ORIGINAL ARTICLES

Personality Differences in Schizophrenia Are Related

to Performance on Neuropsychological Tasks
Ronald J. Gurrera, MD,* Paul G. Nestor, PhD,* Brian F. ODonnell, PhD, Valerie Rosenberg,
and Robert W. McCarley, MD*

Abstract: Altered cognition and personality appear to emerge in

tandem and adversely affect outcome in schizophrenia, yet little
research has been done to determine whether these are related or
independent domains. In this study, the relationship between the Big
Five personality traitsneuroticism, extraversion, openness, agreeableness, conscientiousnessand cognitive and motor performance
in outpatients with chronic, clinically stable schizophrenia (N 30)
and age-matched healthy comparison subjects (N 45) was examined. Subjects completed tests of attention, executive and motor
functions, and the NEO-Five Factor Personality Inventory. Patients
scored significantly higher on neuroticism and lower on extraversion
and agreeableness, but after variance due to neuropsychological
performance was statistically removed from NEO scale scores,
personality dimensions and profiles no longer differed between
groups. Neuropsychological performance and demographic variables, but not diagnosis, uniquely accounted for statistically significant amounts of personality variance, and neuropsychological task
performance was correlated with personality dimensions in both
patients and comparison subjects. These cross-sectional data provide
preliminary evidence that personality dysfunction in schizophrenia
may be mediated by disease-related changes in cognitive operations,
or the neural processes underlying them. Longitudinal studies utilizing more comprehensive measures of neurocognitive performance
are needed to define further the relationship between neuropsychological function and personality in schizophrenia.
Key Words: Schizophrenia, personality, personality traits,
cognitive neuroscience, personality disorders.
(J Nerv Ment Dis 2005;193: 714 721)

*VA Boston Healthcare System, Boston and Brockton, Massachusetts;

Harvard Medical School, Department of Psychiatry, Boston, Massachusetts; University of Massachusetts Boston, Department of Psychology,
Boston, Massachusetts; and Indiana University, Department of Psychology, Bloomington, Indiana.
Supported by the Medical Research Service and the Schizophrenia Center of
the Department of Veteran Affairs, NIMH grant MH-40779 to Dr.
McCarley; and VA Merit Award to Dr. Nestor.
Send reprint requests to Ronald J. Gurrera, MD, VA Boston Healthcare
System (116A), 940 Belmont Street, Brockton, MA 02301.
Copyright 2005 by Lippincott Williams & Wilkins
ISSN: 0022-3018/05/19311-0714
DOI: 10.1097/01.nmd.0000185938.30783.6b

714

ognitive deficits in schizophrenia include impaired attention (Franke et al., 1994; Nuechterlein and Asarnow,
1989) and executive functions (Rund and Borg, 1999), particularly cognitive flexibility and forward planning (Morice
and Delahunty, 1996); and motor abnormalities (Wolff and
ODriscoll, 1999), including fine motor skills (Griffith et al.,
1994; Vrtunski et al., 1989). Cognitive abnormalities may be
present long before illness onset in individuals who develop
schizophrenia (Cuesta et al., 2001).
Altered personality is another consistent feature of
schizophrenia (Gurrera et al., 2000; Solano and De Chavez,
2000) and is often evident before the clinical disorder
manifests itself. Patients with schizophrenia are less extraverted and have higher neuroticism and lower conscientiousness and agreeableness scores than healthy controls (see
Gurrera et al., 2000, for concise review). High neuroticism
and low extraversion (Berenbaum and Fujita, 1994) and
cluster A personality disorder traits (Cuesta et al., 1999) are
most commonly found in individuals who later develop
schizophrenia, but the prevalence of all personality disorder
types is increased in schizophrenia (Lyons and Jerskey, 2002;
Solano and De Chavez, 2000).
Severity and quality of personality deviance affect
long-term outcome in schizophrenia (Gross and Huber, 1993;
Smith et al., 1996; Solano and De Chavez, 2000). Even
so-called normal personality traits contribute to variability
in clinical presentation and treatment response (Smith et al.,
1995) and predict workplace performance (Lysaker et al.,
1998). Poorer workplace function, particularly in the area of
social skills, is also predicted by deficits in cognitive function
(Lysaker et al., 1998). Verbal memory, attention, and card
sorting predict community and social functioning, but positive symptoms do not (Green, 1996).
Despite pari passu premorbid emergence and overlapping effects on illness outcome, little research has been done
to determine whether personality abnormalities and cognitive
deficits in schizophrenia are related or independent domains
(Cuesta et al., 2001). Abnormal cognition is generally attributed to altered brain function caused by the disorder, but there
is no consensus regarding the source of the personality
differences. Whether premorbid personality is an independent
risk factor for developing schizophrenia (van Os and Jones,
2001) or a manifestation of a common, underlying neuropathological process remains an open question.

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

Several recent studies have begun to address this question. In recently admitted inpatients with psychosis, more
severe negative, positive, and disorganization symptoms were
associated with more extreme personality disorder traits
(Cuesta et al., 1999). Sociopathy was positively correlated
with disorganization, whereas schizoid features were correlated only with positive and negative symptom dimensions. In
another study of patients with psychosis (Cuesta et al., 2001),
passive-dependent and schizoid traits, but not anankastic or
sociopathic traits, correlated with measures of memory, attention, and executive function. Moreover, personality and
cognitive measures each accounted for substantial amounts of
the others variance (Cuesta et al., 2001).
Personality and cognitive ability are correlated in the
general population (Harris et al., 1998), and the former can
affect the latter. For example, some individual differences in
task performance have been attributed to temperamentally
based speed-related personality attributes (Brebner and
Stough, 1995). Introverts may perform relatively better on
tasks emphasizing accuracy or involving associative learning
ability (e.g., verbal tasks), while extraverts may be comparatively advantaged on tasks that emphasize speed or require
the acquisition of automatic motor sequences (i.e., performance tasks; Zeidner, 1995). In a community sample selected
on the basis of self-reported schizophrenia spectrum symptoms, variation in several personality dimensions predicted
Wisconsin Card Sort Test (WCST) performance (Tien et al.,
1992). Thus, personality may also influence cognition in
individuals prone to schizophrenia.
Neurocognitive deficits can impact the acquisition of
social skills in schizophrenia (Addington et al., 1998; Green,
1996), so it seems plausible that schizophrenia could impact
cognitive operations essential to healthy personality functioning, thus contributing to premorbid and comorbid personality
features of schizophrenia. In multiple sclerosis, a progressive
demyelinating disorder commonly associated with cognitive
disturbance, frontal lobe function correlates with personality
traits (Benedict et al., 2001). In healthy volunteers, smaller
frontal brain volume is associated with more pathologic
MMPI scores (Matsui et al., 2000), suggesting that even
subclinical reductions in functional brain capacity may produce personality changes.
Considerable evidence indicates that five major trait
dimensions account for most personality variance in normal
and psychiatrically ill populations (John and Srivastava,
1999; Stone, 1993). The NEO Personality Inventory (NEO-PI;
Costa and McCrae, 1992), widely used to assess these five
factors, measures neuroticism, extraversion, openness, agreeableness, and conscientiousness. High neuroticism individuals are tense, irritable, discontented, shy, and moody, and lack
self-confidence. Extraverted individuals are sociable, assertive, energetic, adventurous, enthusiastic, and outgoing. Open
individuals are curious, imaginative, unconventional, artistic,
and excitable, and have wide interests. Agreeable individuals
are forgiving, warm, sympathetic, and modest, and not demanding or stubborn. Conscientious individuals are selfdisciplined, efficient, thorough, dutiful (not careless), deliberate (not impulsive), and organized.
2005 Lippincott Williams & Wilkins

Personality Changes in Schizophrenia

The present study employed multivariate statistical

methods to examine the contribution of neuropsychological
task performance to personality differences between healthy
controls and patients with schizophrenia. A control group
permitted the effects of a schizophrenia diagnosis, and neuropsychological performance, to be evaluated independently.
Based upon previous work (Gurrera et al., 2000), we hypothesized that better task performance would be associated with
healthier personality scoresspecifically, lower neuroticism
and higher openness, agreeableness, and conscientiousness
scores. We also hypothesized that the pattern of associations
with specific neuropsychological tasks would be distinct for
each personality dimension.

METHODS
Design and Procedure
Data collection was cross-sectional and consisted of
formal neuropsychological assessment and a self-administered personality measure.

Subjects
Medicated outpatients with clinically stable DSM-IV
schizophrenia (N 30) and healthy subjects (N 45)
recruited from the community participated in this study,
which was approved by the Human Studies Subcommittee of
the VA Boston Healthcare System. Subjects were age 18 to
55 years and spoke English as a first language. Community
volunteers were recruited through advertisements placed in
local newspapers. Callers were screened for the following
exclusion criteria: any history of ECT, neurological disorder,
head injury, mental retardation, or loss of consciousness; and
current treatment with medications that might affect cognitive
function. Potential subjects were subsequently evaluated with
the Clinicians Version of the Structured Clinical Interview
for DSM-IV Axis I disorders, which was conducted by
doctorate-level psychologists who had been trained and qualified in performing this assessment. Those with DSM-IV
alcohol or drug dependence (ever) or DSM-IV alcohol or
drug abuse within the past year were excluded. After a
complete description of the study was provided to the subjects, written informed consent was obtained.
Patients had significantly lower mean (SD) educational
achievement (12.0 1.9 vs. 14.8 2.2 grades; t 5.70; p
0.001) and socioeconomic status (SES; Hollingshead, 1965)
than controls (4.13 .63 vs. 2.33 1.04; t 9.30; p
0.001), but mean (SD) parental SES (PSES) was similar for
both groups (2.57 1.29 vs. 2.69 1.00; t .439; p 0.646).
Subject groups did not differ with respect to gender (male/
female ratio, 38/7 vs. 24/6; Fisher exact test, p 0.757) or
mean (SD) age (39.7 8.9 vs. 39.2 9.7 years; t .247;
p 0.806). Patients had mean (SD) illness duration of 17.7
(9.7) years (range, 134 years). Clinical stability was defined
operationally as the ability to complete an extensive battery
of neuropsychological tests.

Symptom Measures
Schizophrenia symptom data, measured with the SANS
and SAPS (Andreasen, 1984a, 1984b), were available in 22

715

Gurrera et al.

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

and 20 patients, respectively. Mean (SD) global symptom

scores were 12.14 (5.77) for the SANS and 9.75 (3.45) for the
SAPS.

Medication
Twenty-four patients were receiving antipsychotic
monotherapy, and six patients were treated with two different
antipsychotic medications, in the following frequencies: olanzapine (eight), fluphenazine (six), risperidone (six), haloperidol (four), clozapine (four), loxapine (three), quetiapine
(two), chlorpromazine (two), and ziprasidone (one). Chlorpromazine-equivalent dosages (Stoll, 1998) were used in
analyses. Antipsychotic medication dosage data were available for 27 patients: mean (SD) dosage was 533 (475)
chlorpromazine-equivalent milligrams.

Neuropsychological Variables
A neuropsychological battery was administered to subjects as part of a multidisciplinary, longitudinal study of
schizophrenia. The battery included measures of motor speed,
visual attention, and executive functions, assessed respectively by Finger Tapping, Trail Making, and Wisconsin Card
Sort (Lezak, 1995). These variables were selected because of
their previous associations with personality function (see
introduction), and because they were available for all study
subjects. Finger tapping performance was measured by the
number of taps accomplished within 10 seconds (average of
three trials). Performance on Trails A and B of the HalsteadReitan Battery was defined as time (seconds) to task completion. WCST performance was measured by counting categories
completed, perseverative responses, and errors (perseverative
and nonperseverative). Wechsler Memory Scale, Third Edition,
measures of immediate and delayed auditory and visual memory
and working memory (Wechsler, 1997) were available for a
subset (N 21) of patients.

Personality Measures
The NEO Five Factor Inventory (NEO-FFI), Form S
(Costa and McCrae, 1992), is a self-administered questionnaire consisting of 60 items rated on a 5-point response scale
(strongly disagree to strongly agree); it is a shortened
version of the NEO-PI, which has 180 items. Both instruments measure the dimensions of neuroticism, extraversion,
openness, agreeableness, and conscientiousness. Correlations
between the FFI and PI versions range from .77 to .92, and
coefficients for FFI scales range from .68 to .86 (Costa and
McCrae, 1992). Although originally developed for nonpsychiatric populations, the NEO-PI appears valid for patients
with schizophrenia (Kentros et al., 1997), and the FFI has
also been used to assess individuals with schizophrenia (Gurrera et al., 2000).

Statistical Analyses
Personality dimensions and profiles in each group were
compared by MANOVA before and after extracting personality
variance shared with neuropsychological task performance. For
the latter comparison, multiple linear regression was used to
compute residual scores for each NEO scale. Product moment
correlations between personality and neuropsychological variables, by group, supplemented this analysis.

716

The contributions of diagnosis and neuropsychological

measures to personality variance were then quantified by a
series of hierarchical (forced) step-wise linear regression
analyses, which permitted total variance for each dependent
variable to be uniquely partitioned among the independent
variables (Cohen and Cohen, 1983). Because the ratio of
initial independent variables (13) to subjects (75) was relatively high (statistically undesirable), principal components
analysis (PCA) with Equamax rotation was applied separately
to demographic (age, education, SES, PSES) and neuropsychological task variables to reduce the number of independent
variables in the equation; PCA component scores were substituted for independent variable scores in the regression
analyses. Independent variables were entered in three tiers, in
the following order: demographic, diagnostic, neuropsychological. Diagnosis was coded as 1 control, 2 patient.
Personality scale scores (dependent variables) were converted
to T scores using gender-specific normative FFI scale means
and standard deviations for healthy adults (Costa and McCrae, 1992). Residuals were plotted against predicted values
to evaluate model adequacy (Cohen and Cohen, 1983). The
unique contribution of each independent variable to each
personality scale was evaluated by comparing partial (rp)
and semipartial (rsp), correlations.
Principal components analysis yielded two demographic variable components with eigenvalues 1. The first
component (1.705) accounted for 42.6% of the total variance
and contained the heaviest loadings from educational level
(.921) and SES (.919). The second component (1.363)
accounted for 34.1% of the total variance with principal
loadings for age (.820) and PSES (.811). PCA of neuropsychological measures yielded three components, each corresponding to a different neuropsychological task: component 1
(4.088, 51.1% of total variance), WCST; component 2 (1.595,
19.9% of total variance), Finger Tapping; and component 3
(1.009, 12.6% of total variance), Trail Making. Demographic
and neuropsychological components thus accounted for
76.7% and 83.6%, respectively, of the total variance within
those variable tiers.
For all regression analyses, the F-to-enter probability
was .05, and the F-to-exclude probability was 0.1. Significance levels are two-tailed. Statistical computations were
performed by SPSS 11.5 except for T score conversions,
which were computed manually.

RESULTS
Neuropsychological Measures
Mean neuropsychological test scores differed significantly
between groups (MANOVA multivariate F8,66 8.94; p
0.001; all univariate F1,73 7.06; p 0.010). As expected,
patients performed significantly worse than controls on all measures. Medication dosage was not correlated with neuropsychological components (r .084; p 0.677).

Personality Measures
Personality scores differed significantly between subject groups (multivariate F5,69 4.29; p 0.002). Patients
scored higher on neuroticism (53.08 vs. 44.35; univariate
2005 Lippincott Williams & Wilkins

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

F 1,73 18.72; p 0.001) and lower on extraversion (49.85

vs. 55.01; F1,73 4.61; p 0.035) and agreeableness (46.30
vs. 52.38; F1,73 4.36; p 0.040). There were trends toward
lower mean openness (49.32 vs. 54.02; F1,73 3.07; p
0.084) and conscientiousness (46.14 vs. 50.01; F1,73 3.02;
p 0.086) scores in patients. Repeated measures MANOVA
found a statistically significant group-scale interaction (multivariate F4,70 5.29; p 0.001), indicating that NEO scale
profiles also differed between groups. Personality dimensions
were not associated with negative (.196 r .322; p
0.143) or positive (.176 r .050; p 0.458) symptoms.

Personality and Neuropsychological Function

Residual NEO scale scores, computed by extracting the
variance shared with neuropsychological measures, did not

Personality Changes in Schizophrenia

differ significantly between groups (F5,69 .308; p

0.906). Similarly, residual NEO scale profiles were not
significantly different between groups (F4,70 .357;
p 0.838).
All three hierarchical regression models accounted for
statistically significant amounts of variance in neuroticism
and agreeableness (Table 1, ANOVA tables on left).
For conscientiousness, only the third model (which included neuropsychological components) approached statistical significance.
Education/SES was negatively associated with neuroticism ( .374), but not after diagnosis was added (Table
1, right side). When the neuropsychological tier was added
(model 3), neither diagnosis nor demographic variables were

TABLE 1. Hierarchical Linear Regression Analyses

Sum of
Squares

df

Mean
Square

Neuroticism
1 Regression
Residual
2 Regression
Residual

940.84
5769.95
1428.03
5282.76

2
72
3
71

3 Regression
Residual

2272.26
4438.53

Agreeableness
1 Regression
Residual
2 Regression
Residual

R2

R2

Variable

470.42
80.14
476.01
74.40

5.870

0.004

.374

.140

.140

6.398

0.001

.461

.213

.073

6
68

378.71
65.27

5.802

0.000

.582

.339

.126

Educ/SES
Age/PSES
Educ/SES
Age/PSES
Diagnosis
Educ/SES
Age/PSES
Diagnosis
WCST
Finger Tapping
Trail Making

.374*
.021
.125
.014
.367*
.021
.152
.064
.244*
.202
.434*

0.001*
0.845
0.384
0.895
0.013*
0.880
0.160
0.691
0.041*
0.073
0.001*

1447.55
10363.30
1448.43
10362.42

2
72
3
71

723.78
143.94
482.81
145.95

5.029

0.009

.350

.123

.123

3.308

0.025

.350

.123

.000

3 Regression
Residual

2444.21
9366.64

6
68

407.37
137.74

2.957

0.013

.455

.207

.084

Educ/SES
Age/PSES
Educ/SES
Age/PSES
Diagnosis
Educ/SES
Age/PSES
Diagnosis
WCST
Finger Tapping
Trail Making

.338*
.090
.330*
.090
.012
.257
.191
.229
.133
.163
.369*

0.003*
0.416
0.032*
0.421
0.938
0.094
0.108
0.195
0.302
0.186
0.011*

Conscientiousness
1 Regression
Residual
2 Regression
Residual

431.17
6367.70
515.37
6283.49

2
72
3
71

215.58
88.44
171.79
88.50

2.438

0.095

.252

.063

.063

1.941

0.131

.275

.076

.012

3 Regression
Residual

1089.86
5709.01

6
68

181.64
83.96

2.164

0.057

.400

.160

.084

Educ/SES
Age/PSES
Educ/SES
Age/PSES
Diagnosis
Educ/SES
Age/PSES
Diagnosis
WCST
Finger Tapping
Trail Making

.168
.188
.065
.185
.151
.010
.288*
.042
.288*
.238
.133

0.146
0.104
0.676
0.110
0.333
0.948
0.020*
0.815
0.032*
0.062
0.362

Model

*Statistical significance at 0.05 level.

2005 Lippincott Williams & Wilkins

717

Gurrera et al.

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

significantly related to neuroticism, but worse performance

on WCST ( .244) and Trail Making ( .434) were, and
uniquely contributed more than a third of the total variance
accounted for by the full regression model.
Education/SES was positively associated with agreeableness in the first ( .338) and second ( .330)
models, whereas diagnosis accounted for no variance (Table
1, right side). When neuropsychological tasks were included,
neither demographic nor diagnosis variables were significantly related to this dimension, but better Trail Making
performance was significantly associated with higher agreeableness scores ( .369).
First and second hierarchical regression models yielded
no statistically significant relationships with conscientiousness, but the full model approached statistical significance
(F 2.164; p 0.057; Table 1, left side). Higher scores were
associated with older age and lower parental SES ( .288),
and better WCST ( .288) and Finger Tapping ( .238)
performance (Table 1, right side).
Extraversion was consistently positively associated
with education/SES (1.982 F 5.946; 0.004 p 0.080;
and .368 .377; 0.001 p 0.021), whereas openness
was consistently negatively associated with age/PSES
(2.308 F 4.675; 0.012 p 0.044; and .271
.285; 0.012 p 0.027) across all three regression
models. In other words, higher extraversion scores were
associated with greater educational achievement and better
SES, while higher openness scores were associated with
younger age and better parental SES. Regression models
accounted for 14.2% to 14.9% of total extraversion variance,
and 11.5% to 16.9% of total openness variance. Neuropsychological task performance was not related to either
dimension.
Partial (rp) and semipartial (rsp) correlations produced
by the third regression model for each personality dimension
were compared. Partial and semipartial correlations are derived directly from their corresponding regression analyses
and represent different ways of partitioning covariance. Because they do not represent separate statistical tests (the same
significance level applies to all partial coefficients resulting
from the same regression analysis), they permit a more detailed
analysis of covariance without increasing the likelihood of a
type I error.
Poorer performance on Trail Making and WCST was
significantly associated with higher neuroticism scores. Whereas
the magnitudes of r and rsp for each of these variables were very
similar (.453 vs. .333 and .240 vs. .205, respectively), for
diagnosis, r was much larger than rsp (.452 vs. .039), indicating that the association between diagnosis and neuroticism
was redundant with those with Trail Making and WCST.
Trail Making uniquely accounted for 11.1%, and WCST
performance 4.2%, of neuroticism variance. Trail Making
was also significantly associated with agreeableness (rsp
.283; p 0.011), uniquely sharing 8.0% of its variance.
WCST (rsp .243; p 0.032) and age/PSES (rsp .265;
p 0.020) uniquely shared 5.9% and 7% of conscientiousness variance, respectively.

718

In all three regression models, higher education and

SES were consistently related to higher extraversion scores,
whereas younger age and higher PSES were significantly
related to higher openness scores. In each case, r and rsp had
similar magnitudes (.368 vs. .265 and .282 vs. .250,
respectively) but r and rsp for diagnosis did not (.244 vs.
.042 and .201 vs. .109). Each uniquely accounted for
similar amounts of personality dimension variance (7.0%
and 6.2%).
Personality dimensions correlated significantly with individual neuropsychological measures in each subject group
(Table 2), indicating that task performance was significantly
associated with personality dimensions in both patients and
controls. In fact, of the 18 statistically significant correlations
obtained collectively, 11 (61.1%) were found in the control
group; at an level of .05, chance alone would account for
only two of these. Neuroticism showed the most similar
correlation pattern across groups, whereas patterns for agreeableness and conscientiousness were more divergent. All
correlations were modest (.440 r .431).
In patients whose memory function had been assessed
(N 21), immediate visual memory was correlated with
openness (r .501; p 0.021) and conscientiousness (r
.494; p 0.023), but immediate auditory memory was not
associated with any personality dimension (r .299; p
0.188). Similarly, delayed visual memory correlated with
neuroticism (r .443; p 0.044), openness (r .606; p
0.004), and agreeableness (r .442; p 0.045), but
TABLE 2. Intragroup Correlations Between Personality T
Scores and Individual Neuropsychological Task Performance
Variables
N
Patient group
Finger Tapping, right hand
Finger Tapping, left hand
Trail Making A (seconds)
Trail Making B (seconds)
WCST categories
completed
WCST perseverative
responses
WCST nonperseverative
errors
WCST perseverative errors
Control group
Finger Tapping, right hand
Finger Tapping, left hand
Trail Making A, (seconds)
Trail Making B, (seconds)
WCST, categories
completed
WCST, perseverative
responses
WCST, nonperseverative
errors
WCST, perseverative errors

.125 .052 .272 .036 .065

.419* .131 .354* .101
.191
.386* .040
.186 .329
.085
.239 .031 .060 .234
.151
.128 .138
.189 .003 .043
.078
.431*
.028

.189 .440*
.023

.355* .185

.092 .341 .056

.196 .398*

.324 .159

.047
.140
.271
.241
.224
.087 .029
.197
.049
.204
.250 .252 .039 .117 .098
.304* .115
.059 .234 .319*
.340* .235
.289* .197
.307*
.259 .333* .203 .345* .167
.322* .151 .184 .384* .245
.277 .288* .175 .312* .189

*Statistical significance at 0.05 level.

2005 Lippincott Williams & Wilkins

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

delayed auditory memory was not associated with personality

measures (r .248; p 0.278). Working memory correlated only with openness (r .457; p 0.037). No
correlations were significant at a Bonferroni probability of
.002 (corresponding to an level of .05 and 25 correlation pairs).

DISCUSSION
The finding that patients with clinically stable schizophrenia have significantly altered personality profiles
higher neuroticism and lower scores on other NEO scalesis
consistent with prior studies using healthy comparison groups
or normative data (Gurrera et al., 2000). A principal finding
of the present study is that group personality differences
disappear when the variance shared with neuropsychological
task performance is statistically controlled, indicating that
personality differences may be related to abnormal cognition
associated with schizophrenia.
A second major finding of the present study is that
diagnosis, in itself, uniquely accounted for very little personality variance. Neuroticism was the only dimension to which
diagnosis made a statistically significant contribution, but
after neuropsychological variables were added to the regression model, diagnosis was no longer significantly associated
even with this dimension. Since patients with more severe
illness will also tend to have more extreme personality
deviations and more impaired cognition, one of the risks of a
cross-sectional study is that simultaneous variation within
these domains will be misinterpreted as being causally related
when it is not. A major strength of the multiple linear
regression technique, supplemented by analysis of partial and
semipartial correlation coefficients, is that this collinearity
can be statistically controlled so that unique independentdependent variable relationships can be measured. If diagnosis had been the principal source of variance for both personality and task performance in this study (i.e., patients had
schizophrenia-related altered personality and also schizophrenia-related impaired cognition), diagnosis would have been
related consistently to personality across all regression models. In addition, its semipartial correlation coefficients would
have been similar in magnitude to its first-order correlation
coefficients and substantially larger than the semipartial correlation coefficients for task performance. However, the opposite pattern was found, suggesting that neuropsychological
task performance is, at least quantitatively, a much more
important source of personality variance than is diagnosis.
The observation that neuropsychological task performance variables and personality dimensions were correlated
more frequently in the control group supports the interpretation that diagnosis per se is not the principal source of
personality variance in schizophrenia. Rather, these data
seem more consistent with the view that neuropsychological
function contributes to personality variation, regardless of
diagnosis. Correlations were modest in size across groups, the
pattern of associations between tasks and personality dimensions showed similarities and differences between groups,
and most of the total personality variance in this sample was
not explained by the independent variables included in this
2005 Lippincott Williams & Wilkins

Personality Changes in Schizophrenia

study. Therefore, the relationship of neuropsychological

function to individual differences in personality is likely
complex, with the independent variables examined in the
present study comprising only a small part of the picture.
WCST and Trail Making significantly predicted neuroticism, Trail Making significantly predicted agreeableness,
and there was a trend for WCST to predict conscientiousness.
Thus, each of these personality dimensions was associated
with performance on a different subset of neuropsychological
tasks. Neuropsychological performance did not predict extraversion or openness. These results are consistent with our
hypothesis that personality dimensions have distinct relationships with the neuropsychological functions indexed by the
tasks in this study.
This study is among the first to use statistical methods
to explore the relationship between personality dimensions
and neuropsychological deficits in schizophrenia. One previous study (Lysaker et al., 1998) of patients with schizophrenia spectrum psychoses found no correlations between a
different neuroticism measure (Eysencks EPQ-Neuroticism)
and WCST performance, but EPQ-Extraversion was negatively correlated with perseverative errors and positively
correlated with categories completed. In a subsequent and
larger study (Lysaker et al., 1999), high neuroticism patients
made significantly more perseverative errors and completed
fewer WCST categories. The WCST assesses functions that
are believed to depend in large part on diverse frontal lobe
routines (Golden et al., 1998). An apparent neurogenic frontal
lobe syndrome characterized by impaired executive control is
associated with elevated neuroticism and reduced agreeableness and conscientiousness in multiple sclerosis patients
(Benedict et al., 2001). Thus, this study broadly replicates and
extends previous findings that diminished executive function
is associated with higher neuroticism scores and lower conscientiousness; in addition, these results indicate that impaired visual attention is related to lower agreeableness.
In a subset of patients, visual but not auditory memory
was correlated with several personality dimensions. However, whereas correlations with openness and conscientiousness were consistent with an overall pattern of better neuropsychological performance being associated with healthier
personality scores, other visual memory correlations contradicted this pattern. Also, no subset correlation attained Bonferroni-adjusted statistical significance. Thus, the meaning
of these findings is unclear, and they should be regarded
cautiously.
This study has a number of limitations that make it
advisable to consider the results preliminary. Despite our
relatively successful efforts to recruit female patients, such
that the proportion of female patients in the present study was
substantially higher than their representation among VA outpatients, their comparatively small number precluded an
analysis of potential gender effects, which may be important.
The neuropsychological measures examined in the
present study were selected from among those collected for a
larger, multicenter study of schizophrenia, and selection was
based on considerations of data availability (study size) and
relevance to the study hypotheses. To ensure an adequately

719

Gurrera et al.

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

large sample, only some neuropsychological task data could

be included. The absence of more comprehensive neuropsychological performance data (e.g., memory) in the regression
models does not diminish the principal finding of this study
that personality differences between patients with schizophrenia and controls disappear when neuropsychological performance is statistically controlled, and that diagnosis makes
little, if any, measurable unique contribution to personality
differences in patients and controls. On the contrary, adding
more neuropsychological variables to the regression models
would most likely have reduced the relative contribution of
diagnosis even further. However, a more comprehensive set
of neuropsychological measures would provide a more complete and accurate picture of how neuropsychological function relates to personality.
The cross-sectional design of this study provides no
basis for conclusions about cause-and-effect relationships
between neuropsychological function and personality variation. The statistical analyses indicate that a diagnosis of
schizophrenia in itself does not contribute measurably to
personality differences in this disorder, whereas neuropsychological task performance deficits do account for significant amounts of variance in at least some personality dimensions. Longitudinal studies over time would provide a clearer
picture of how these two domains are related, and how each
impacts the other.
Personality assessment was based on subjects responses to questionnaires, so variability in individual response style and accuracy may have distorted the results.
However, individual inconsistencies would have tended to
obscure any relationships actually present. The ratio of sample size to independent variables (75/6, or 12.5) is slightly
less than the minimum of 15 required (Stevens, 2002) for
reliable cross-validation (generalization to other population
samples), so these results need to be confirmed in a larger
sample.
No relationship between medication dosage and neuropsychological task performance was found. Others have also
failed to demonstrate a consistent antipsychotic medication
effect on motor performance in schizophrenia (Goode et al.,
1981; Vrtunski et al., 1989). If present, such an effect would
tend to obscure, rather than account for, relationships between personality dimensions and motor performance. There
was also no evidence of any systematic relationship between
schizophrenia symptoms and personality dimensions. However, schizophrenia symptom data were available for only
66.7% to 73.3% of patients, so the possibility of a type II
error exists and is another reason the present findings should
be considered preliminary.

CONCLUSION
These data provide preliminary evidence that neuropsychological performance deficits associated with schizophrenia may contribute to personality alterations that also characterize this disorder. Individuals personal background and
educational experience also appear to contribute significantly
to other personality features, such that individual differences
in personal background and neuropsychological function dis-

720

tinctly affect personality dimensions. These results also suggest that the relationships between neuropsychological function and personality may be independent of diagnosis.
REFERENCES
Addington J, McCleary L, Munroe-Blum H (1998) Relationship between
cognitive and social dysfunction in schizophrenia. Schizophr Res. 34:
59 66.
Andreasen NC (1984a) Scale for the Assessment of Negative Symptoms
(SANS). University of Iowa: Iowa City.
Andreasen NC (1984b) Scale for the Assessment of Positive Symptoms
(SAPS). University of Iowa: Iowa City.
Benedict RH, Priore RL, Miller C, Munschauer F, Jacobs L (2001) Personality disorder in multiple sclerosis correlates with cognitive impairment.
J Neuropsychiatry Clin Neurosci. 13:70 76.
Berenbaum H, Fujita F (1994) Schizophrenia and personality: Exploring the
boundaries and connections between vulnerability and outcome. J Abnorm
Psychol. 103:148 158.
Brebner J, Stough C (1995) Theoretical and empirical relationships between
personality and intelligence. In DH Saklofske, M Zeidner (Eds), International Handbook of Personality and Intelligence (Chapter 16, pp 321
347). New York: Plenum Press.
Cohen J, Cohen P (1983) Applied Multiple Regression/Correlation Analysis
for the Behavioral Sciences (pp 79 132). Hillsdale, New Jersey: Lawrence
Erlbaum Associates.
Costa PT Jr, McCrae RR (1992) NEO PI-R: Professional Manual (Revised
NEO Personality Inventory (NEO PI-R) and NEO Five-Factor Inventory
(NEO-FFI)). Odessa (FL): Psychological Assessment Resources, Inc.
Cuesta MJ, Peralta V, Caro F (1999) Premorbid personality in psychoses.
Schizophr Bull. 25:801 811.
Cuesta MJ, Peralta V, Zarzuela A (2001) Are personality traits associated
with cognitive disturbance in psychosis? Schizophr Res. 51:109 117.
Franke P, Maier W, Hardt J, Hain C, Cornblatt BA (1994) Attentional
abilities and measures of schizotypy: Their variation and covariation in
schizophrenic patients, their siblings and normal control subjects. Psychiatry Res. 54:259 272.
Golden CJ, Kushner T, Lee B, McMorrow MA (1998) Searching for the
meaning of the Category Test and the Wisconsin Cart Sort Test: A
comparative analysis. Int J Neurosci. 93:141150.
Goode DJ, Manning AA, Middleton JF, Williams B (1981) Fine motor
performance before and after treatment in schizophrenic and schizoaffective patients. Psychiatry Res. 5:247255.
Green MF (1996) What are the functional consequences of neurocognitive
deficits in schizophrenia? Am J Psychiatry. 153:321330.
Griffith JM, Adler LE, Freedman R (1994) Fine motor performance in
schizophrenia. Neuropsychobiology. 29:179 184.
Gross G, Huber G (1993) Premorbid personality in schizophrenia: The
contribution of European long-term studies. Neurol Psychiatry Brain Res.
2:14 20.
Gurrera RJ, Nestor PG, ODonnell BF (2000) Personality traits in schizophrenia: Comparison with a community sample. J Nerv Ment Dis. 188:
3135.
Harris JA, Vernon PA, Jang KL (1998) A multivariate genetic analysis of
correlations between intelligence and personality. Dev Neuropsychol.
14:127142.
Hollingshead AB (1965) Two Factor Index of Social Position. New Haven
(CT): Yale University Press.
John OP, Srivastava S (1999) The Big Five trait taxonomy: History, measurement and theoretical perspectives. In LA Pervin, OP John (Eds),
Handbook of Personality: Theory and Research (2nd ed, Chapter 4, pp
102138). New York: Guilford Press.
Kentros M, Smith TE, Hull J, McKee M, Terkelsen K, Capalbo C (1997)
Stability of personality traits in schizophrenia and schizoaffective disorder: A pilot project. J Nerv Ment Dis. 185:549 555.
Lezak MD (1995). Neuropsychological Assessment (3rd ed). New York:
Oxford University Press.
Lyons MJ, Jerskey BA (2002) Personality disorders: Epidemiological findings, methods and concepts. In MT Tsuang, M Tohen (Eds), Textbook in
Psychiatric Epidemiology (2nd ed, Chapter 21, pp 563599). New York:
Wiley-Liss, Inc.

2005 Lippincott Williams & Wilkins

The Journal of Nervous and Mental Disease Volume 193, Number 11, November 2005

Lysaker PH, Bell MD, Kaplan E, Bryson G (1998) Personality and

psychosocial dysfunction in schizophrenia: The association of extraversion and neuroticism to deficits in work performance. Psychiatry
Res. 80:61 68.
Lysaker PH, Bell MD, Kaplan E, Greig TC, Bryson GJ (1999) Personality
and psychopathology in schizophrenia: The association between personality traits and symptoms. Psychiatry. 62:36 48.
Matsui M, Gur RC, Turetsky BI, Yan MX-H, Gur RE (2000) The relation
between tendency for psychopathology and reduced frontal brain volume in
healthy people. Neuropsychiatry Neuropsychol Behav Neurol. 13:155162.
Morice R, Delahunty A (1996) Frontal/executive impairments in schizophrenia. Schizophr Bull. 22:125137.
Nuechterlein KH, Asarnow RF (1989) Perception and cognition. In: HI
Kaplan, BJ Sadock (Eds), Comprehensive Textbook of Psychiatry IV
(5th ed, Vol 1, Chapter 3, pp 241256). Baltimore (MD): Williams &
Wilkins.
Rund BR, Borg NE (1999) Cognitive deficits and cognitive training in
schizophrenic patients: A review. Acta Psychiatr Scand. 100:8595.
Smith TE, Grabstein E, Kentros M, Kulchycky S, Borgaro S (1996) Axis II
diagnoses and treatment refractoriness in schizophrenia. Psychiatr Q.
67:51 64.
Smith TE, Shea MT, Schooler NR, Levin H, Deutsch A, Grabstein E (1995)
Studies of schizophrenia: Personality traits in schizophrenia. Psychiatry.
58:99 112.

2005 Lippincott Williams & Wilkins

Personality Changes in Schizophrenia

Solano JJR, De Chavez MG (2000) Premorbid personality disorders in

schizophrenia. Schizophr Res. 44:137144.
Stevens JP (2002) Applied Multivariate Statistics for the Social Sciences
(4th ed). Mahwah, New Jersey: Lawrence Erlbaum Associates.
Stoll AL (1998) The Psychopharmacology Reference Card: Antipsychotic
Treatment Guide. Belmont (MA): McLean Hospital.
Stone MH (1993) Abnormalities of Personality: Within and Beyond the
Realm of Treatment. New York: WW Norton & Co.
Tien AY, Costa PT, Eaton WW (1992) Covariance of personality, neurocognition and schizophrenia spectrum traits in the community. Schizophr
Res. 7:149 158.
van Os J, Jones PB (2001) Neuroticism as a risk factor for schizophrenia.
Psychol Med. 31:1129 1134.
Vrtunski PB, Simpson DM, Meltzer HY (1989) Voluntary movement dysfunction in schizophrenics. Biol Psychiatry. 25:529 539.
Wechsler D (1997) The Wechsler Memory Scale Revised. New York: Psychological Corp.
Wolff AL, ODriscoll GA (1999) Motor deficits and schizophrenia: the
evidence from neuroleptic-naive patients and populations at risk. J Psychiatry Neurosci. 24:304 314.
Zeidner M (1995) Personality trait correlates of intelligence. In DH Saklofske, M Zeidner (Eds), International Handbook of Personality and
Intelligence (Chapter 15, pp 299 319). New York: Plenum Press.

721