Journal of Abnormal Psychology 2003, Vol. 112, No. 3, 403– 414

Copyright 2003 by the American Psychological Association, Inc. 0021-843X/03/$12.00 DOI: 10.1037/0021-843X.112.3.403

Antisaccade Performance of 1,273 Men: Effects of Schizotypy, Anxiety, and Depression Nikolaos Smyrnis, Ioannis Evdokimidis, Nicholas C. Stefanis, Dimitrios Avramopoulos, and Theodoros S. Constantinidis

Alexios Stavropoulos “251” Hellenic Air Force General Hospital

National University of Athens

Costas N. Stefanis National University of Athens

A total of 1,273 conscripts of the Greek Air Force performed antisaccades and completed self-reporting questionnaires measuring schizotypy and current state-dependent psychopathology. Only 1.0% of variability in antisaccade performance indices was related to psychometric scores in the population and could be attributed more to current state-dependent symptoms such as anxiety rather than to schizotypy. In contrast, a specific increase of error rate and response latency variability and a high correlation of these 2 variables was observed in a group with very high schizotypy scores. This effect was independent of anxiety and depression, suggesting that a specific group of psychosis-prone individuals has a characteristic deviance in antisaccade performance that is not present in the general population.

1995; Sweeney, Mintun, & Kwee, 1996), event-related potential studies (Evdokimidis, Liakopoulos, Constantinidis, & Papageorgiou, 1996; Everling, Krappmann, & Flohr, 1997), transcranial magnetic stimulation studies (Muri, Hess, & Meienberg, 1991), and animal studies (Funahashi, Chafee, & Goldman-Rakic, 1993; Schlag-Rey, Amador, Sanchez, & Schlag, 1997) point to the importance of frontal areas in the performance of the antisaccade task. It has been shown that patients with schizophrenia produce significantly more errors in the antisaccade task than healthy controls (Clementz, McDowell, & Jisook, 1994; Fukushima et al., 1988; Fukushima, Fukushima, Miyasaka, & Yamashita, 1994; Fukushima, Morita, Fukushima, Chiba, & Yamashita, 1990; Katsanis, Kortenkamp, Iacono, & Grove, 1997; McDowell & Clementz, 1997; Sereno & Holzman, 1995). Concerning the response latency for making an antisaccade, some studies report longer mean latencies for correct antisaccades for patients with schizophrenia compared with controls (Fukushima et al., 1988, 1990, 1994; Sereno & Holzman, 1995), whereas other studies report no difference in mean latency between these groups (Clementz et al., 1994; Katsanis et al., 1997). In addition, deficits of patients with schizophrenia in the performance of this task have been linked to dorsolateral prefrontal cortical hypoactivity (Nakashima et al., 1994), frontal ventricular enlargement (Fukushima et al., 1988), and deficits in the performance of executive tasks such as the Wisconsin Card Sorting Test (Rosse, Schwartz, Kim, & Deutsch, 1993). In summary, there is evidence that at least a subgroup of patients with schizophrenia make more errors in the antisaccade task and maybe are slower to respond. The study of antisaccades was introduced in the investigation for potential biological markers of psychosis proneness. Some studies reported that first-degree relatives of persons with

In the antisaccade task introduced by Hallet (1978), participants were instructed to look in the opposite direction of a visually presented stimulus. Normal individuals can perform this task successfully, although in a small percentage of error (PE) trials, normal individuals make a saccade toward the target (prosaccade) and then a saccade in the opposite direction (antisaccade). Evidence from lesion studies (Guitton, Buchtel, & Douglas, 1985; Pierrot-Deseilligny, Rivaud, Gaymard, & Agyd, 1991), studies in normal humans using functional brain imaging (O’Driscoll et al.,

Nikolaos Smyrnis, University Mental Health Research Institute, and Cognition and Action Group, Neurology Department, National University of Athens, Athens, Greece; Ioannis Evdokimidis, Cognition and Action Group, Neurology Department, National University of Athens; Nicholas C. Stefanis, Dimitrios Avramopoulos, Theodoros S. Constantinidis, and Costas N. Stefanis, University Mental Health Research Institute, National University of Athens; Alexios Stavropoulos, Hellenic Air Force Medical Directorate, “251” Hellenic Air Force General Hospital, Athens, Greece. A preliminary report of this article was presented at the second annual International Congress on Hormones, Brain and Neuropsychopharmacology, Rhodes, Greece, August 2001. This work was supported by Grant EKBAN 97 to Costas N. Stefanis from the General Secretariat of Research and Technology of the Greek Ministry of Development. The technical support for this project was provided by Intrasoft Co. We thank the following colleagues, in alphabetical order, who helped in data acquisition and preprocessing: Katerina Eustratiadi, Ioannis Giouzelis, Georgios Kastrinakis, Emanouil Kattoulas, Catherine Paximadis, and Christos Theleritis. We also thank the commanders in chief and the military personnel of the “124” Base for Basic Training of the Hellenic Air Force. Correspondence concerning this article should be addressed to Nikolaos Smyrnis, Neurology Department, National University of Athens, Aeginition Hospital, 72 Vassilisis Sofias Avenue, Athens Gr-11528, Greece. E-mail: [email protected] 403

404

SMYRNIS ET AL.

schizophrenia produced more antisaccade errors than controls, but in general, first-degree relatives made less errors than persons with schizophrenia (Clementz et al., 1994; Curtis, Calkins, Grove, Feil, & Iacono, 2001; Karoumi et al., 2001; Katsanis et al., 1997; McDowell & Clementz, 1997; McDowell, Myles-Worsley, Coon, Byerley, & Clementz, 1999), whereas others reported no difference (Crawford et al., 1998; Thaker, Cassady, Adami, Moran, & Ross, 1996) or a difference that was present only when first-degree relatives had also schizophrenia spectrum symptoms (Thaker et al., 2000). A finer distinction was proposed on the basis of the observation that the firstdegree relatives of patients with schizophrenia who had a significantly higher error rate than controls in the antisaccade task also had a significantly higher error rate, whereas the first-degree relatives of patients with schizophrenia who had lower error rates also had lower error rates (no difference from controls; Crawford et al., 1998; Curtis et al., 2001; McDowell & Clementz, 1997; McDowell et al., 1999). The antisaccade mean latency was reported to be longer for first-degree relatives of persons with schizophrenia in some studies (McDowell et al., 1999; Thaker et al., 1996, 2000), whereas other studies report no difference of mean latency between normal controls and first-degree relatives (Curtis et al., 2001; Karoumi et al., 2001). These differences in performance, although attenuated, were also reported to be present in their first-degree relatives, leading to the hypothesis that the phenotype of vulnerable individuals to develop psychosis could also include deficits in antisaccade performance (McDowell et al., 1999). Another approach to study populations that are at risk for developing schizophrenia is to identify individuals with a Diagnostic Statistical Manual of Mental Disorders (DSM–IV; American Psychiatric Association, 1994) Axis II schizotypal personality disorder or individuals within the general population that score highly on self-reporting questionnaires that measure schizotypy. These individuals are at greater risk of developing psychosis than the general population (Chapman, Chapman, Kwapil, Eckbland, & Zinser, 1994). Two major theoretical approaches exist to explain the link between schizotypy and schizophrenia. The first theory of schizotaxia (Meehl, 1989) proposes that schizotaxia is a conjectured neural integrative defect due to a dominant schizogene that gives rise to the schizotypal personality. This genetic profile in synergy with other polygenic potentiators and adverse life experiences gives rise in a small percentage of these individuals to the clinical syndrome of schizophrenia. In a theory closely related to Meehl’s theory of schizotaxia, Matthysse, Holzman, and Lange (1986) proposed that schizotypal traits are some of the phenotypic expressions of an underlying vulnerability to schizophrenia, whereas other such expressions that are independent from schizotypy are the deficits in tests of neurocognitive function and eye-movement tasks like the antisaccade task. These theories predict that in a distinctive group of individuals with high schizotypy who have not developed schizophrenia, there would be indications of eyemovement function deficits similar to those observed in schizophrenia. In agreement with this hypothesis, it was reported that individuals with high schizotypy, as measured in self-reporting questionnaires, produce more errors in their antisaccade performance than normal controls and don’t differ in mean response

latency (Gooding, 1999; O’Driscoll, Lenzenweger, & Holzman, 1998). The second theoretical approach to schizotypy favored by Eysenck (Eysenck & Eysenck, 1976) states that personality traits such as those that define psychoticism are a continuum from health to schizophrenia with no need to introduce arbitrary cutoff points above which schizotypy lies as a different entity (Claridge, 1994). According to this view, certain dimensions of personality are to be found in the general population and their extremes lead to the symptoms of a disease state such as schizophrenia (van Os, Verdoux, Bijl, & Ravelli, 1999). Within this framework, schizotypy is decomposed in dimensions using a factor model, and the different factors identified in samples of the general population (e.g., a positive symptom factor loading on unusual perceptual experiences) reflect the corresponding factor structure of the clinically identified syndrome of schizophrenia (in the previous example, the positive symptoms of schizophrenia). So far, all studies of antisaccade performance in patients with schizophrenia have been retrospective. If the antisaccade performance deficits are liability indicators for the manifestation of the disease, they should probably precede and be unrelated to the onset of the clinical manifestations. In the past 2 years, we completed the first (screening) phase of a large study of eye-movement and cognitive tests for a sample of 2,075 young men in the Greek Air Force (Athens Study of Psychosis Proneness and Incidence of Schizophrenia; ASPIS). The study is prospective in nature and longitudinal, designed to follow this population and investigate whether oculomotor functions such as the antisaccade and cognitive functions have predictive value for the later development of a clinical psychiatric disorder. In this study, we provide population data on the antisaccade task. Although these data are limited by the specific task conditions, age and gender, they provide the first of its kind database of oculomotor function data that could be of interest in the growing field of study of antisaccade performance in different clinical populations. In this article, we present results from the study of the relation of antisaccade task performance to psychopathological parameters such as schizotypy anxiety and depression in a large population. Previous studies have found antisaccade performance differences in individuals with very high schizotypy scores in self-reporting questionnaires compared with controls (Gooding, 1999; O’Driscoll et al., 1998). Our study aimed to extend the investigation of this relation in a large population. This approach gave us the opportunity to study two opponent hypotheses on the effects of schizotypy on eye-movement performance—that of a correlation across the population (favoring the dimensional theory of schizotypy) and that of a specific relation within a high risk group (favoring the categorical theory of schizotypy). Furthermore, the concurrent examination of the effects on antisaccade performance of state-dependent psychopathological symptoms such as anxiety and depression provided us with the opportunity to examine the specificity of the relation of schizotypy and antisaccade performance. This issue is of importance in view of the hypothesis that schizotypy characteristics reflect traits that predispose individuals for the development of psychosis, thus ideally, one would expect antisaccade performance to be specifically related to schizotypy and not to anxiety and depression symptoms in the population.

ANTISACCADE PERFORMANCE OF 1,273 MEN

Method Participants and Materials A sample of 2,130 young male participants aged 18⫺24 were recruited from the Greek Air Force. A total of 2,075 individuals (97.4% of those approached) agreed to participate in the study after giving written informed consent. Each participant performed eye-movement tasks (smooth eye pursuit, saccade, antisaccade, visual fixation) and cognitive tasks (Continuous Performance Task [CPT-IP]; Cornblatt, Risch, Faris, Friedman, & Erlenmeyer-Kimling, 1998; verbal n-back test, spatial n-back test; Gevins et al., 1996). We also administered to 1,657 of these participants (80.0% of participants) a battery of self-rated questionnaires assessing, among other variables, (a) schizotypy with the Perceptual Aberration Scale (PAS; Chapman et al., 1978) and the Schizotypal Personality Questionnaire (SPQ; Raine, 1991); (b) current psychopathology with the Hopkins Symptom Checklist 90 —revised scale (SCL90⫺R; Derogatis, Lipman, Rickels, Uhlenhuth, & Covi, 1974), which was translated and standardized for the Greek population (Ntonias et al., 1990); and (c) personality characteristics with the Temperament and Character Inventory—Revised (TCI-140⫺R; Cloninger, Svrakic, & Przybeck, 1993). Confirmatory factor analysis revealed that a four-dimensional model of SPQ best fitted the observed data (Stefanis et al., 2003). The first factor loaded on the subscales of odd beliefs and odd perceptual experiences (F cognitive perceptual, cog-per). The second factor loaded on suspiciousness, social anxiety, no close friends, and constricted affect (F negative). The third factor loaded on odd behavior and odd speech (F disorganization), and the fourth factor loaded on social anxiety, suspiciousness, and ideas of reference (F paranoid). A follow-up phase was initiated for all participants to record those participants that were admitted in the neurology or psychiatry services of the Air Force General Hospital during service time. During this phase of the study, 43 participants with a neurological or psychiatric disorder were identified. Most of these individuals had a diagnosis before entering the service, but they did not report it during the initial medical assessment of conscripts entering military service. This assessment is a standard interview-based procedure performed by a team of medical doctors of all specialties (military personnel). Among the 43 participants, 7 were diagnosed with psychosis. These 43 individuals were not excluded from the population data set in the presentation that follows, but their exclusion did not change any of the findings. A total of 1,629 (98.3% of the 1,657) participants completed the psychometric evaluation battery and complied with the criteria for successful performance of the antisaccade task (see below for definition of these criteria). The four validity items of the TCI-140⫺R questionnaire were used to exclude 384 participants (23.5%) who responded incorrectly in at least one of these items, thus ensuring the collaboration of the responders in these self-report scales. These items were of the type please answer this question by ticking box 4, and a wrong answer would indicate that the individual was responding at random. The analyses were performed on the remaining population of 1,273 participants (valid responders; M ⫽ 20.9 years, SD ⫽ 1.9 years).

to the participant, and the upper part of the monitor was used to project the eye-position data so that the experimenter could have online control of the quality of the eye-movement signal. Eye movements were recorded from the right eye only using the IRIS (SCALAR MEDICAL BV, Delft, the Netherlands) infrared device. A 12-bit A/D converter was used for data acquisition (Advantech PC-Lab Card 818L; Advantech Co., Ltd., Taipei, Taiwan). Eye-movement data were sampled at 600 Hz and stored in the computer hard drive for offline data processing.

Procedure Each trial of the antisaccade task started with the appearance of a central fixation target (white cross; 0.5° ⫻ 0.5°). After a variable period of 1⫺2 s, the central target was extinguished and an identical target appeared randomly at one of nine target amplitudes from the center (2°⫺10° at 1° intervals) and in one of two directions (left or right). The participants were instructed to make an eye movement as quickly as possible to the opposite direction from that of the peripheral target and to hold that position until the central fixation target reappeared 1.5 s later. Each participant completed a block of 90 experimental trials. An additional few trials were administered to familiarize the participants with the task.

Antisaccade Indices An interactive computer program (created using the TestPoint; Capital Equipment Corporation, Billerica, MA) was used for detection and measurement of saccades from the eye-movement record. The first derivative of the position record was used to calculate the instantaneous velocity curve. This curve in turn was used to detect the onset of saccadic eye movements using the criterion that a number of five consecutive velocity values (8-ms duration at 600 Hz) were above a predefined noise level (the noise level was determined by taking the root mean square of the signal in 15 windows of 20 values each, covering the first 500 ms of the 1⫺2 s period of central fixation, and then taking the median value of these 15 values). The program used the return of the instantaneous velocity to the noise level to mark the end of the saccade. A criterion of change of velocity sign was used to exclude blinks (which present as increases of velocity in one direction followed immediately by increases in the opposite direction). Finally, the program marked saccadic eye movements and blinks on the position record for inspection by a human observer. Figure 1 presents traces of individual saccadic eye-movement records from 1 participant of whom the program marked the onset and end of saccades. In cases of discrepancy, the human observer could switch to a manual identification of events on the eye-movement position record. For each individual, the following antisaccade performance indices were evaluated: 1.

PEs. For the present analysis, an error was defined as a movement in the direction of the peripheral target after its presentation followed by a corrective movement (in the direction opposite of the target), which occurred almost always (in over 99.0% of the error trials). The distribution of PEs was normalized via an arcsine transformation for percentages, which was used for all subsequent analyses (Snedecor & Cochran, 1980). In the analysis that follows, PE denotes original data and PE(t) denotes arcsinetransformed data.

2.

Mean latency for correct antisaccades (LCM).

3.

Standard deviation of the latency for correct antisaccades (LCSD).

4.

Mean latency for error prosaccades (LEM).

5.

Standard deviation of the latency for error prosaccades (LESD).

Apparatus All participants were tested in a military facility during the first 2 weeks of their service in the Greek Air Force. Five personal computer setups were used for measuring the eye-movement performance of 5 participants simultaneously. Each eye-movement setup consisted of an adjustable chair for the participant, a specially designed table that had a head holder mounted on one end and a 17.00 in. (43.18 cm) computer monitor mounted on the other end (1 m distance from the head holder). A steel frame at the edges of the table was used to cover the whole area between the participant and the computer monitor with black cloth, thus creating a dark booth. The upper part of the monitor was outside the covered area and was visible to the experimenter. The lower part of the monitor was used to project targets

405

406

SMYRNIS ET AL.

Figure 1. A: The experimental design for the antisaccade task and traces of valid responses from 1 participant (Trace 1, correct antisaccade; Traces 2 to 4, error prosaccades with correction). B: Invalid trials that were rejected from analysis (Trace 1, saccades during the last 100 ms of central fixation; Trace 2, blink during the last 100 ms of central fixation; Trace 3, blink during saccade execution; Trace 4, head movements resulting in slow shifts of signal). C: The definition of latency for correct antisaccade (LC) and latency for error prosaccade (LE) in two traces. We excluded trials with artifacts (blinks, etc.) in the analysis period extending from 100 ms before the appearance of the peripheral target to the end of the first saccade as well as trials for which an eye movement occurred in the 100-ms period before the appearance of the peripheral target. In addition, we excluded trials with a response latency that was not within the window of 80⫺600 ms (to avoid including predictive movements). Figure 1 presents traces of valid and invalid individual saccadic eye-movement records. The analyses included only individuals who performed at least 40 valid antisaccade trials (n ⫽ 1,629 participants).

Statistical Analyses Descriptive statistics and Pearson product⫺moment correlations were used for all variables of interest. Bonferroni adjustments for significance

levels of correlations were used. Thus, for the antisaccade index correlations we used p ⬍ .0033, which is the adjusted level of .05 significance for 15 comparisons, and for the antisaccade correlation with schizotypy, anxiety, and depression, we used p ⬍ .0008, which is the adjusted .05 level for 62 comparisons. To analyze the data set as a whole and address the question of the relation between the psychometric variables on the one hand (PAS scores, SPQ scores, and SCL90⫺R anxiety and depression scores) and the antisaccade performance indices on the other hand, we used a multivariate analysis, namely, canonical correlation analysis. This analysis allows the comparison of two sets of correlated variables by comparing a weighted sum of the first set (a linear combination of the variables) with a weighted sum of the second set. Thus, from the original two sets of variables, the analysis derives two weighted sums that are maximally correlated between them. These weighted sums are called canonical variates, and the squared correlation of them is called canonical root. The process of canonical correlation analysis goes on to define more than one canonical root that can describe a complex relation between the two data sets. Each new root explains an additional proportion of variance between the two sets that become smaller and smaller with each new root. Actually, the program computed as many roots as the number of variables in the first set of variables (defined arbitrarily). The significance of each new root is tested, and only the significant roots are retained and interpreted. The meaning of each root is then evaluated by computing the canonical weights for each variable of each one of the two sets. In general, the larger the absolute value of the weight, and the larger the specific contribution of this variable to the root, thus the larger the contribution of this variable to the intercorrelation of the two sets of variables. In our analysis, these canonical weights are presented as z scores with a mean of zero and variance of one. In the group analysis, we used the criterion of two standard deviations (2SD) to derive groups with a very high schizotypy score. We thus derived a group of 68 individuals with a PAS score above 2SD from the population mean, a group of 39 individuals with an SPQ score above 2SD from the population mean, and four groups of individuals with similarly defined high scores in each of the four factors of SPQ. We then compared each group mean in each of the five antisaccade indices with the corresponding population mean that was derived by excluding the high schizotypy individuals using a t test. For each schizotypy group, we thus performed five t tests. We used a p level of .01 to correct for multiple comparisons (.05/5 ⫽ .01). For significant effects, we also performed the analysis after removing the effect of anxiety and depression from the antisaccade measurements. In order to do this, we performed a regression of each antisaccade index in the population (n ⫽ 1,273) using the SCL90⫺R anxiety and depression scores as independent variables. We then used the residuals of the regression in the group analysis for the corresponding antisaccade index. Thus, the residuals of PE(t), LCM, and LCSD were used to compare

Table 1 Summary Statistics for Antisaccade Indices Index

M

SD

2.5% percentile

97.5% percentile

PEa (%) LCMa LCSDa LEMb LESDb EDUa (years)

22.7 268.3 54.5 204.8 44.8 12.6

16.4 38.9 17.7 36.9 26.7 1.86

1.6 202.7 28.2 151.7 4.7 9

63.2 352.9 97.2 288.5 119 16

Note. Values are measured in milliseconds unless otherwise noted. PE ⫽ percentage of errors; LCM ⫽ mean latency for correct antisaccades; LCSD ⫽ standard deviation of latency for correct antisaccades; LEM ⫽ mean latency for error prosaccades; LESD ⫽ standard deviation of latency for error prosaccades; EDU ⫽ years of formal education. a n ⫽ 1,273. b n ⫽ 1,263.

ANTISACCADE PERFORMANCE OF 1,273 MEN the performance of the high PAS group with the population. All statistical analyses were performed using the STATISTICA software (StatSoft Inc., 1999).

Results Antisaccade Indices and Their Interrelationships Table 1 presents the summary statistics for the five antisaccade indices, and Figure 2 shows their distributions. There was a significant decrease in LEM compared with correct antisaccades

407

(paired t test), t(1,262) ⫽ 55.6, p ⬍ .0001. The same decrease was observed for the variance of error latency compared with the variance of latency for correct antisaccades (paired t test), t(1,262) ⫽ 12.75, p ⬍ .0001. The correlation statistics among the antisaccade indices and between the antisaccade indices and the level of education are given in Table 2. All antisaccade indices of performance were not related to the years of formal education (EDU). The error rate was positively correlated with mean latency of hits and its variance and negatively correlated with the mean latency of errors.

Figure 2. Population distributions of the antisaccade indices. PE ⫽ percentage of errors; LCM ⫽ mean latency for correct antisaccades; LCSD ⫽ standard deviation of the latency for correct antisaccades; LEM ⫽ mean latency for error prosaccades; LESD ⫽ standard deviation of the latency for error prosaccades.

SMYRNIS ET AL.

408

Figure 3B shows a scatterplot of the relation of error rate, PE(t), and the SPQ score. Only the variability of response latency for correct antisaccades (LCSD) was marginally significantly correlated with schizotypy measured with PAS (uncorrected p ⫽ .001).

Table 2 Antisaccade Index Correlations Index 1. 2. 3. 4. 5.

PE(t) LCM LCSD LEM LESD

EDU

2

3

⫺.05 ⫺.02 ⫺.04 ⫺.06 ⫺.07

.12* — .66* .42* .29*

.19* — — .32* .31*

4

5

⫺.22* — — — .53*

⫺.02 — — — —

Note. Pearson product–moment correlation coefficients representing the relationships between the antisaccade indices of performance and between these indices and years of formal education (EDU). PE(t) ⫽ arcsinetransformed percentage of errors; LCM ⫽ mean latency for correct antisaccades; LCSD ⫽ standard deviation of latency for correct antisaccades; LEM ⫽ mean latency for error prosaccades; LESD ⫽ standard deviation of latency for error prosaccades. * p ⬍ .05 (corrected for multiple comparisons).

Antisaccade Indices, Schizotypy, Anxiety, and Depression: Population Analysis Both PAS and SPQ had a unimodal distribution in the population (PAS score, M ⫽ 7.46, SD ⫽ 5.12; SPQ score, M ⫽ 27.36, SD ⫽ 12.30). The PAS and the SPQ scores were highly correlated, r ⫽ .65, p ⬍ .0001. The PAS score was also highly correlated with all factors of the SPQ, in particular with the positive symptoms factor as expected, r ⫽ .73, p ⬍ .0001. Table 3 presents the correlation coefficients among PAS, SPQ (overall and the four factors), and the anxiety and depression scores from the SCL90⫺R, with all the antisaccade indices and the years of EDU. The major observations could be summarized as follows: 1.

2.

The variability of response latency for correct antisaccades (LCSD) was significantly correlated with the anxiety score of SCL90⫺R (corrected p ⬍ .05). Anxiety was also correlated with antisaccade error rate, PE(t), but this correlation did not survive correction for multiple comparisons (uncorrected p ⫽ .0034).

3.

Schizotypy was highly positively correlated with both anxiety and depression.

4.

Lower EDU was correlated with higher scores of schizotypy. EDU was not correlated with anxiety or depression.

5.

Using the total number of 1,629 individuals did not result in a change of the correlation coefficients of schizotypy with antisaccade performance indices (small case numbers in PAS and SPQ rows of Table 3).

We next performed a canonical correlation analysis to estimate the relation of antisaccade performance indices (first set of variables) to psychometric scores including schizotypy, anxiety, and depression (second set of variables). In two separate analyses, we used either the PAS and SPQ total scores or the PAS and the negative and disorganized factors of the SPQ that were the most related to antisaccade indices from the four SPQ factors. The results were very similar, so we present the analysis results for PAS and negative plus disorganized factor scores from SPQ. The canonical analysis showed that the correlation between the two sets was highly significant (canonical R ⫽ .18), ␹2(25) ⫽ 57.10,

The antisaccade indices were very weakly or not at all correlated with schizotypy. Figure 3A shows a scatterplot of the relation of error rate, PE(t), and PAS score, and

Table 3 Antisaccade Correlations With Schizotypy, Anxiety, and Depression SCL 90–R Variable

PE(t)

LCM

PAS PAS total SPQ SPQ total F cog-per. F negative F disorganized F paranoid SCL90–R Anxiety Depression

.05 .06 .02 .01 .00 .05 .02 ⫺.04

.04 .05 .02 .01 .01 .05 .02 .00

.08 .06

.02 ⫺.04

LCSD .09 .11 .06 .07 .04 .09 .06 .02 .10* .03

LEM

LESD

EDU

Anxiety

Depression

.02 .05 .04 .04 .02 .05 .04 .01

.06 .06 .03 .03 .04 .03 .03 .00

⫺.10* ⫺.20 ⫺.07 ⫺.12 ⫺.07* ⫺.10* ⫺.08 ⫺.01

.38* .33 .47* .44 .43* .35* .44* .40*

.35* .31 .47* .44 .41* .37* .44* .40*

⫺.02 ⫺.04

.05 .00

.04 .00

— .77*

.77* —

Note. Pearson product–moment correlation coefficients representing the relationships between the antisaccade indices of performance, years of formal education (EDU), and Symptom Checklist 90—Revised scale (SCL90–R) anxiety and depression scores with schizotypy indices. PE(t) ⫽ arcsine-transformed percentage of errors; LCM ⫽ mean latency for correct antisaccades; LCSD ⫽ standard deviation of latency for correct antisaccades; LEM ⫽ mean latency for error prosaccades; LESD ⫽ standard deviation of latency for error prosaccades; PAS ⫽ Perceptual Aberration Scale; PAS total ⫽ correlation coefficients using the total sample of 1,629 individuals; SPQ ⫽ Schizotypal Personality Questionnaire; SPQ total ⫽ correlation coefficients using the total sample of 1,629 individuals; F ⫽ factors from SPQ confirmatory factor analysis; cog-per. ⫽ cognitive–preceptual. * p ⬍ .05 (corrected for multiple comparisons).

ANTISACCADE PERFORMANCE OF 1,273 MEN

409

1.

In the first variable set of antisaccade indices, the variability of latency for correct antisaccades (LCSD) was the highest contributing variable to the total correlation between the two sets, with minor contributions from the PE(t) and the variability of latency for error saccades (LESD).

2.

In the second variable set of psychometric evaluation scores, the anxiety score of SCL90⫺R was the highest contributing variable, with depression being second, although the three measures of schizotypy had the lowest contributions that did not vary considerably among them.

On the one hand, the multivariate analysis thus showed that anxiety and to a lesser degree depression but not schizotypy scores were the major factors contributing to the relation of antisaccade task performance with psychometric evaluation scores. On the other hand, the major factor from the antisaccade task contributing to the relation was the variability of response latency for correct antisaccades and not the PEs in performance.

Antisaccade Indices in High Schizotypy: Group Analysis In this analysis, we compared antisaccade performance indices in the group of individuals with scores above 2SD in each of the schizotypy measures with the population values. The results of the analysis for PAS are presented in Table 5. It can be observed that both the PE and the variability of response latency for corrects Figure 3. A: Scatterplot of arcsine-transformed percentage of errors, PE(t), in y axis versus Perceptual Aberration Scale (PAS) score in x axis. B: Scatterplot of arcsine-transformed PE(t) in y axis versus Schizotypal Personality Questionnaire (SPQ) in x axis.

p ⫽ .0002. The total redundancies though were very small (1.24% for the first set of antisaccades variables and 1.19% for the second set of psychometric scores). The total redundancy for the first set of variables can be explained as the proportion of variance in the second set, which can be explained given the first set. The opposite is true for the second redundancy measure. Thus, the two sets of variables were significantly but very weakly correlated. Next, we proceeded at defining how many canonical roots were significant in the model, thus exploring the complexity of the structure of the correlation matrix between the two sets of variables. The program computed five roots (as many as the variables in the first set) and evaluated the model significance by subtracting the first root, then the first and the second, and so forth. This way, if the first root was the only one that was significant, then its subtraction should yield insignificant model correlation Rs for all subsequent models. This was the case in our analysis (R ⫽ .1), ␹2(16) ⫽ 18.41, p ⫽ .930, after the exclusion of root one (R ⫽ .06), ␹2(9) ⫽ 6.34, p ⫽ .70, and after the exclusion of root one and two, and so forth. Thus, the two data sets were significantly correlated using a single combination of weighted sums for each set. We next investigated what was the unique contribution of each variable within each set to the canonical root by computing canonical weights (see the Method section). Table 4 presents the canonical weights for the set of antisaccade variables and the set of psychometric variables in the model. Two major observations can be made:

Table 4 Canonical Weights for Antisaccade and Psychometric Variable Sets Canonical weightsa

Index First set Antisaccade index PE(t) LCM LCSD LEM LESD

0.37 0.23 0.87 0.12 0.36 Second set

Psychometric index PAS SPQ F negative F disorganized SCL90–R Anxiety Depression

0.49 0.49 0.51 1.13 0.65

Note. This table presents the resulting canonical weights from the canonical correlation analysis. The weights are associated with the first canonical root that was the only significant root in the model (see the Method section). PE(t) ⫽ arcsine-transformed percentage of errors; LCM ⫽ mean latency for correct antisaccades; LCSD ⫽ standard deviation of latency for correct antisaccades; LEM ⫽ mean latency for error prosaccades; LESD ⫽ standard deviation of latency for error prosaccades; PAS ⫽ Perceptual Aberration Scale; SPQ ⫽ Schizotypal Personality Questionnaire; SCL90 – R ⫽ Symptom Checklist 90 —Revised scale. a Absolute value.

SMYRNIS ET AL.

410 Table 5 High Perceptual Abberation Scale (PAS) Group Versus Population Index PE (%) PE(t) LCM LCSD LEM LESD

Ma 22.2 0.466 266.7 54.1 204.5 44.4

M 29.9 0.555 277.7 62.2 208.9 50.2

SD 20.3 0.238 39.4 20.7 35.1 25.6

t (1,270)

t (1,260)

3.08* 2.08 3.25* 1.02 1.86

2.58b 1.99 2.66* — —

Note. Values are measured in milliseconds unless otherwise noted. This table shows population means and the high PAS group means and standard deviations for all antisaccade indices of performance. The t tests used for testing the hypothesis show which high PAS group means differ from the population means. The last column presents the same t tests after partialing out the Symptom Checklist 90 —revised scale anxiety and depression from the antisaccade indices (see Method section). Dashes indicate that data were not reported. PE ⫽ percentage of errors, PE(t) ⫽ arcsine-transformed percentage of errors; LCM ⫽ mean latency for correct antisaccades; LCSD ⫽ standard deviation of latency for correct antisaccades; LEM ⫽ mean latency for error prosaccades; LESD ⫽ standard deviation of latency for error prosaccades. a For the general population. b The p for this test was marginally significant at uncorrected .012. * p ⬍ .05 (corrected for multiple comparisons).

latency for correct antisaccades was again significantly higher in the high PAS group compared with the population at the .01 level of significance (see last two columns of Table 5). Figure 4 shows the relation of PE and variability of latency for correct antisaccades in the high PAS group. The correlation coefficient was .43 ( p ⫽ .0002), which was approximately double the value observed for the population (see Table 1), indicating that in this specific group an increasing error rate is highly correlated with an increase in the variability of antisaccade latencies. Using the high SPQ group (n ⫽ 39 participants), we did not find any significant differences from the population for any of the antisaccade indices of performance. There was an increase in the variability of response latency for correct antisaccades in this group compared with the population (group M ⫽ 62.5 [SD ⫽ 23.6] ms and population M ⫽ 54.3 [SD ⫽ 17.7] ms), although this effect did not reach significance at p ⫽ .01, t(1,269) ⫽ 2.16, p ⫽ .036. Moreover, when covarying for anxiety and depression, the effect disappeared, t(1,269) ⫽ 1.33, p ⫽ .189. Using the high 2SD groups for the four factors of SPQ also did not produce any significant results.

Discussion Antisaccade Performance Indices

antisaccades were significantly higher in this group compared with the population, whereas the other antisaccade indices had no significant effect (the LCM was higher in the high PAS group compared with the population, but the effect did not reach the .01 level of significance). When we performed the analysis on the residuals of the antisaccade indices after the regression using SCL90⫺R anxiety and depression (see the Method section), the error rate became marginally significant, whereas the variability of

The ASPIS study provided us with a population-based oculomotor database that will serve for a longitudinal study. This population will be followed and interviewed at regular intervals in the coming years for the development of psychiatric symptoms. Our sample of young men is by definition not representative of the general population; however, it is generally representative of young Greek men because of the fact that military service is obligatory in Greece, and thus, all men from all areas

Figure 4. Scatterplot of the correlation between arcsine-transformed data percentage of errors, PE(t), and standard deviation of the latency for correct antisaccades (LCSD) for the high Perceptual Aberration Scale group. The solid line represents the regression line, and the dotted lines show 95.0% confidence intervals for the regression line.

ANTISACCADE PERFORMANCE OF 1,273 MEN

must participate barring disabilities. As such, the military provides not only a large pool of participants, but also one that facilitates evaluation of psychosis in a longitudinal study. In this study, we present population-derived measures of antisaccade performance. These measures such as the mean error rate and its variance in the population (see Table 1, Figure 1) could be of importance in the growing literature of antisaccade use as a tool for investigating frontal lobe dysfunction in clinical populations including patients with schizophrenia (Everling & Fischer, 1998). So far, all indications of a dysfunction in this task, as well as in smooth eye pursuit, come from group comparisons between a control group and a patient group. An indication of the potential problems that arise from this approach is that, for example, error rates in antisaccade performance in different studies range from 0% to 30% for controls and 0% to 60% for patients with schizophrenia (Everling & Fischer, 1998). Although the antisaccade task varies from study to study and task-related differences could account in part for these variations, such huge differences in error rates are still difficult to explain. A potential confounding factor in antisaccade performance is EDU. We found that the level of EDU did not have a significant effect on any of the antisaccade variables measured in our study. The other potential difference among control groups is age. In a study of saccades and antisaccades in different age groups, it was shown that the basic parameters of the antisaccade task such as error rate and response latency do vary with age (Fischer, Biscaldi, & Gezeck, 1997). However, the major differences in antisaccade performance were observed between children and adults. Interesting to note, observing the variation of error rate with age in that study, it can be seen that after the age of 20 the error rate stabilized around 20.0%, which is close to the mean error rate of our population. Finally, to our knowledge, the issue of gender differences in antisaccade performance of healthy adults has not been addressed in the relevant literature. Thus, we believe that our data could offer (within the limitations of the specific task parameters) a control standard for future studies of antisaccade performance in clinical populations.

Antisaccade Performance in Schizotypy, Anxiety, and Depression Previous studies have compared smaller samples of individuals with high and low schizotypy and showed that high schizotypy was related to increased error rate in performance of this task, whereas mean response latencies for correct antisaccades and error prosaccades were not different between the two groups (Gooding, 1999; O’Driscoll et al., 1998). Our results showed that the correlations of antisaccade performance measures with schizotypy were negligible in the population. Furthermore, using a multivariate analysis, we confirmed that state-dependent variables anxiety and depression were the most significant contributors in this weak relation to antisaccade performance and not schizotypy. We then defined high schizotypy groups and compared their performance with the population. In this analysis, we confirmed that the high PAS group indeed had a higher error rate and higher latency variability for correct antisaccades than the population. Moreover, the positive relation of these two measures in this group was much stronger than the equivalent relation in the population. In contrast,

411

high scores in SPQ and the four factors of SPQ did not show a difference from the population in any of the antisaccade indices. When partialing out the effects of state-dependent variables such as anxiety and depression, the variability of latency for the high PAS group remained significantly higher than that of the population, whereas the error rate difference became marginally significant. Thus, we could say that our group analysis results basically confirm previous results showing a deviance in performance of the antisaccade task in a group of individuals with high schizotypy identified using the PAS scale (Gooding, 1999; O’Driscoll et al., 1998). We could then try to combine the group analysis and the population analysis results to form a hypothesis on the relation of schizotypy and deficits in eye-movement functioning. It seems that antisaccade performance deviance is not related to increasing schizotypy scores in the population as would be predicted by factorial models (Claridge, 1994). These models would predict that increasing schizotypy scores would result in increasing predisposition to psychosis that in turn would be related to an increasing deviance in antisaccade performance. At the end of this relation lie the patients suffering from schizophrenia. In contrast, our results favor categorical models (Meehl, 1989), predicting that a small subgroup of individuals with very high scores in schizotypy (high PAS group) would express a similar deviance in antisaccade performance that is also present in the disease, although no such relation would be present in the population. It is interesting to note here that we observed the deviance in antisaccade function only in the high PAS group and not in the high SPQ group or the high SPQ factor groups. This is in accordance with previous work showing that the PAS scale assesses a qualitative latent entity that is found at the highest 5%⫺10% of the score distribution, thus identifying schizotypy as a category (Lenzenweger, 1994). In contrast, the SPQ scale has been decomposed in three factors, and it has been used to assess different dimensions of schizotypy in the population, favoring the completely dimensional view of psychopathology (Reynolds, Raine, Mellingen, Venables, & Mednick, 2000). Another finding in our population analysis was the influence on antisaccade performance of state-dependent psychopathological factors, namely, SCL90⫺R anxiety and depression. It was observed that these factors were significantly related to antisaccade performance even more so than schizotypy in the general population. Furthermore, excluding the effect of anxiety and depression in the high schizotypy groups reduced the significance of the difference in antisaccade performance between these groups and the population. In all studies comparing individuals with schizophrenia and controls in antisaccade performance, the level of anxiety or depression was not rigorously studied as potential confounding factors. Considering the fact that increases in error rate were observed for patients with depression (Katsanis et al., 1997; McDowell & Clementz, 1997; Sereno & Holzman, 1995) and for patients with obsessive-compulsive disorder (McDowell & Clementz, 1997; Tien, Pearlson, Machlin, Bylsma, & Hoehn-Saric, 1992) and that patients with schizophrenia also present with symptoms of anxiety and depression during the course of their illness, we believe that our results raise the issue of studying the specific contribution of state-dependent variables anxiety and depression in the antisaccade performance of these patients. The confounding effects of anxiety and depression on performance and the strong

412

SMYRNIS ET AL.

relation of these factors with schizotypy scores in the population could be viewed under the hypothesis of a continuum in the population between a schizophrenia spectrum and the affective disorders (van Os et al., 1999). Still, the existence of a high schizotypy group in which a deviance in performance of the antisaccade task was independent of anxiety and depression is, we believe, in favor of the categorical distinction of schizotypy in the population.

Variability of Response Latency We observed in our analysis that the variability of response latency for correct antisaccades was the most significant factor in the correlation of antisaccade performance with psychometric scores in the population. In addition, it was found that individuals with a very high schizotypy score (PAS) had a significant increase in the variability of their response latency for correct antisaccades. This effect was present even when we excluded the effects of anxiety and depression. Finally, in this high PAS schizotypy group, the relation of PE to the variability of response latency for correct antisaccades was twice as high as that found in the general population. Thus, individuals with a high PAS score who have a large variability in the response latency of their correct antisaccades also are the ones who make more errors in the task. In a study measuring performance on the CPT-IP in a group of individuals with schizophrenia, in a group of people with depression, and in two control groups (one with other psychiatric diagnoses and one with healthy individuals), it was found that both the index measuring errors in performance (d prime) and the response latency variability were significantly higher in patients with schizophrenia and patients with depression relative to controls, although mean response latency did not differ among all groups (van den Bosch, Rombouts, & van Asma, 1996). Furthermore, there was a very high correlation between errors in performance and response latency variability, which was specific only for the patients with schizophrenia and depression, whereas in the control groups this correlation was nonsignificant. In the schizophrenia and depression groups, the errors in performance (indicated by a decreasing d prime score) resulted in an increase in response latency variability. Van den Bosch, Rombouts, and van Asma (1996) pointed to the importance of response latency variability in determining performance in the CPT-IP of patients with schizophrenia. It would be interesting then to study the response latency variability of the antisaccades in patients with schizophrenia. Our prediction would be that the variability would be larger than that of controls, and furthermore, we would expect a positive correlation of this variability with error rate. According to a model for volitional saccade generation (Carpender, 1988; Carpender & Williams, 1995), the variability from trial to trial in saccadic response latency reflects the variability in the slope of the rise time between a resting state and a threshold at which a decision is made for saccade generation, whereas the median latency reflects the time at which this predetermined threshold is reached. In fact, the variability of response latency in a decision eye-movement task in the monkey has been correlated with the variability in the rise time of neuronal activity for neurons in the frontal eye field (FEF), whereas the neuronal activity threshold for saccade generation remained constant (Hanes & Schall, 1996) from trial to trial. Although at this point any

relation of our behavioral data with neurophysiological data is premature, it would be interesting to speculate that the increasing variability of response latency observed in high schizotypy may be related to neural events. Such a link of a behavioral measure with the neuronal activity in FEF could be of significance in schizophrenia considering the hypothesis of a frontal lobe dysfunction in this disease (Weinberger, 1988).

Methodological Issues Two potential caveats related to the present study include the exclusion of 23.0% of the psychometric data of the sample as well as the validity of the PAS and SPQ scores, given their translation into Greek for the first time. Although the EDU background of the participants was diverse, the majority of individuals who did not comply with the validity criterion of the TCI-140⫺R questionnaire (23.0% of the individuals who were screened) had fewer years of EDU than average. Analysis results including these nonvalid responders were generally not different from the results excluding the responders as presented in the Results section. Moreover, by applying strict exclusion criteria, we tried to minimize the potential false positive results in our schizotypy measures that could introduce noise in the measurements. The participants who we excluded had a lower level of EDU, thus the possibility for them not being able to understand the written material to a sufficient level was larger. Regarding the validity of the PAS and SPQ, used here for the first time in the Greek population, the scores obtained in the present sample were quite comparable with those achieved by English-speaking populations who had the questionnaires administered in English. Specifically, although the mean PAS score we report was higher than that reported for young American men (7.42 vs. 6.88, respectively; Thomas Kwapil, personal communication, November 2000), the American sample consisted of undergraduate college students who had completed over 12 years of EDU. When this subgroup was used in our population (n ⫽ 468 individuals), the mean PAS score dropped to 6.88, albeit with a higher standard deviation than reported in the U.S. study (6.06 vs. 4.12, respectively). In addition, the means and standard deviations for the SPQ schizotypy questionnaire scores were not different from the means and standard deviations reported in normative data for this questionnaire (Raine, 1991).

Conclusion This is the first study to investigate the performance of oculomotor tasks in a large sample of healthy individuals, producing the beginnings of a database for indices of performance in these tasks. The antisaccade performance was not related to schizotypy when studying the whole population. There was, however, a specific increase in error rate and variability of response latency for correct antisaccades in a subgroup of individuals with the highest scores in PAS. This finding is in favor of the hypothesis of a qualitative difference between a psychosis-prone group and the general population. Finally, anxiety and depression seemed to affect antisaccade performance in the population, and this could be an important factor to consider in future studies of antisaccade performance in psychiatric disorders, considering the fact that anxiety and depres-

ANTISACCADE PERFORMANCE OF 1,273 MEN

sion symptoms are common during the course of most of these disorders.

References American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Carpender, R. H. S. (1988). Movements of the eyes (2nd ed.). London: Pion. Carpender, R. H. S., & Williams, M. L. L. (1995, September 7). Neural computation of log likelihood in control of saccadic eye movements. Nature, 377, 59⫺62. Chapman, L. J., Chapman, J. P., Kwapil, T. R., Eckbland, M., & Zinser, M. C. (1994). Putatively psychosis-prone subjects 10 years later. Journal of Abnormal Psychology, 103, 171⫺183. Chapman, L. J., Chapman, J. P., & Raulin, M. L. (1978). Body image aberration in schizophrenia. Journal of Abnormal Psychology, 87, 399⫺407. Claridge, G. (1994). Single indicator of risk for schizophrenia: Probable fact or likely myth? Schizophrenia Bulletin, 20, 151⫺168. Clementz, B. A., McDowell, J. E., & Jisook, S. (1994). Saccadic system functioning among schizophrenia patients and their first-degree biological relatives. Journal of Abnormal Psychology, 103, 277⫺287. Cloninger, C. R., Svrakic, D. M., & Przybeck, T. R. (1993). A psychobiological model of temperament and character. Archives of General Psychiatry, 50, 975⫺990. Cornblatt, B. A., Risch, N. J., Faris, G., Friedman, D., & ErlenmeyerKimling, L. (1988). The Continuous Performance Test, Identical Pairs version (CPT–IP): I. New findings about sustained attention in normal families. Psychiatry Research, 26, 223–238. Crawford, T. J., Sharma, T., Puri, B. K., Murray, R. M., Berridge, D. M., & Lewis, S. W. (1998). Saccadic eye movements in families multiply affected with schizophrenia: The Maudsley family study. American Journal of Psychiatry, 155, 1703⫺1709. Curtis, C. E., Calkins, M. E., Grove, W. M., Feil, K. J., & Iacono, W. G. (2001). Saccadic disinhibition in patients with acute and remitted schizophrenia and their first degree relatives. American Journal of Psychiatry, 158, 100⫺106. Derogatis, L. R., Lipman, R. S., Rickels, K., Uhlenhuth, E. H., & Covi, L. (1974). The Hopkins Symptom Checklist (HSCL): A self-report symptom inventory. Behavioral Science, 19, 1⫺15. Evdokimidis, I., Liakopoulos, D., Constantinidis, T. S., & Papageorgiou, C. (1996). Cortical potentials with antisaccades. Electroencephalography and Clinical Neurophysiology, 98, 377⫺384. Everling, S., & Fischer, B. (1998). The antisaccade: A review of basic research and clinical studies. Neuropsychologia, 36, 885⫺899. Everling, S., Krappmann, P., & Flohr, H. (1997). Cortical potentials preceding pro- and antisaccades in man. Electroencephalography and Clinical Neurophysiology, 102, 356⫺362. Eysenck, H. J., & Eysenck, S. B. G. (1976). Psychoticism as a dimension of personality. London: Hodder & Stoughton. Fischer, B., Biscaldi, M., & Gezeck, S. (1997). On the development of voluntary and reflexive components in human saccade generation. Brain Research, 754, 285⫺297. Fukushima, J., Fukushima, K., Chiba, T., Tanaka, S., Yamashita, I., & Kato, M. (1988). Disturbances of voluntary control of saccadic eye movements in schizophrenic patients. Biological Psychiatry, 23, 670⫺677. Fukushima, J., Fukushima, K., Miyasaka, K., & Yamashita, I. (1994). Voluntary control of saccadic eye movement in patients with frontal cortical lesions and parkinsonian patients in comparison with that of schizophrenics. Biological Psychiatry, 36, 21⫺30. Fukushima, J., Morita, N., Fukushima, K., Chiba, T., & Yamashita, I. (1990). Voluntary control of saccadic eye movements in patients with

413

schizophrenic and affective disorders. Journal of Psychiatric Research, 24, 9⫺24. Funahashi, S., Chafee, M. V., & Goldman-Rakic, P. S. (1993, October 21). Prefrontal neuronal activity in Rhesus monkeys performing a delayed antisaccade task. Nature, 365, 753⫺756. Gevins, A. S., Smith, M. E., Le, J., Leong, H., Bennett, J., Martin, N., et al. (1996). High resolution evoked potential imaging of the cortical dynamics of human working memory. Electroencephalography and Clinical Neurophysiology, 98, 327–348. Gooding, D. C. (1999). Antisaccade task performance in questionnaire identified schizotypes. Schizophrenia Research, 35, 157⫺166. Guitton, D., Buchtel, H. A., & Douglas, R. M. (1985). Frontal lobe lesions in man cause difficulties in suppressing reflexive glances and in generating goal-directed saccades. Experimental Brain Research, 58, 455⫺472. Hallet, P. E. (1978). Primary and secondary saccades to goals defined by instructions. Vision Research, 18, 1276⫺1296. Hanes, D. P., & Schall, J. D. (1996, October 18). Neural control of voluntary movement initiation. Science, 274, 427⫺430. Karoumi, B., Saoud, M., d’Amato, T., Rosenfeld, F., Denise, P., Gutknecht, C., et al. (2001). Poor performance in smooth pursuit and antisaccadic eye-movement tasks in healthy siblings of patients with schizophrenia. Psychiatry Research, 101, 209⫺219. Katsanis, J., Kortenkamp, S., Iacono, W. G., & Grove, W. M. (1997). Antisaccade performance in patients with schizophrenia and affective disorder. Journal of Abnormal Psychology, 106, 468⫺472. Lenzenweger, M. F. (1994). Psychometric high-risk paradigm, perceptual aberrations and schizotypy: An update. Schizophrenia Bulletin, 20, 121⫺135. Matthysse, S., Holzman, P. S., & Lange, K. (1986). The genetic transmission of schizophrenia: Application of mendelian latent structure analysis to eye tracking dysfunctions in schizophrenia and affective disorder. Journal of Psychiatric Research, 20, 57⫺65. McDowell, J. E., & Clementz, B. A. (1997). The effect of fixation condition manipulations on antisaccade performance in schizophrenia: Studies of diagnostic specificity. Experimental Brain Research, 115, 333⫺344. McDowell, J. E., Myles-Worsley, M., Coon, H., Byerley, W., & Clementz, B. A. (1999). Measuring liability for schizophrenia using optimized antisaccade stimulus parameters. Psychophysiology, 36, 138⫺141. Meehl, P. E. (1989). Schizotaxia revisited. Archives of General Psychiatry, 46, 935⫺944. Muri, R. M., Hess, C. W., & Meienberg, O. (1991). Transcranial magnetic stimulation of the human frontal eye field by magnetic pulses. Experimental Brain Research, 86, 219⫺223. Nakashima, Y., Momose, T., Sano, I., Katayama, S., Nakajima, T., Niwa, S., & Matsushita, M. (1994). Cortical control of saccade in normal and schizophrenic subjects: A PET study using a task-evoked rCBF paradigm. Schizophrenia Research, 12, 259⫺264. Ntonias, S., Karastergiou, A., & Manos, N. (1990). Standardization of the Symptom Checklist 90⫺R rating scale in a Greek population. Psychiatriki, 2, 42⫺48. O’Driscoll, G. A., Alpert, N. M., Matthysse, S. W., Levy, D. L., Rauch, S. L., & Holzman, P. S. (1995). Functional neuroanatomy of antisaccade eye movements investigated with positron emission tomography. Proceedings of the National Academy of Sciences, USA, 92, 925⫺929. O’Driscoll, G. A., Lenzenweger, M. F., & Holzman, P. S. (1998). Antisaccades and smooth pursuit eye tracking and schizotypy. Archives of General Psychiatry, 55, 837⫺843. Pierrot-Deseilligny, C. P., Rivaud, S., Gaymard, B., & Agyd, Y. (1991). Cortical control of reflexive visually-guided saccades. Brain, 114, 1473⫺1485. Raine, A. (1991). The SPQ: A scale for the assessment of schizotypal personality in a non-clinical sample: The role of task demand. Schizophrenia Bulletin, 17, 555⫺564.

414

SMYRNIS ET AL.

Reynolds, C. A., Raine, A., Mellingen, K., Venables, P. H., & Mednick, S. A. (2000). Three-factor model of schizotypal personality: Invariance across culture, gender, religious affiliation, family adversity, and psychopathology. Schizophrenia Bulletin, 26, 603⫺618. Rosse, R. B., Schwartz, B. L., Kim, S. Y., & Deutsch, S. I. (1993). Correlation between antisaccade and Wisconsin Card Sorting Test performance in schizophrenia. American Journal of Psychiatry, 150, 333⫺335. Schlag-Rey, M., Amador, N., Sanchez, H., & Schlag, J. (1997, November 27). Antisaccade performance predicted by neuronal activity in the supplementary eye field. Nature, 390, 398⫺401. Sereno, A. B., & Holzman, P. S. (1995). Antisaccades and smooth pursuit eye movements in schizophrenia. Biological Psychiatry, 37, 394⫺401. Snedecor, G. W., & Cochran, W. G. (1980). Statistical methods (7th ed.). Ames: Iowa State University Press. StatSoft Inc. (1999). STATISTICA for Windows [Computer manual]. Tulsa, OK: Author. Stefanis, N. C., Smyrnis, N., Avramopoulos, D., Evdokimidis, I., Tzoufras, I., & Stefanis, C. N. (2003). Factorial composition of self rated schizotypal traits amongst young males undergoing military training: The ASPIS. Manuscript submitted for publication. Sweeney, J. A., Mintun, M. A., & Kwee, S. (1996). Positron emission tomography study of voluntary saccadic eye movements and spatial working memory. Journal of Neurophysiology, 75, 454⫺468.

Thaker, G. K., Cassady, S., Adami, H., Moran, M., & Ross, D. E. (1996). Eye movements in spectrum personality disorders: Comparison of community subjects and relatives of schizophrenic patients. American Journal of Psychiatry, 153, 362⫺368. Thaker, G. K., Ross, D. E., Cassady, S. L., Adami, H., Medoff, D. R., & Sherr, J. (2000). Saccadic eye movement abnormalities in relatives of patients with schizophrenia. Schizophrenia Research, 45, 235⫺244. Tien, A. Y., Pearlson, G. D., Machlin, S. R., Bylsma, F. W., & HoehnSaric, R. (1992). Oculomotor performance in obsessive-compulsive disorder. American Journal of Psychiatry, 149, 641⫺646. van den Bosch, R. J., Rombouts, R. P., & van Asma, M. J. O. (1996). What determines continuous performance task performance? Schizophrenia Bulletin, 22, 643⫺651. van Os, J., Verdoux, H., Bijl, R., & Ravelli, A. (1999). Psychosis as an extreme of continuous variation in dimensions of psychopathology. In W. F. Gattaz & H. Hafner (Eds.), Search for the causes of schizophrenia (pp. 59⫺79). Berlin, Germany: Springer-Verlag. Weinberger, D. R. (1988). Schizophrenia and the frontal lobe. Trends in Neuroscience, 11, 367⫺370.

Received August 27, 2001 Revision received December 12, 2002 Accepted December 20, 2002 䡲