
tions through an intercom.
Eye Movement Task
Subjects were instructed to look to visual targets whenever
they appeared. Trials began with a center fixation target that
remained illuminated for 1.5 to 2.5 seconds before peripheral
targets were presented at an unpredictable location in the left or
right visual field (target steps of 10°, 20°, or 30° from center
fixation which were maintained for 1.5 seconds). Peripheral
targets were presented coincident with the termination of the
central fixation target. A brief tone alerted subjects to the
reilluminaton of the central fixation cue, signaling the start of the
next trial (this tone was not coincident with peripheral target
presentation). Fifty-four trials were administered. Electrooculog-
raphy (EOG) recordings (Grass Neurodata 12 Acquisition Sys-
tem, Astro-Med, Inc., West Warwick, Rhode Island) were ob-
tained to assess responses across a broader field of view than can
be obtained with infrared reflection techniques. Electrodes were
placed at the lateral and nasal canthus of each eye to record eye
movements. Blinks were monitored using electrodes placed
above and below the left eye. Data were digitized on-line at 500
Hz and stored for off-line analysis.
To minimize the potential impact of baseline drift in EOG
signals over the course of testing, eye position recordings were
calibrated for each trial independently. This was done using data
obtained when subjects fixated the central fixation cue and the
peripheral target during each trial. Recordings were analyzed
using software developed in our laboratory. Data were smoothed
with a finite-impulse response filter with a passband of 0 to 16 Hz,
a smooth transition band of 16 to 70 Hz, and a stop band for
frequencies greater than 70 Hz to reduce noise artifacts with a
minimum of signal distortion. An algorithm identified saccades as
beginning when eye velocity rose above 30°/s and continuing
until the eye velocity returned below that level. In rare cases
when saccades drifted slowly to a final resting eye position (at a
speed less than 30°/s which typically marked the endpoint of
saccades), eye movement traces were manually edited to mark
the end of saccades at the point when a final resting eye position
was achieved. A technician blind to subject characteristics visu-
ally reviewed recordings from each trial to ensure that saccades
were measured correctly by our computer algorithms. Trials
were rejected if a blink or intrusive saccade occurred between
100 milliseconds prior to target presentation and the end of the
primary saccade. Saccades of latencies less than 70 milliseconds
were excluded to rule out anticipatory eye movements. The
following parameters of saccades were measured: 1) latency, the
time interval (in milliseconds) between the presentation of
peripheral targets and the onset of saccades; 2) saccade error, the
difference in degrees of visual angle between the target location
and the endpoint of saccades; and 3) peak velocity of saccades in
degrees per second.
Reliability of Saccadic Eye Movement Parameters Over Time
Intraclass correlations (ICCs) for saccade parameters were
calculated separately for patients and healthy subjects to evaluate
consistency of measurements across assessments. Patients and
healthy individuals demonstrated robust ICCs for all eye move-
ment parameters (all ICCs ? .61 and ? .83, respectively, for the
two groups; all p’s ? .001), indicating a high-level reliability of
oculomotor measurements over time.
A visual fixation task was administered to subjects with at least
20/40 uncorrected vision to determine the presence of ocular
drift and intrusive saccades during sustained visual fixation.
These studies were performed with an infrared reflection system
(Model 210, Applied Science Laboratories, Bedford, Massachu-
settes) to measure small saccades (to .20°) and slow drift at a
level of precision which is not possible with EOG recordings.
Subjects were instructed to fixate a central cue for 15 seconds and
to continue fixating that location from memory for 15 seconds
after the central fixation point was extinguished. Subjects then
fixated targets presented at 15° of visual angle to the left and right
of center, each for 15 seconds. The number of square wave jerk
intrusions during fixation (pairs of saccades in which an intrusive
saccade takes the eyes from a target of interest followed by a
second that refoveates the object of regard within 450 millisec-
onds) during central fixation were computed. Measurement of
eye velocity (in degrees per second), excluding saccades, was
used to quantify foveopetal drift toward central fixation during
fixation of eccentric targets.
There were neither significant group by direction nor group
by target step effects for any measure. Therefore, because the
aim of this study was to compare patients and healthy individuals
over time, data from leftward and rightward saccades and
different target steps were pooled before conducting statistical
analyses. Random effect regression models assessed differences
in eye movement parameters over time between patients and
healthy individuals and between patients taking haloperidol and
those taking risperidone. These models provide greater statistical
power to detect change over time and they explicitly test for the
impact of missing data on estimates of change over time-
(Hedeker and Gibbons 1997). Time was a random effect and
diagnosis or medication type was treated as a fixed effect. Linear
and quadratic effects of time were examined to identify both
gradual changes over time as well as abrupt changes with
treatment that were relatively stable over follow-up. To test for
Table 3. Clinical Ratings at Baseline and 6-Week, 26-Week, and 52-Week Follow-Up
(n ? 44)
(n ? 44)
(n ? 28)
(n ? 26)
Data presented are mean (SD).
the Assessment of Positive Symptoms; HAM-D, Hamilton Depression Inventory (24-item).
J.L. Reilly et al
BIOL PSYCHIATRY 2005;57:145–154 147
Table 4. Mean (SD) Saccade and Fixation Parameters for Patients and Healthy Individuals Across Target Displacements at Baseline and 6-Week, 26-Week, and 52-Week Follow-Up
Baseline6-Week Follow-Up 26-Week Follow-Up52-Week Follow-Up
(n ? 44)
(n ? 39)
(n ? 44)
(n ? 39)
(n ? 28)
(n ? 27)
(n ? 26)
(n ? 28)
Peak Velocity (degree/s)
(n ? 35)
(n ? 24)
(n ? 35)
(n ? 24)
(n ? 24)
(n ? 17)
(n ? 23)
(n ? 17)
Velocity of Ocular Drift
Mean Number of Square
Wave Jerks across
148 BIOL PSYCHIATRY 2005;57:145–154
J.L. Reilly et al
any systematic influence of missing data, random effect pattern-
mixture models were fit adding another fixed parameter corre-
sponding to whether a subject had complete or missing data at
the 26-week or 52-week follow-up testing (all subjects had
complete data for the first two testing). Groups were matched on
demographic variables at all follow-ups.
Comparisons of Patients and Healthy Individuals
Mean values (SD) for patients and healthy individuals on sac-
cade and fixation measures over time are presented in Table 4.
The random effects model regressing saccade
latency on the linear and quadratic terms of time, diagnosis, and
their interactions detected significant effects for the linear
[F(1,81) ? 24.23, p ? .001] and quadratic [F(1,62) ? 14.54, p ?
.001] effects of time and the interaction of time with diagnosis
[F(1,45) ? 13.42, p ? .001 and F(1,45) ? 10.20, p ? .003,
respectively]. As shown in Figure 1, patients had significantly
shorter saccade latencies at baseline than healthy subjects (p ?
.05), but the groups did not differ at any of the follow-up testing.
Consistent with those findings, patients’ latencies increased
significantly from baseline to 6-week follow-up (p ? .001) but
did not change thereafter. Latencies of healthy individuals did
not change significantly over time.
Figure 2 shows a leftward shift in the distribution at baseline
of patients’ saccade latencies relative to healthy individuals
(Figure 2A). The figure shows a leftward shift in the distribution
at baseline saccade latencies relative to healthy individuals
(Figure 2A). While there was remarkable consistency over time
among healthy individuals in their latencies (Figure 2B), there
was a distinct rightward shift in the distribution of patients’
latencies after treatment that was largely driven by those patients
taking risperidone (Figure 2C). Exploratory analyses of baseline
data for each target step amplitude indicated patients had
relatively shorter latencies to targets presented closer to central
fixation (p ? .01 for 10°, p ? .07 for 20°, and p ? .21 for 30°).
Accuracy. The random effects model for saccade accuracy
detected a significant linear effect of time [F(1,81) ? 8.45, p ?
.005] and of the interaction between time and diagnosis [F(1,109) ?
16.89, p ? .001]. As shown in Figure 3, patients and healthy
individuals did not differ in saccade accuracy at baseline; how-
ever, patients’ saccades were significantly less accurate than
healthy subjects’ saccades at all follow-up occasions (p’s ? .003).
Patients’ saccades tended to be more hypometric (undershooting
targets) after 6 weeks of treatment (p ? .08), and declines in
accuracy from baseline were significant at all subsequent follow-
ups (p’s ? .05).
Peak Velocity. The random effects model for peak velocity
of saccades detected a significant linear effect of time [F(1,81) ?
12.25, p ? .001] and of the interaction between time and
diagnosis [F(1,109) ? 7.00, p ? .01]. While patients’ peak
velocities were comparable to healthy subjects at baseline, their
velocities slowed significantly after starting treatment (p ? .05).
Because patients’ saccades were hypometric relative to healthy
individuals (as reported above) and smaller saccades typically
reach lower peak velocities (Leigh and Kennard 2004), the
Figure 1. Mean (SE) of saccade latencies (milliseconds) for patients and
healthy individuals at baseline and 6-week, 26-week, and 52-week follow-
up. *p ? .05 for between-group comparison.
Figure 2. Cumulative percentage of trial-wise saccade latencies. The top
panel (A) shows the comparison of patients (triangles) and healthy individ-
uals (circles) at baseline. Note the leftward shift in the patient distribution,
indicating a greater percentage of speeded latencies. The middle panel (B)
latencies from baseline (solid line) to 6-week follow-up (broken line) as
indicated by the nearly overlapping distributions. The bottom panel (C)
shows the rightward shift from the baseline distribution for patients taking
risperidone as opposed to those taking haloperidol.
J.L. Reilly et al
BIOL PSYCHIATRY 2005;57:145–154 149
analysis of peak velocity data was repeated with peak velocity of
each group corrected for saccade amplitude. After this correc-
tion, the group by time interaction in saccade velocities was no
Visual Fixation. The random effects model on the number of
square wave jerks detected a significant main effect for time (p ?
.05) but no significant effect for diagnosis or any interaction
between time and diagnosis. There was a slight increase in the
number of square wave jerks for all subjects over time (Table 4).
For eye drift velocity data when subjects held eccentric gaze, the
random effects regression failed to detect any significant effects
for time, diagnosis, or their interaction.
Comparison of Patients Receiving Haloperidol
Mean values (SD) for patients treated with haloperidol and
risperidone on saccade measures over time are presented in
revealed significant effects for the linear [F(1,35) ? 26.07, p ?
.001] and quadratic [F(1,25) ? 18.45, p ? .001] effects of time
and their interactions with drug type [F(1,19) ? 9.33, p ? .01] and
[F(1,19) ? 8.79, p ? .01], respectively. There were no differences
at baseline between those who later received haloperidol and
risperidone in saccade latencies. However, patients treated with
risperidone had longer saccade latencies compared with those
treated with haloperidol at all follow-ups (p’s ? .05) (see Figure
4). Consistent with that effect, patients treated with risperidone
had significant increases in saccade latencies from baseline to
later follow-ups (p’s ? .01), while those treated with haloperidol
did not. The trial-wise distribution of saccade latencies shows
that risperidone-treated patients had a rightward shift in their
latency distribution relative to baseline compared with the
The random effects model for saccade latency
and healthy individuals at baseline and 6-week, 26-week, and 52-week
follow-up. **p ? .01 for between-group comparison.
Table 5. Mean (SD) Saccade Parameters for Patients Treated with Haloperidol and Risperidone Across Target Displacements
at Baseline and 6-Week, 26-Week, and 52-Week Follow-Up
Baseline 6-Week Follow-Up 26-Week Follow-Up52-Week Follow-Up
(n ? 13)
(n ? 24)
(n ? 13)
(n ? 24)
(n ? 8)
(n ? 15)
(n ? 10)
(n ? 13)
Figure 4. Mean (SE) of saccade latencies (milliseconds) for haloperidol-
treated and risperidone-treated patients at baseline and 6-week, 26-week,
and 52-month follow-up. *p ? .05, **p ? .01 for between-group compari-
150 BIOL PSYCHIATRY 2005;57:145–154
J.L. Reilly et al
negligible change in the latency distribution among haloperidol-
treated patients (Figure 2C).
For saccade error, the pattern mixture model
indicated a significant linear effect of time [F(1,33) ? 8.29, p ?
.01] and a time by drug interaction [F(1,46) ? 4.47, p ? .05].
Patients taking risperidone showed reduced saccade accuracy
relative to their baseline performance at the 26-week and 52-
week follow-ups (p’s ? .05), whereas there was no significant
change from baseline to any later assessments among haloperi-
Peak Velocity. The random effects regression model of peak
velocity data detected a significant effect of time [F(1,35) ? 8.32,
p ? .01], suggesting a comparable reduction over time in peak
saccade velocity among patients treated with either risperidone
or haloperidol. When saccade velocity was adjusted for ampli-
tude, only a marginal main effect for time [F(1,35) ? 3.26, p ?
.08] was detected, indicating minimal reduction over time in peak
saccade velocity after accounting for effects of reduced saccade
shown in Table 3, patients improved significantly on clinical
ratings from baseline to follow-up, irrespective of type of anti-
psychotic medication (all p’s ? .05). Patients treated with halo-
peridol and those treated with risperidone did not differ in their
degree of clinical improvement on any clinical rating. There were
no significant associations between change in clinical ratings and
change in eye movements. Furthermore, no associations were
found between medication dose or eye movements.
Never-treated patients in their first episode of schizophrenia
had speeded saccadic latencies to visual targets relative to
matched healthy individuals. Following six weeks of treatment
with antipsychotic medication, saccade latencies fell within the
normal range among those patients treated with risperidone but
not those treated with haloperidol. These changes in saccade
latencies after treatment were maintained throughout the 1-year
follow-up interval. In addition, risperidone but not haloperidol
treatment was associated with a modest, but persistent, reduction
in saccade accuracy. Patients and healthy individuals had similar
peak saccade velocity at baseline and there was minimal evi-
dence of a reduction of patients’ saccade velocity after starting
treatment. Finally, patients’ ability to maintain central or periph-
eral gaze was normal before and after initiating treatment.
Speeded Saccadic Response to Visual Targets
The findings reported here are consistent with impairment in
the attentional regulation of sensorimotor systems during the
acute phase of schizophrenia. The persistence of speeded sac-
cade latencies to unpredictable targets among patients treated
with haloperidol suggests that this deficit is stable in schizophre-
nia but that it may be minimized by the atypical antipsychotic
risperidone. Shortened saccade latencies to novel targets and
intact gaze maintenance among schizophrenia patients suggest
diminished control of neurophysiologic systems supporting au-
tomatic components of spatial attention and oculomotor pro-
gramming. The distinction between patients’ ability to maintain
ocular stability when centrally or peripherally fixating a target
and their speeded responses to the appearance of unpredictable
visual stimuli is important because it implicates dysregulation of
the corticofugal inputs to brainstem regions rather than an
intrinsic disturbance within these regions. Relative to neocortical
attentional systems, brainstem and cerebellar physiology are
largely responsible for maintenance of ocular stability and other
basic saccade parameters including saccade accuracy and peak
velocity (Robinson et al 1993; Van Gisbergen et al 1981). These
were intact among patients before treatment, arguing against a
primary disturbance in cerebellar or pontine visuomotor systems.
Therefore, we propose that our findings are best explained by a
model of reduced neocortical regulation of brainstem systems
responsible for driving saccades to novel visual inputs.
Evidence from single cell recording and neuronal inactivation
studies with nonhuman primates and evidence from lesion,
functional imaging, and transcranial magnetic resonance imaging
(TMS) studies with humans have defined the role of several
cortical and subcortical regions in triggering saccades (Leigh and
Kennard 2004; Pierrot-Deseilligny et al 2002) Figure 5 illustrates
the cortical areas and their projections to subcortical regions
involved in saccade generation. Neocortical areas include the
dorsolateral prefrontal cortex (DLPFC) and the frontal, supple-
mentary, and parietal eye fields. The cortical eye fields and the
DLPFC have direct projections to the superior colliculus, and
there is a tight linkage between activity in cortical eye fields and
superior colliculus in relation to saccade generation (Helminski
and Segraves 2003; Munoz 2002; Pare and Wurtz 1997; Schall and
Bichot 1998; Segraves and Goldberg 1987; Sommer and Wurtz
1998). There is also an indirect pathway from the DLPFC and
frontal eye fields to the superior colliculus via the caudate
nucleus and substantia nigra pars reticulata (SNpr) (Hikosaka
et al 2000). Cortical input via these pathways to the colliculus
stimulates fixation neurons located in the rostral pole of the
colliculus to hold gaze on objects of interest and maintains a
balance between fixation neurons and saccade neurons located
in more caudal portions (Figure 5B) (Munoz and Fecteau 2002).
While the cortical eye fields provide excitatory input that
directly drives saccade neurons in the superior colliculus and
indirectly releases the colliculus from tonic inhibition from the
SNpr, additional input to the visual fixation zone of the colliculus
is provided by DLPFC (Hikosaka et al 2000). Single cell record-
ings in nonhuman primates have identified fixation neurons in
DLPFC that have an inverse activity profile to saccade-related
neurons in the superior colliculus and frontal eye fields, suggest-
ing that fixation-related DLPFC neurons tonically inhibit saccade
neurons in the colliculi and frontal eye fields (Tinsley and
Everling 2002). Thus, hypofunction of the fixation neurons in
DLPFC could lead to reduced excitation of subcortical fixation
systems, resulting in a hyperexcitability of orienting responses in
the presence of novel stimuli. Given evidence of hypofrontality
among neuroleptic-naive first-episode patients (Andreasen et al
1997) and multiple lines of work suggesting prefrontal impair-
ment in schizophrenia, dysfunction in prefrontal circuitry is a
likely cause of the abnormal visual orienting observed in our
The corticofugal pathway from neocortex to the colliculus via
the striatum does not appear to be responsible for the speeded
latencies among patients. Haloperidol has a high affinity for D2
receptors in the caudate nucleus and is known to markedly affect
striatal physiology and motor function. Therefore, the stability of
shortened latencies among patients treated with haloperidol
makes it unlikely that projections to or from the striatum are
responsible for the speeded response to visual targets reported
here. It is also unlikely that a neurophysiologic disturbance
below the level of the superior colliculus in the pons or midbrain
could account for the findings of the present study, because
disturbance at the level of pontine burst and pause cells, to
J.L. Reilly et al
BIOL PSYCHIATRY 2005;57:145–154 151
which the superior colliculus projects, would result in significant
alterations in saccade velocity and the stability of visual fixation.
Our data regarding ocular stability in schizophrenia is consistent
with other reports (Gooding et al 2000; Kissler and Clementz 1998).
Finally, the normal saccade velocity and accuracy among patients
prior to treatment indicate that the ability of the colliculus to trigger
saccades and the cerebellar regulation of saccade accuracy are
A model of reduced prefrontal inhibitory regulation of sub-
cortical and brainstem systems responsible for oculomotor ori-
enting fits well with findings from paradigms in which shortened
saccade latencies are experimentally elicited. Reduced saccade
latencies result when DLPFC is experimentally inactivated with
TMS (Muri et al 1999; Tinsley and Everling 2002). When a
temporal delay or gap is introduced between fixation point offset
and peripheral target onset in saccade paradigms, faster saccadic
reactions are seen due to a disengagement of attentional systems
that reduces activity in fixation neurons, thereby releasing the
oculomotor system to respond more rapidly to visual input
(Fischer and Boch 1983). Indeed, prior investigations have
demonstrated that patients with schizophrenia have decreased
saccadic response times in gap paradigms compared with
healthy individuals (Clementz 1996; Larrison-Faucher et al 2004;
Sereno and Holzman 1996), consistent with dysfunction of
DLPFC and its inputs to neocortical and subcortical circuitry.
Components of other sensorimotor pathways might account
for the speeded saccadic responses, but these also seem less
likely. Faster saccadic response times in schizophrenia could
reflect speeded transfer of visual input into visuomotor systems.
However, studies of visual evoked potentials (Butler et al 2001;
Foxe et al 2001) and functional magnetic resonance imaging
(fMRI) investigations of early visual information processing
(Braus et al 2002) have reported reduced low-level, early-stage
visual processing rather than speeded processing, particularly in
the dorsal visual pathways that receive magnocellular inputs and
are involved in orienting responses to salient stimuli. Therefore,
it is not likely that faster processing of inputs through visual
pathways is responsible for the significantly reduced saccade
latencies reported here.
To our knowledge, this is the first study to document
abnormalities in visually guided saccades among acutely ill,
never-treated patients with schizophrenia. Previous investiga-
tions have reported normal performance on visually guided
saccade tasks in schizophrenia. However, there are important
methodological differences between these studies and the
present investigation. Most prior studies examining visually
guided saccades in schizophrenia have been limited by small
samples and potentially by confounding influences of treat-
ment and/or disease chronicity (Broerse et al 2001a, 2001b;
Crawford et al 1995a, 1995b; Fukushima et al 1994, 1990a,
1990b; Hutton et al 2002; Iacono et al 1981; Karoumi et al
1998; Levin et al 1981; Maruff et al 1998; Mather and Putchat
1982). Among the few studies that examined visually guided
saccades in unmedicated first-episode patients, methodologi-
cal differences such as the inclusion of an auditory tone
presented simultaneously with the peripheral target presenta-
tion (Hutton et al 1998, 2001) and use of visual targets for
which target appearance was predictable (Muller et al 1999)
could account for the failure of earlier investigations to detect
the difference in saccade latencies reported here.
Treatment Effects of Antipsychotics on Visually
Differences in saccade latency after treatment with risperi-
done, but not haloperidol, suggest an improvement in neocorti-
cal control of brainstem oculomotor systems associated with
atypical antipsychotic treatment. The modest reduction in sac-
cade accuracy among risperidone-treated patients indicates that
the enhancement of visual attentional systems, as suggested by
the normalization of saccade latencies after treatment, may occur
at the cost of some reduction in fine motor control in the
oculomotor system. Haloperidol treatment did not result in a
significant change in saccade accuracy, suggesting a differential
impact of these medications on aspects of motor programming
Figure 5. Lateral view of human cerebral cortex and projections to superior colliculus (SC) involved in saccade triggering. (A) Cortical and subcortical areas
saccade neurons in caudal portions. Directional coding is shown with upward direction in superior and downward direction in inferior regions.
152 BIOL PSYCHIATRY 2005;57:145–154
J.L. Reilly et al
under control of the cerebellar vermis. This explanation is
consistent with functional imaging studies that demonstrate a
significantly greater reduction in cerebellar blood flow with
risperidone treatment compared with haloperidol treatment
(Miller et al 2001).
In contrast to prior reports of normal saccade latencies
among patients treated with typical antipsychotics, patients in
the present study treated with haloperidol had saccade laten-
cies that remained shortened even after clinical stabilization.
This difference could be accounted for by several factors.
First, it is possible that the higher treatment dosages used in
older studies could increase saccade latencies via effects on
striatum or other brain regions. The average medication dose
from these earlier studies (Fukushima et al 1994, 1990a;
Karoumi et al 1998; Mather and Putchat, 1982) was two to four
times higher than the dosage prescribed in the current study.
Our study also differed in having a relatively large sample of
patients early in the course of illness that could have increased
our ability to detect effects. From another perspective, any
interpretation of a greater effect of risperidone than haloper-
idol in normalizing visual orienting systems needs to be
considered with caution. Although patients in each medica-
tion group did not differ on demographic or clinical ratings at
baseline or at any follow-up, the nonrandom assignment to
treatment necessitates caution in the interpretation of differ-
ential medication effects on these saccade parameters.
Investigations of oculomotor neurophysiology provide a reli-
able quantitative strategy for investigating attentional deficits in
schizophrenia and potentially for monitoring effects of antipsy-
chotic medications on cognitive processes and motor systems
over the course of treatment. This study suggests that a dysregu-
lation of prefrontal control of visual attentional systems is present
in schizophrenia and that this abnormality may be reduced by
the atypical antipsychotic risperidone. Future studies of patients
randomized to treatment, evaluated with paradigms in which
disengagement of attentional systems regulating the oculomotor
system can be experimentally manipulated (e.g., gap overlap
conditions), and using event-related neuroimaging methods can
provide further clarification and validation of findings observed
in the present study.
This publication was supported by funds received from
MH62134, MH45156, MH 01433 and NIH/NCRR/GCRC Grant
M01 RR00056. We thank Drs. Cameron S. Carter, M.D.,
Gretchen Haas, Ph.D., and Matcheri S. Keshavan, M.D. and the
clinical core staff of the Center for the Neuroscience of Mental
Disorders (MH45156) for their assistance in diagnostic and
psychopathological assessments. Kira Lathrop, MAMS, provided
assistance preparing the illustrations in Figure 5.
Ammons CH, Ammons RB (1962): The Quick Test (QT): provisional manual.
Iowa City, IA: University of Iowa Press.
Iowa City, IA: University of Iowa Press.
Andreasen NC, O’Leary DS, Flaum M, Nopoulos P, Watkins GL, Boles Ponto
LL, et al (1997): Hypofrontality in schizophrenia: Distributed dysfunc-
tional circuits in neuroleptic-naive patients. Lancet 349:1730–1734.
processing in neuroleptic-naive first-episode schizophrenic patients: A
functional magnetic resonance imaging study. Arch Gen Psychiatry 59:
Broerse A, Crawford TJ, den Boer JA (2001a): Parsing cognition in schizo-
and risperidone on cognition in schizophrenia? A saccadic eye move-
Broerse A, Holthausen EA, van den Bosch RJ, den Boer JA (2001b): Does
frontal normality exist in schizophrenia? A saccadic eye movement
study. Psychiatry Res 103:167–178.
(2001): Dysfunction of early-stage visual processing in schizophrenia.
Cegalis JA, Leen D, Solomon EJ (1977): Attention in schizophrenia: An anal-
gap interval among schizophrenia patients. Exp Brain Res 111:121–130.
Clementz BA, Sweeney JA (1990): Is eye movement dysfunction a biological
Cornblatt BA, Keilp JG (1994): Impaired attention, genetics, and the patho-
physiology of schizophrenia. Schizophr Bull 20:31–46.
Crawford TJ, Haeger B, Kennard C, Reveley MA, Henderson L (1995a): Sac-
cadic abnormalities in psychotic patients. I. Neuroleptic-free psychotic
patients. Psychol Med 25:461–471.
Crawford TJ, Haeger B, Kennard C, Reveley MA, Henderson L (1995b): Sac-
cadic abnormalities in psychotic patients. II. The role of neuroleptic
treatment. Psychol Med 25:473–483.
Fischer B, Boch R (1983): Saccadic eye movements after extremely short
reaction times in the monkey. Brain Res 260:21–26.
Foxe JJ, Doniger GM, Javitt DC (2001): Early visual processing deficits in
cal mapping. Neuroreport 12:3815–3820.
of saccadic eye movement in patients with frontal cortical lesions and
parkinsonian patients in comparison with that in schizophrenics. Biol
the control of voluntary saccadic eye movements in schizophrenic pa-
tients. Biol Psychiatry 28:943–958.
Voluntary control of saccadic eye movements in patients with schizo-
phrenic and affective disorders. J Psychiatr Res 24:9–24.
Goldberg ME, Colby CL (1992): Oculomotor control and spatial processing.
works in primate association cortex. Annu Rev Neurosci 11:137–156.
Gooding DC, Grabowski JA, Hendershot CS (2000): Fixation stability in
schizophrenia, bipolar, and control subjects. Psychiatry Res 97:119–128.
Hamilton M (1960): A rating scale for depression. J Neurol Neurosurg Psychi-
ture models for missing data in longitudinal studies. Psychol Methods
Helminski JO, Segraves MA (2003): Macaque frontal eye field input to sac-
Hikosaka O, Takikawa Y, Kawagoe R (2000): Role of the basal ganglia in the
control of purposive saccadic eye movements. Physiol Rev 80:953–978.
Hollingshead AB (1975): Four-Factor Index of Social Status. New Haven, CT:
Yale University, Department of Sociology.
Huey ED, Wexler BE (1994): Abnormalities in rapid, automatic aspects of
attention in schizophrenia: Blunted inhibition of return. Schizophr Res
Hutton SB, Crawford TJ, Puri BK, Duncan LJ, Chapman M, Kennard C, et al
(1998): Smooth pursuit and saccadic abnormalities in first-episode
schizophrenia. Psychol Med 28:685–692.
Hutton SB, Cuthbert I, Crawford TJ, Kennard C, Barnes TR, Joyce EM (2001):
Saccadic hypometria in drug-naive and drug-treated schizophrenic pa-
tients: A working memory deficit? Psychophysiology 38:125–132.
first-episode schizophrenia. Neuropsychologia 40:1729–1736.
Iacono WG, Tuason VB, Johnson RA (1981): Dissociation of smooth-pursuit
J.L. Reilly et al
BIOL PSYCHIATRY 2005;57:145–154 153
time task that schizophrenics perform well. Arch Gen Psychiatry Download full-text
Karoumi B, Ventre-Dominey J, Vighetto A, Dalery J, d’Amato T (1998): Sac-
Kissler J, Clementz BA (1998): Fixation stability among schizophrenia pa-
tients. Neuropsychobiology 38:57–62.
Larrison-Faucher AL, Matorin AA, Sereno AB (2004): Nicotine reduces anti-
saccade errors in task impaired schizophrenic subjects. Prog Neuropsy-
Leigh RJ, Kennard C (2004): Using saccades as a research tool in the clinical
neurosciences. Brain 127:460–477.
Levin S, Holzman PS, Rothenberg SJ, Lipton RB (1981): Saccadic eye move-
ments in psychotic patients. Psychiatry Res 5:47–58.
Levy DL, Holzman PS, Matthysse S, Mendell NR (1994): Eye tracking and
schizophrenia: A selective review. Schizophr Bull 20:47–62.
control of schizophrenics: A deficit in sensory processing, not strictly in
Miller DD, Andreasen NC, O’Leary DS, Watkins GL, Boles Ponto LL, Hichwa RD
Muller N, Riedel M, Eggert T, Straube A (1999): Internally and externally
guided voluntary saccades in unmedicated and medicated schizo-
Munoz DP (2002): Commentary: Saccadic eye movements: Overview of
Munoz DP, Fecteau JH (2002): Vying for dominance: Dynamic interactions
control visual fixation and saccadic initiation in the superior colliculus.
Role of the prefrontal cortex in the control of express saccades. A trans-
cranial magnetic stimulation study. Neuropsychologia 37:199–206.
Nuechterlein KH, Dawson ME, Green MF (1994): Information-processing
abnormalities as neuropsychological vulnerability indicators for schizo-
O’Driscoll GA, Wolff AL, Benkelfat C, Florencio PS, Lal S, Evans AC (2000):
Functional neuroanatomy of smooth pursuit and predictive saccades.
Overall J, Gorman D (1962): The Brief Psychiatric Rating Scale. Psychol Rep
Pierrot-Deseilligny C, Ploner CJ, Muri RM, Gaymard B, Rivaud-Pechoux S
Posner MI, Early TS, Reiman E, Pardo PJ, Dhawan M (1988): Asymmetries in
hemispheric control of attention in schizophrenia. Arch Gen Psychiatry
Rosano C, Krisky CM, Welling JS, Eddy WF, Luna B, Thulborn KR, et al
(2002): Pursuit and saccadic eye movement subregions in human
frontal eye field: A high-resolution fMRI investigation. Cereb Cortex
tral sulcus: Chemoarchitecture of a region corresponding to the frontal
eye fields. Brain Res 972:16–30.
Schall JD, Bichot NP (1998): Neural correlates of visual and motor decision
Segraves MA, Goldberg ME (1987): Functional properties of corticotectal
neurons in the monkey’s frontal eye field. J Neurophysiol 58:1387–
Sereno AB, Holzman PS (1996): Spatial selective attention in schizophrenic,
affective disorder, and normal subjects. Schizophr Res 20:33–50.
Sommer MA, Wurtz RH (1998): Frontal eye field neurons orthodromically
activated from the superior colliculus. J Neurophysiol 80:3331–3335.
Spitzer R, Williams J, Gibbons M, First M (1987): Structured Clinical Interview
for DSM-III-R (SCID). New York: New York State Psychiatric Institute.
Straube A, Riedel M, Eggert T, Muller N (1999): Internally and externally
guided voluntary saccades in unmedicated and medicated schizo-
Sweeney JA, Bauer KS, Keshavan MS, Haas GL, Schooler NR, Kroboth PD
parison of risperidone and haloperidol in antipsychotic-naive schizo-
phrenic patients. Neuropsychopharmacology 16:217–228.
Sweeney JA, Mintun MA, Kwee S, Wiseman MB, Brown DL, Rosenberg DR,
et al (1996): Positron emission tomography study of voluntary saccadic
eye movements and spatial working memory. J Neurophysiol 75:454–
the gap effect. Prog Brain Res 140:61–72.
Van Gisbergen JA, Robinson DA, Gielen S (1981): A quantitative analysis of
154 BIOL PSYCHIATRY 2005;57:145–154
J.L. Reilly et al