Smooth pursuit eye movements in 1,087 men: effects of schizotypy, anxiety, and depression

Preview:

Citation preview

Exp Brain Res (2007) 179:397–408

DOI 10.1007/s00221-006-0797-8

RESEARCH ARTICLE

Smooth pursuit eye movements in 1,087 men: eVects of schizotypy, anxiety, and depression

Nikolaos Smyrnis · Ioannis Evdokimidis · Asimakis Mantas · Emmanouil Kattoulas · Nicholas C. Stefanis · Theodoros S. Constantinidis · Dimitrios Avramopoulos · Costas N. Stefanis

Received: 9 December 2005 / Accepted: 6 November 2006 / Published online: 30 November 2006© Springer-Verlag 2006

Abstract Individuals with schizotypal personality dis-order or high scores in questionnaires measuringschizotypy are at high risk for the development ofschizophrenia and they also share some of the samephenotypic characteristics such as eye-tracking dys-function (ETD). The question arises whether theseindividuals form a distinct high-risk group in the gen-eral population or whether schizotypy and ETD co-vary in the general population with no distinct cutoVpoint for a high-risk group. A large sample of militaryconscripts aged 18–25 were screened using oculomotor,cognitive and psychometric tools for the purposes of aprospective study on predisposing factors for the devel-opment of psychosis. Schizotypy measured using theperceptual aberration scale (PAS) and the schizotypalpersonality questionnaire (SPQ), anxiety and depres-sion, measured using the Symptom Checklist 90-R, hadno eVect on pursuit performance in the total sample.Small groups of individuals with very high scores inschizotypy questionnaires were then identiWed. These

groups were not mutually exclusive. The high PASgroup had higher root-mean-square error scores (aquantitative measure for pursuit quality) than thetotal sample, and the high disorganized factor of SPQgroup had lower gain and higher saccade frequenciesin pursuit than the total sample. The presence of sig-niWcant diVerences in pursuit performance only forpredeWned high schizotypy groups favors the hypothe-sis that individuals with high schizotypy might presentone or more high-risk groups, distinct from the generalpopulation, that are prone to ETD as that observed inschizophrenia.

Keywords Eye movements · Smooth pursuit deWcit · Population · Personality traits · Psychosis proneness

Introduction

Impairment in the performance of smooth eye pursuitmovements [eye-tracking dysfunction (ETD)] has beenconsistently reported for schizophrenic patients andtheir Wrst-degree relatives (Holzman et al. 1973, 1974;Holzman and Levy 1977; Levy et al. 1993). Holzmanand Matthysse (1990) proposed that smooth eye pur-suit could serve as a biological marker for detectinggenetic susceptibility to schizophrenic spectrum disor-ders. In this respect it becomes an important issuewhether ETD is present before the onset of schizo-phrenia, for example, in high-risk populations. Suchhigh-risk populations are, beyond the Wrst-degree rela-tives of patients, those individuals with clinically diag-nosed schizotypal personality disorder or individualswithin the general population that score high on self-reporting questionnaires that measure schizotypy. It

N. Smyrnis · N. C. Stefanis · T. S. Constantinidis · D. Avramopoulos · C. N. StefanisUniversity Mental Health Research Institute, Athens, Greece

N. Smyrnis · I. Evdokimidis · A. Mantas · E. KattoulasCognition and Action Group, Neurology Department, National and Kapodistrian University of Athens, Eginitio Hospital, Athens, Greece

N. Smyrnis (&) · N. C. StefanisPsychiatry Department, National and Kapodistrian University of Athens, Eginitio Hospital, 74 Vas. SoWas Ave., 11528 Athens, Greecee-mail: smyrnis@med.uoa.gr

123

398 Exp Brain Res (2007) 179:397–408

has been shown that schizotypal individuals are at agreater risk of developing psychosis than the generalpopulation (Chapman et al. 1994).

Two major theoretical frameworks have been pro-posed to explain the link between schizotypy andschizophrenia. The theory of “schizotaxia” of Meehl(Meehl 1989) proposes that schizotaxia is a conjecturedneural integrative defect due to a dominant “schizo-gene” that gives rise to the schizotypal personality.This genetic proWle in synergy with other polygenicpotentiators and adverse life experiences gives rise tothe clinical syndrome of schizophrenia in a small per-centage of these individuals. In a theory closely relatedto Meehl’s theory of schizotaxia, Matthysse et al.(1986) proposed that schizotypal traits are one of thephenotypic expressions of an underlying vulnerabilityto schizophrenia, while other such expressions are thedeWcits in tests of neurocognitive function and eyemovement tasks, like the smooth eye pursuit task. Thistheoretical framework predicts that, in a distinctivegroup of individuals with high schizotypy, that have notdeveloped schizophrenia, there would be indications ofeye movement function deWcits similar to thoseobserved in schizophrenia.

The second theoretical framework of schizotypyfavored by Eysenck (Eysenck and Eysenck 1976) pro-poses that personality traits, such as those that deWnepsychosis, are a continuum from health to schizophre-nia with no need to introduce arbitrary cutoV points,above which schizotypy lies as a diVerent entity (Cla-ridge 1994). According to this view certain dimensionsof personality are to be found in the general populationand their extremes lead to the symptoms of a diseasestate such as schizophrenia. Within this framework,schizotypy is decomposed in dimensions using a factormodel and the diVerent factors identiWed in the sam-ples of the general population (for example, a positivesymptom factor loading on unusual perceptual experi-ences) reXect the corresponding factor structure of theclinically identiWed syndrome of schizophrenia (in theprevious example, the positive symptoms of schizo-phrenia). This model then would predict that otherphenotypic expressions of schizophrenia, such as ETD,would also co-vary with psychosis in the population,showing a gradual increase in the levels found inpatients. Thus one would predict, based on the co-vari-ation of phenotypic expressions of schizophrenia in thenormal population, that a correlation of ETD andschizotypy would be evident in the population.

All previous studies on schizotypy and smooth eyepursuit used predeWned groups of schizotypes and nor-mal controls to study the eVects of schizotypy onsmooth eye pursuit performance. It was reported that

students with qualitatively assessed ETD had a higherprobability of schizotypal personality disorder diag-nosed using DSMIII criteria (Siever et al. 1984). Indi-viduals with DSMIII schizotypal personality disorderhad a higher probability of qualitatively assessed ETDcompared to normal controls and compared to individ-uals with personality disorders not in the schizophreniaspectrum (Siever et al. 1990, 1994). Using the Chap-man self-reporting questionnaires such as the percep-tual aberration scale (Chapman et al. 1978), it wasshown that a group of individuals with very high scoresin schizotypy had a higher probability of ETD mea-sured as low quality pursuit compared to normal con-trols (O’Driscoll et al. 1998; Gooding et al. 2000). Inaddition, a group of individuals with very high scores inthe schizotypal personality questionnaire (Raine 1991)had lower quality of smooth eye pursuit compared tocontrols (Lencz et al. 1993). In contrast to these Wnd-ings, one study showed that individuals with very highschizotypy scores had no diVerence in the mean qualityof their pursuit compared to controls, but there wasmore variability in the quality of the pursuit records inthis group (Simons and Katkin 1985). Finally, it wasreported that only a subgroup of individuals with clini-cal characteristics of schizotypy, that were also Wrst-degree relatives of patients with schizophrenia, had asigniWcantly higher probability of presenting with qual-itatively measured ETD compared to normal controls(Thaker et al. 1996).

In the majority of studies on schizotypy andsmooth eye pursuit, the ETD was detected only withqualitative assessment of the smooth eye pursuitrecords. Siever et al. (1994) used quantitative mea-sures of pursuit performance, namely pursuit gain andsaccade frequency, and observed no diVerencebetween schizotypes and controls. Gooding et al.(2000) measured pursuit performance using the root-mean-square error that is the diVerence of the eyemovement signal to the target movement signal. Thisquantitative measure assesses global pursuit quality.In the study of Gooding et al. (2000) root-mean-square error was higher and pursuit gain was lower inthe high schizotypy groups compared to controls butthere was no diVerence in saccade frequency in thetwo groups.

The Athens Study of Psychosis Proneness and Inci-dence of Schizophrenia (ASPIS, Smyrnis et al. 2003;Stefanis et al. 2004) involved the collection of datafrom eye-movement and cognitive tests in a sample of2,130 young men in the Greek Air Force. The study isprospective in nature and longitudinal, designed to fol-low these individuals and investigate if oculomotorfunctions, such as smooth eye pursuit, and cognitive

123

Exp Brain Res (2007) 179:397–408 399

functions have predictive value for the later develop-ment of a clinical psychiatric disorder. In this report,we provide data on quantitative measures of smootheye pursuit performance in this sample. Althoughthese data are limited by the speciWc task conditions,age and gender, they provide a unique database of ocu-lomotor function in a large sample of apparentlyhealthy individuals.

This unique database oVered us the opportunity toaddress the question of whether within a large sampleof apparently healthy individuals there is one or moresmall groups that exhibit a combination of neurophysi-ological (ETD) and psychological (high schizotypy)phenotypic expressions of high risk for the develop-ment of schizophrenia. The theory of schizotaxia pre-dicts the existence of such groups and also predicts thatthese groups share genetic similarities with patientssuVering from schizophrenia. The negative predictionof the same theory is that all other individuals of thepopulation do not share these phenotypic expressionsof genetic susceptibility with the schizotaxia group. Analternative prediction would be that diVerent pheno-typic expressions of high risk for schizophrenia arepresent in an attenuated form in the population. Thusgenetic susceptibility co-varies with the magnitude ofthe phenotypic expressions and there are no distincthigh-risk groups. All previous studies have used a pri-ori deWned groups to study ETD in schizophrenia spec-trum disorders. Having data for eye movement tasksand psychometric scores in a large sample, we wereable to study ETD in the total sample and observewhether phenotypic expressions such as ETD andschizotypy co-vary in the sample or whether ETD ispresent only in a small subgroup in the total samplewith high schizotypy. In the latter case we would beable to derive speciWc high-risk groups within our sam-ple that would combine diVerent phenotypic character-istics common to schizophrenia and predict a highergenetic susceptibility for these individuals to developpsychosis.

A potential confounding factor in the study of ETDas a biological marker in schizophrenia spectrum disor-ders is whether ETD is related to anxiety and depres-sion and thus does not reXect a phenotypic expressionof vulnerability that should be stable in time. Thepotential confounding eVect of depression and anxietybecomes even more important in this sample of youngarmy recruits who were in the Wrst phase of theirmilitary training. Military training is a stressfulcondition that could result in the development of anxi-ety or depression symptoms. We thus tested the rela-tion of ETD to anxiety and depression in addition toschizotypy.

Methods

Participants and materials

A sample of 2,130 young men aged 18–25 were recruitedfrom the Greek Air Force. All participants gave writteninformed consent. A total of 2,075 individuals performedthe following eye-movement tasks: smooth eye-pursuit,saccade, antisaccade, visual Wxation. We also adminis-tered to a randomly selected subset of 1,657 of these sub-jects (80% of participants) a battery of self-ratedquestionnaires assessing among other variables (a)schizotypy with the perceptual aberration scale (PAS,Chapman et al. 1978) and the schizotypal personalityquestionnaire (SPQ, Raine 1991); (b) current psychopa-thology with the Symptom Checklist 90 revised scale(SCL90-R, Derogatis et al. 1974) translated and stan-dardized in the Greek population (Ntonias et al. 1990);(c) personality characteristics with the Temperament andCharacter Inventory (TCI-140-R, Cloninger et al. 1993).

A 2-year follow-up phase was initiated for all partici-pants to record those that were admitted in the Neurol-ogy or Psychiatry services of the Air Force GeneralHospital during service time. During this phase of thestudy 43 subjects with a neurological or psychiatric disor-der were identiWed. Most of these individuals had a diag-nosis before entering the service but they did not reportit during the initial medical assessment of conscriptsentering military service. This assessment is a standardinterview-based procedure performed by a team of med-ical doctors of all specialties (military personnel).Among the 43, seven subjects were diagnosed with psy-chosis. These 43 individuals were not excluded from thetotal sample data set in the presentation that followedbut their exclusion did not change any of the Wndings.

The four validity items of the TCI-R questionnairewere used to exclude 296 subjects (20.8% of the 1,657)that responded incorrectly in at least one of theseitems, thus ensuring the collaboration of the respond-ers in these self-report scales. These items were of thetype “please answer this question by ticking box 4”anda wrong answer would indicate that the individual wasresponding at random. Another 274 subjects (20.1% ofthe remaining 1,361) did not comply with the criteriafor successful performance of the pursuit task (seebelow for deWnition of these criteria). The analyseswere performed on the remaining sample of 1,087 sub-jects (valid responders).

Apparatus

A detailed description of the apparatus for eyemovement measurements is given in a prior report

123

400 Exp Brain Res (2007) 179:397–408

(Smyrnis et al. 2003). Eye movements were recordedfrom the right eye only using the IRIS SCALAR®infrared device. A 12-bit A/D converter was used fordata acquisition (Advantech® PC-Lab Card 818L).Eye movement data were sampled at 600 Hz andwere stored in the PC hard disk for oV-line dataprocessing.

Procedure

Each subject initially performed a calibration proce-dure. A target (white cross 0.5 £ 0.5°) appeared at thecenter of the visible screen and then at constant inter-vals of 2 s moved successively 10° to the right back tothe center, 10° to the left and back of the center. Thiscycle was repeated twice and then two more cycleswere performed at distances of 5°. The subjects wereinstructed to look at the visual target. We thusrecorded four saccades for each of four positions (left/right, 5/10°). If necessary left/right diVerences in ampli-tude were corrected with a manual adjustment of theIRIS device gain control and the calibration procedurewas repeated.

Then the subject was instructed to follow a target(white cross 0.5 £ 0.5°) that was moving horizontallyon the computer monitor in constant speed. The visualangle of the moving target was §10° from the center ofthe screen. We used Wve target speeds (10, 20, 30, 40and 50 deg/s). The subject completed Wve cycles foreach target speed consisting of the target moving 20° tothe left and then 20° to the right at constant speed. Thetarget started moving at the speed of 10 deg/s. Aftercompletion of the Wve cycles the target increased speedand the subject continued tracking at 20 deg/s then at30 deg/s, etc., until all Wve target speeds were pre-sented. There was no stop between changes in targetspeed. In this analysis we report data on the Wrst threeof the target speeds (10–30 deg/s). The reason forexcluding the speed of 40 and 50 deg/s was that at thesetarget speeds a large proportion of subjects wouldchange to a diVerent strategy and would not pursue thetarget but would make large predictive saccades fromone corner of the screen to the other.

Quantitative assessment of smooth eye pursuit

A PC program was used to calculate the root-mean-square error between the eye position and the targetposition record at each target speed (10, 20, and30 deg/s). The root-mean-square error is a global mea-sure of pursuit accuracy and increases with increasingdissimilarity between the eye position and the targetposition. We excluded root-mean-square error

measurements where the eye position signal had satu-rated for more than 10% of the total signal duration.We also excluded the top 2.5% of root-mean-squareerror scores at each target speed that were consideredas extremely low quality pursuit signals.

An interactive PC program (created using the Test-Point® CEC) was used for detection and measure-ment of pursuit gain and number of saccades from theeye movement record. First the calibration amplitudedata were entered into a polynomial model (using the“Wtpolynomial” function of Testpoint) and a linear Wtwas used to adjust for small left/right hemiWeld diVer-ences in amplitude measurement that could not bedetected by the experimenter. This linearization pro-cess was important in order to make sure that left andright pursuit displacement was measured with thesame accuracy. The program then selected a period of133 ms centered at the point where the target crossedthe center of the screen (two periods per run, tenperiods for each target speed). This window wasselected in order to measure eye velocity at the pri-mary position (Leigh and Zee 1991). The operatormanually discarded the period if an artifact wasdetected (i.e., a blink). If the period was artifact free,then the program computed the instantaneous veloc-ity by simple numeric diVerentiation. No Wltering wasapplied to instantaneous velocity. After computingthe instantaneous velocity the program derived amedian value of it for all the period scanned. Then theprogram identiWed particular points where the instan-taneous velocity exceeded the median value by morethan twofold, an abrupt increase in instantaneousvelocity that is synonymous to an abrupt accelerationof the eye. This large acceleration cannot beexplained by smooth pursuit at a constant speed; thus,by inference a saccade has began. Although this crite-rion is not the one typically used in the saccade detec-tion (a saccade velocity over a certain threshold), wefound that it was very good at detecting saccadeswithout over-including saccades in the high pursuitspeed records or missing saccades in the low speedrecords. The program considered these instances asoccurrences of saccadic eye movements and mea-sured their latency from the period onset and theirduration. The saccades were marked on the pursuitrecord and the operator could accept or reject themby inspection. After excluding all time segments withsaccades, the remaining pure pursuit segments weremarked on the position trace and were used to mea-sure mean velocity (amplitude diVerence divided bythe total segment time) for the segment and Wnally bydividing the mean velocity with the correspondingtarget velocity to derive a gain value for the particular

123

Exp Brain Res (2007) 179:397–408 401

segment. For each individual, the following smootheye pursuit performance indices were evaluated: (1)median pursuit gain at each target speed (10, 20,30 deg/s); (2) saccade frequency at each target speedwhich was the total number of saccades for all periodswithout artifacts divided by the sum of these timeperiods. The indices of pursuit gain and saccade num-ber for a particular speed for a particular subject wereconsidered valid if they were derived from at leastthree periods of measurements free of artifacts. Theanalysis was performed only for subjects that hadvalid measurements in all nine indices of pursuit per-formance (see section on Participants and materials).

Statistical analysis

Repeated measures analysis of variance (ANOVA)was used to test for diVerences between target speedsfor root-mean-square error, gain and saccade fre-quency in the total sample and to test for the eVects oflevel of education (Wve levels) on these pursuit indices.

We then used two approaches to analyze the rela-tion of psychometric scores to pursuit performance.The Wrst approach is to test the eVects of psychometricscore variables on pursuit performance globally withinthe total sample without grouping into subgroups withspeciWc characteristics. For this purpose we used thegeneral linear model to test for the eVects of the psy-chometric variables (schizotypy, anxiety and depres-sion) on each one of the three pursuit performancemeasures (root-mean-square error, pursuit gain, sac-cade frequency) in the total sample using the three tar-get speeds as the repeated measures within subjectfactor. The general linear model is a generalization ofthe linear regression model, such that eVects can betested for categorical predictor variables (an ANOVAdesign), as well as for continuous predictor variables (amultiple regression design) and in designs with multi-ple dependent variables as well as with a single depen-dent variable. We also considered a general linearmodel where four SPQ factor scores were entered ascontinuous predictors instead of the total SPQ score.

The second approach was based on the deWnition ofspeciWc subgroups within the total sample. In thisgroup analysis we used the criterion of 2 standard devi-ations (2 SD) to derive groups with a very high schizo-typy score. Note that these groups were not mutuallyexclusive. We thus derived a high PAS score group(N = 55 subjects), a high SPQ score group (N = 26 sub-jects) and four groups of individuals with similarly deW-ned high scores in each of the four factors of SPQ(cognitive perceptual group N = 34 subjects; negativegroup N = 41 subjects; disorganization group N = 23

subjects; paranoid group N = 19 subjects). The highPAS and high SPQ groups were partially overlapping,that is 13 individuals belonged to both groups. Twomore groups of high anxiety (N = 54 subjects) and highdepression (N = 45 subjects) were derived fromSCL90-R scores on anxiety and depression. Againthese groups were partially overlapping, that is 25 indi-viduals belonged to both groups. Repeated measuresANOVA with within subjects factor the target speedand between subjects factor the group (total sampleversus high score group) was used to assess the eVect ofgroup on each one of the pursuit indices of perfor-mance (root-mean-square error, gain, saccade fre-quency). All statistical analyses were performed usingthe STATISTICA 6.0 software (STATSOFT Inc.1984–2001).

Results

General characteristics and psychometric proWle of participants

Table 1 shows the descriptive statistics for age and edu-cation of the participants. It also presents the descrip-tive statistics for the PAS score, SPQ total score andSCL90-R anxiety and depression scores in the totalsample. In a previous paper we described in detail theconWrmatory factor analysis that was used to decom-pose SPQ (Stefanis et al. 2004). On the basis of thisanalysis a four factor model was the best in describingthis data set. The Wrst factor loaded on the subscales ofodd beliefs and odd perceptual experiences (cognitive-perceptual). The second factor loaded on suspicious-ness, social anxiety, no close friends and constrictedaVect (negative). The third factor loaded on oddbehavior and odd speech (disorganization) and thefourth factor loaded on social anxiety, suspiciousnessand ideas of reference (paranoid).

Table 1 Mean and standard deviation (SD) for the total sampledata

PAS perceptual aberration scale, SPQ schizotypal personalityquestionnaire, Anxiety: symptom checklist 90-R anxiety scale,Depression: symptom checklist 90-R depression scale

Mean SD

Age 20.9 1.9Education 12.6 1.87PAS 7.54 5.27SPQ 27.34 12.41Anxiety 8.22 6.59Depression 13.86 8.48

123

402 Exp Brain Res (2007) 179:397–408

Pursuit performance

The root-mean-square error increased with increasingtarget speed (F2,2172 = 451, P < 10¡5) (Fig. 1), the gaindecreased (F2,2172 = 595, P < 10¡5) (Fig. 2) and the sac-cade frequency increased (F2,2172 = 337, P < 10¡5)(Fig. 3). The level of education had no signiWcant eVecton root-mean-square error (F4,1082 = 0.5, P > 0.7) orinteraction with target speed (F8,2164 = 1, P > 0.4). Thelevel of education also had no signiWcant eVect on pur-suit gain (F4,1082 = 0.25, P > 0.9) or interaction with tar-get speed (F8,2164 = 126, P > 0.2). Finally, there was nosigniWcant eVect of level of education on saccade fre-quency (F4,1082 = 0.6, P > 0.6) neither a signiWcant inter-action with target speed (F8,2164 = 1.12, P > 0.3).

EVect of psychometric scores on pursuit performance in the total sample

Root-mean-square error

As described in the Methods section, this analysisfocused on the eVects of psychometric scores on root-

mean-square error in the total sample. In the Wrst anal-ysis we used as independent variables (predictors) theSPQ total score, the PAS score and the SCL90-R anxi-ety and depression scores and as the repeated mea-sures dependent variables, the root-mean-squareerrors at the three diVerent speeds. Part A of Table 2presents the results of the general linear model analy-sis. It can be seen that none of the psychometric vari-ables had a signiWcant main eVect on root-mean-squareerror and there was no signiWcant interaction of theeVect of these variables and target speed. In the secondanalysis we used as independent variables the four fac-tors of SPQ. Part A of Table 3 presents the results ofthis analysis. It can be seen that none of the SPQ fac-tors had signiWcant main eVect on root-mean-squareerror and there was no signiWcant interaction betweenthese variables and target speed.

Pursuit gain

The same general linear model analysis was used tostudy the eVect of psychometric scores on pursuit gain inthe total sample. In the Wrst analysis we used as indepen-dent variables (predictors) the SPQ total score, the PASscore and the SCL90-R anxiety and depression scoresand as the repeated measures dependent variables, thepursuit gains at the three diVerent speeds. Part B ofTable 2 presents the results of the general linear modelanalysis. It can be seen that none of the psychometricvariables had a signiWcant main eVect on pursuit gainand there was no signiWcant interaction of the eVect ofthese variables and target speed. Part B of Table 3 pre-sents the results of the second analysis that used the fourfactors of SPQ as independent variables. It can be seenthat none of the SPQ factors had a signiWcant main eVecton pursuit gain and there was no signiWcant interactionbetween these variables and target speed.

Fig. 1 Histograms showing the root-mean-square error (RMSE)at 10, 20 and 30 deg/s for the high perceptual aberration scale(PAS) schizotypy group (PAS > 2 SD) compared to the totalsample for the three diVerent target speeds. Bars represent stan-dard errors of the mean

0

40

80

120

160

200

240

10 deg/sec 20 deg/sec 30deg/sec

RM

SE

Total Sample

High PAS

Fig. 2 Histograms showing the pursuit gain (GAIN) at 10, 20 and30 deg/s for the group with high scores (>SD) in the disorganiza-tion factor of the schizotypal personality questionnaire (SPQ)compared to the total sample for the three diVerent target speeds.Bars represent standard errors of the mean

0.5

0.6

0.7

0.8

0.9

1.0

10 deg/sec 20 deg/sec 30 deg/sec

GA

IN

Total Sample

Disorganization

Fig. 3 Histograms showing the saccade frequency (SF) in sac-cades per second at 10, 20 and 30 deg/s for the group with highscores (>SD) in the disorganization factor of the schizotypal per-sonality questionnaire (SPQ) compared to the total sample forthe three diVerent target speeds. Bars represent standard errorsof the mean

0

1

2

3

4

5

10 deg/sec 20 deg/sec 30 deg/sec

SF (s

ac /

sec)

Total Sample

Disorganization

123

Exp Brain Res (2007) 179:397–408 403

Saccade frequency

Finally the general linear model analysis was used tostudy the eVect of psychometric scores on saccade fre-quency in the total sample. In the Wrst analysis we usedas independent variables (predictors) the SPQ totalscore, the PAS score and the SCL90-R anxiety anddepression scores and as the repeated measures depen-dent variables, the saccade frequencies at the threediVerent speeds. Part C of Table 2 presents the resultsof the general linear model analysis. It can be seen thatnone of the psychometric variables had a signiWcantmain eVect on saccade frequency and there was no sig-niWcant interaction of the eVect of these variables andtarget speed. Part C of Table 3 presents the results ofthe second analysis that used the four factors of SPQ asindependent variables. It can be seen that none of the

SPQ factors had a signiWcant main eVect on saccadefrequency and there was no signiWcant interactionbetween these variables and target speed except for thenegative factor. The univariate analysis showed thatthere was no signiWcant eVect of the negative SPQ fac-tor on saccade frequency at 10 deg/s (F1,1080 = 0.6,P > 0.4) nor at 20 deg/s (F1,1080 = 1.33, P > 0.2) but onlyat 30 deg/s (F1,1080 = 7.13, P < 0.05). We then used lin-ear regression with negative SPQ factor scores as theindependent variable and saccade frequency at 30 deg/sas the dependent variable to investigate this signiW-cant eVect of the negative SPQ factor scores on saccadefrequency only at the target speed of 30 deg/s. Theregression conWrmed a positive relation, thus suggest-ing an increase in saccade frequency with increasingnegative schizotypy in the total sample. The regressionr2 though was 0.005, which means that the variance in

Table 2 The table presents the results of general linear model analysis on the eVects of psychometric variables

PAS perceptual aberration scale, SPQ schizotypal personality questionnaire, Anxiety: SCL90-R anxiety scale, Depression: SCL90-Rdepression scale

Predictors Main eVect F Main eVect P Interaction F Interaction P

Model A: repeated measures factor: root-mean-square error (three target speeds) PAS 1.45 0.22 1.01 0.36SPQ 3.68 0.06 0.10 0.90Anxiety 0.07 0.78 1.68 0.18Depression 1.02 0.31 1.70 0.18

Model B: repeated measures factor: pursuit gain (three target speeds)PAS 0.67 0.41 0.60 0.54SPQ 0.90 0.34 0.03 0.96Anxiety 1.06 0.30 0.06 0.93Depression 0.01 0.91 0.65 0.52

Model C: repeated measures factor: saccade frequency (three target speeds)PAS 0.08 0.76 1.13 0.32SPQ 1.33 0.24 0.12 0.88Anxiety 0.50 0.47 0.06 0.94Depression 0.22 0.63 0.47 0.62

Table 3 The table presents the results of general linear model analysis on the eVects of the SPQ factors

SigniWcant results at the level of <0.05 are marked with bold characters

Predictors Main eVect F Main eVect P Interaction F Interaction P

Model A: repeated measures factor: root-mean-square error (three target speeds)Cognitive-perceptual 0.88 0.34 0.65 0.52Negative 1.68 0.19 0.83 0.43Disorganization 0.02 0.87 2.56 0.08Paranoid 0.39 0.53 0.86 0.42

Model B: repeated measures factor: pursuit gain (three target speeds)Cognitive-perceptual 1.60 0.20 0.81 0.44Negative 0.07 0.79 0.96 0.38Disorganization 1.04 0.30 0.39 0.67Paranoid 0.005 0.94 0.81 0.44

Model C: repeated measures factor: saccade frequency (three target speeds)Cognitive-perceptual 0.01 0.89 0.59 0.54Negative 3.05 0.08 4.48 0.01Disorganization 0.08 0.77 0.96 0.38Paranoid 0.17 0.67 0.12 0.88

123

404 Exp Brain Res (2007) 179:397–408

saccade frequency explained by the variation in thenegative factor of schizotypy was 0.5%. In other wordsthis eVect of increasing negative schizotypy on saccadefrequency, although statistically signiWcant, wasextremely small.

DiVerence in pursuit performance between predeWned high psychometric score groups and the total sample

Root-mean-square error

In this analysis we used a set of repeated measuresANOVAs to compare the root-mean-square error(repeated measures dependent variable) of groupswith scores >2 SD in PAS, SPQ total score, SPQ fac-tor scores, SCL90-R anxiety and depression with theroot-mean-square error for the total sample (seeMethods section). None of these ANOVA analysesshowed a signiWcant diVerence in root-mean-squareerror between the high psychometric score groupsand the total sample with one exception. There was asigniWcant increase of root-mean-square error in thehigh PAS group compared to the total sample(F1,1085 = 3.87, P < 0.05). Moreover, the nonsigniWcantinteraction eVect of group versus target speed (F2, 2170= 1.02, P > 0.3) conWrmed that this increase in root-mean-square error in the high PAS group was presentfor all target speeds as seen in Fig. 2.

Pursuit gain

In this analysis we used a set of repeated measuresANOVAs to compare the pursuit gain (repeated mea-sures dependent variable) of groups with scores >2 SDin PAS, SPQ total score, SPQ factor scores, SCL90-Ranxiety and depression with the gain for the total sam-ple (see Methods section). None of these ANOVAanalyses showed a signiWcant diVerence in pursuit gainbetween the high psychometric score groups and thetotal sample with one exception. There was a signiW-cant decrease of pursuit gain in the high SPQ disorga-nization factor group compared to the total sample(F1,1083 = 5.66, P < 0.02). Moreover, the nonsigniWcantinteraction eVect of group versus target speed(F2,2166 = 0.95, P > 0.3) conWrmed that this decrease inpursuit gain in the high disorganization group was pres-ent for all target speeds as seen in Fig. 2.

Saccade frequency

In this analysis we used a set of repeated measuresANOVAs to compare the saccade frequency (repeatedmeasures dependent variable) of groups with scores

>2 SD in PAS, SPQ total score, SPQ factor scores,SCL90-R anxiety and depression with the saccade fre-quency for the total sample (see Methods section).None of these ANOVA analyses showed a signiWcantdiVerence in pursuit gain between the high psychomet-ric score groups and the total sample with one excep-tion. There was a signiWcant increase in saccadefrequency in the high SPQ disorganization factor groupcompared to the total sample (F1,1083 = 5.26, P < 0.03).Moreover, the nonsigniWcant interaction eVect ofgroup versus target speed (F2,2166 = 0.82, P > 0.4) con-Wrmed that this increase in saccade frequency in thehigh disorganization group was present for all targetspeeds as seen in Fig. 3.

Discussion

Pursuit performance in a large sample of young men

In this report we present the data from the analysis ofpursuit records in a large sample of young adults. Inprevious reports (Smyrnis et al. 2003, 2004) we pre-sented the analysis on the antisaccade and eye-Wxationperformance of the same sample. All this data arederived from the oculomotor database of the ASPISstudy. The Wnal aim for this project is the follow-up onthese individuals at regular intervals in the comingyears for the development of psychiatric symptoms.

Our sample of young men is by deWnition not repre-sentative of the general population; however, it is gen-erally representative of young Greek males sincemilitary service is obligatory in Greece and thus allmales from all areas must participate barring disabili-ties. As such, the military provides not only a largepool of subjects, but also one which facilitates evalua-tion of psychosis in a longitudinal study. The pursuitperformance data provided in this report could be ofimportance in the large literature that uses smooth eyepursuit as a tool for investigating dysfunction in schizo-phrenia. So far all indications of a dysfunction in thistask come from group comparisons between a controlgroup and a patient group. It would be then of impor-tance to have large sample data on this task althoughthe issue remains that the speciWc pursuit task used var-ies from study to study and the deWnition of measuresof pursuit performance also diVer from study to study(Levy et al. 1993), thus limiting any attempt of general-ization of our derived mean values of pursuit perfor-mance to normative values.

Our sample included young adults (18–25 years ofage). It has been shown that pursuit performancereaches adult levels at the age of 17–18 and then

123

Exp Brain Res (2007) 179:397–408 405

remains stable until the age of 65 (Katsanis et al. 1998).Younger children show worse pursuit performancethan adolescents and adults (Katsanis et al. 1998),while older adults (older than 65 years) also show adeterioration of pursuit performance (Kaufman andAbel 1986). Thus, our sample is a representative esti-mate of performance within the adult range. Two otherconfounding factors are education and gender. Wefound that the level of education did not have a signiW-cant eVect on any of the pursuit indices measured inour study. Finally, to our knowledge, the issue of gen-der diVerences in the pursuit performance of healthyadults has not been systematically addressed in the rel-evant literature although in one study diVerences ofpursuit performance between men and women did notreach statistical signiWcance (Simons and Katkin 1985).

Pursuit quality and schizotypy

In this study root-mean-square error was used to assessthe pursuit quality in a quantitative way. This measureis considered analogous to the qualitative assessmentof pursuit records that has been widely used in earlierstudies (Levy et al. 1993). There was no relation ofroot-mean-square error and psychometric variablescores in the total sample, thus refuting the hypothesisof a phenotypic continuum of ETD and schizotypy aspsychosis proneness predictors in the population (seeIntroduction). Furthermore, as predicted by the schizo-taxia hypothesis, only a small group of individuals withvery high scores in PAS showed a signiWcant increaseof root-mean-square error for all target speeds, thusshowing ETD. It should be emphasized here that ourresults should not be seen as an “acid test” to rejectone or the other of the hypotheses of a phenotypic con-tinuum and of schizotaxia. What we claim is that ourdata indeed are compatible with the schizotaxiahypothesis and favor the existence of a schizotypalgroup with ETD. In this respect our data conWrm pre-vious reports on eye-tracking abnormalities in the indi-viduals with very high scores in schizotypy measuredwith PAS (O’Drsicoll et al. 1998; Gooding et al. 2000)and extend these Wndings to conWrm the absence ofsigns of ETD in the majority of individuals that do nothave a very high score on PAS. In a previous report(Smyrnis et al. 2003) we showed that the same highPAS group had a higher rate of antisaccade errors anda higher variability in the latency of antisaccade eyemovements while again there was no relation of anti-saccade performance indices and schizotypy in thetotal sample. In a recent study Holahan and O’Driscoll(2005) found that individuals with high scores in PAShad a signiWcantly higher antisaccade error rate and

ETD and suggested that a group with predominantlypositive symptoms of schizotypy presented with a com-bination of antisaccade and smooth pursuit deWcits.The converging evidence from these studies is that aphenotypic group that shares positive psychotic-likeexperiences and eye movement abnormalities in boththe saccadic and smooth pursuit systems might be dis-tinct from the population of healthy individuals. In thisline of reasoning it is important to point that the PASscale assesses a qualitative latent entity that is found atthe highest 5–10% of the score distribution, thus iden-tifying schizotypy as a category (Lenzenweger 1994).In contrast the SPQ scale has been decomposed intothree factors and it has been used to assess diVerentdimensions of schizotypy in the population favoringthe completely dimensional view of psychopathology(Reynolds et al. 2000). Thus, PAS is by design a betterinstrument to detect a phenotypic distinct group.

Pursuit gain, saccade frequency and schizotypy

In addition to the measurement of global pursuit qual-ity using the root-mean-square error, we measuredspeciWc indices of pursuit performance, namely thegain and the saccade frequency. For these measures weused only the middle portion of the pursuit record thatcorresponds to the eye crossing the primary position.We chose this window because it is considered as theoptimal window where the pursuit system should func-tion perfectly (Leigh and Zee 1991). Any deviances onthese measures are indicative of a speciWc deWcit in thepursuit eye movement system. In the relevant litera-ture it has been discussed that speciWc measures of pur-suit performance are not as good indicators of ETD inschizophrenia as the more global measures (Levy et al.1993). The two studies that used speciWc measures ofpursuit performance in schizotypy, such as gain andsaccade frequency, report contradicting results. Sieveret al. (1994) found no diVerence in pursuit gain andsaccade frequency for individuals with schizotypal per-sonality disorder. Gooding et al. (2000) found thatindividuals with high scores in self-reporting schizotypyscales had signiWcantly lower pursuit gain than controlsand no diVerence in saccade frequency.

We found no relation of gain or saccade frequencywith the psychometric variables except for an interac-tion eVect of saccade frequency and target speed forthe negative factor of SPQ. There was an increase ofsaccade frequency with an increase in negative schizo-typy and this eVect was observed only at the highestspeed of 30 deg/s. The magnitude of this eVect thoughwas extremely small (the common variance explainedwas less than 1%).

123

406 Exp Brain Res (2007) 179:397–408

Using the high score groups resulted in a signiWcanteVect only for the high SPQ disorganization factorgroup. This group had lower pursuit gain and highersaccade frequency at the primary position compared tothe total sample for all target speeds. The dysfunctionsin this group are in speciWc areas of cognition. Interest-ingly, a signiWcant correlation of the disorganizationfactor of SPQ and Wxation instability was found in ourprevious study (Smyrnis et al. 2004). It could be arguedthen that another high-risk group, partially distinctfrom the high PAS group (the two groups overlappedfor nine individuals), is related to a speciWc instabilityin the eye Wxation and pursuit system. In accordancewith this view Holahan and O’Driscoll (2005) foundthat ETD but not antisaccade performance deWcitswere present in a group with high scores in the negativesymptoms of schizotypy measured with the physicalAnhedonia scale. Individuals with very high scores inthis scale have cognitive deWcits that are also observedin schizophrenia. Thus a speciWc deWcit in smooth eyepursuit could be associated with the proWle of cognitivedysfunction found in the disorganization cluster ofschizophrenia symptoms.

If ETD is related to cognitive dysfunction and disor-ganization symptoms in schizophrenia spectrum disor-ders then it is of importance to know the neuralsubstrate of ETD in schizophrenia. Although there is avery large literature on smooth pursuit abnormalitiesin schizophrenia and the schizophrenia spectrum disor-ders, there are very few data on the possible neuralsubstrate of ETD dysfunction in schizophrenia. It isknown from neuroimaging studies that smooth eyepursuit performance activates an extended network ofcortical and subcortical areas including the frontal eyeWelds, supplementary eye Welds, intraparietal sulcus,precuneous, extrastriate areas (medial temporal cortexand medial superior temporal cortex) and the cingulatecortex (Berman et al. 1999; O’Driscoll et al. 2000).O’Driscoll et al. (1999) used PET in a group of rela-tives of patients with schizophrenia and showed thatthose relatives with ETD had signiWcantly less activa-tion of the frontal eye Welds compared to relatives withnormal pursuit. Tregellas et al. (2004) used fMRI andreported that medicated patients with schizophreniacompared to controls showed increased activation inthe posterior hippocampi and the right fusiform gyrousin smooth eye pursuit. In a region of interest analysis ofareas known to be activated during smooth eye pursuit,they observed decreased activation in the frontal eyeWeld, cingulate gyrous and medial occipital cortex. Inanother study Hong et al. (2005) examined the neuralsubstrate in predictive smooth eye pursuit in medicatedpatients with schizophrenia and normal controls and

showed that in this case, where subjects do not rely onretinal motion signals for pursuit, the patients haddecreased activation in the frontal and supplementaryeye Welds, medial superior temporal cortex and cingu-lated gyrous. In yet another study Lencer et al. (2005)examined MT and MST in smooth eye pursuit andshowed reduced activity in medicated patients withschizophrenia. Finally in a recent study comparing Wrstepisode never-medicated patients with schizophreniaand controls, Keedy et al. (2006) found reduced activ-ity in all cortical areas involved in smooth eye pursuitcontrol and suggested a system level dysfunction. Allthese studies point to a diVuse dysfunction of corticalareas involved in pursuit in schizophrenia. There areyet no studies of the neural substrate of ETD in schizo-typy. In conclusion there is no clear evidence of speciWccortical dysfunction related to ETD in schizophreniaand the question of cortical dysfunction related toETD in schizotypy has not yet been investigated.

Pursuit performance and depression–anxiety

The analysis of pursuit performance conWrmed thatthere was no eVect of the levels of current state anxietyor depression on pursuit performance in the total sam-ple. This was true both for the root-mean-square errormeasuring global pursuit quality and speciWc measuressuch as gain and saccade frequency. Furthermore therewas no diVerence in any of the pursuit indices of per-formance in the high anxiety and high depressiongroups compared to the total sample. Since anxiety anddepression symptoms are often present in individualswith schizotypal personality characteristics, the lack ofrelation of these symptoms with pursuit performance isfurther evidence for the speciWcity of the relation ofETD and high schizotypy.

Conclusions

Smooth eye pursuit performance was assessed in alarge sample of young men. Pursuit performance dete-riorated with increasing target speed, as expected. Thelevel of education and current state psychopathology(anxiety-depression) had no eVect on pursuit perfor-mance in this population. Increasing schizotypy wasnot related to pursuit performance in the total sample.SpeciWc subgroups though with very high schizotypyhad some indication of an ETD compared to the totalsample favoring the hypothesis of schizotaxia. Theresults from this study, in combination with our previ-ous Wndings for antisaccade and Wxation performancedeWcits, favor the hypothesis that psychometrically

123

Exp Brain Res (2007) 179:397–408 407

deWned subpopulations with speciWc eye movementfunction deWcits within the general population, mightpresent high-risk groups for the development of psy-chosis and oVer the possibility to study candidate genesin schizophrenia in these groups.

Acknowledgments This work was supported by the grant“EKBAN 97” to Professor C.N. Stefanis from the General Secre-tariat of Research and Technology of the Greek Ministry ofDevelopment. “Intrasoft Co” provided the technical support forthis project. We would like to thank the following colleagues, inalphabetical order, that helped in data acquisition and prepro-cessing: Katerina Eustratiadi, Ioannis Giouzelis, Georgios Kastri-nakis, Catherine Paximadis, Christos Theleritis.

References

Berman RA, Colby CL, Genovese CR, Voyovodic JD, Luna B,Thulborn KR, Sweeney JA (1999) cortical networks subserv-ing pursuit and saccadic eye movements in humans: an fMRIstudy. Hum Brain Map 8:209–225

Derogatis LR, Lipman RS, Rickels K, Uhlenhuth EH, Covi L(1974) The Hopkins symptom checklist (HSCL): a self-report symptom inventory. Behav Sci 19:1–15

Chapman LJ, Chapman JP, Raulin ML (1978) Body image aber-ration in schizophrenia. J Abnorm Psychol 87:399–407

Chapman LJ, Chapman JP, Kwapil TR, Zinser MC (1994) Puta-tively psychosis prone subjects 10 years later. J AbnormPsychol 103:171–183

Claridge G (1994) Single indicator of risk for schizophrenia: prob-able fact or likely myth? Schizophr Bull 20:151–168

Cloninger CR, Svrakic DM, Przybeck TR (1993) A psychobiolog-ical model of temperament and character. Arch Gene Psy-chiatr 50:975–990

Eysenck HJ, Eysenck SBG (1976) Psychoticism as a dimension ofpersonality. Hodder & Stoughton, London

Gooding DC, Miller MD, Kwapil TR (2000) Smooth pursuit eyetracking and visual Wxation in psychosis-prone individuals.Psychiatr Res 93:41–54

Holahan AV, O’Driscoll GA (2005) Antissacade and smoothpursuit performance in positive- and negative-symptomschizotypy. Schizophr Res 76:43–54

Holzman PS, Levy DL (1977) Smooth pursuit eye movementsand functional psychoses: a review. Schizophr Bull 3:15–27

Holzman PS, Matthysse S (1990) The genetics of schizophrenia: areview. Psychol Sci 1:279–286

Holzman PS, Proctor LR, Hughes DW (1973) Eye-tracking pat-terns in schizophrenia. Science 181:179–181

Holzman PS, Proctor LR, Levy DL, Yasillo NJ, Meltzer HY,Hurt SW (1974) Eye-tracking dysfunctions in schizophrenicpatients and their relatives. Arch Gen Psychiatr 31:143–151

Hong LE, Tagamets M, Avila M, Wonodi I, Holcomb H, ThakerGK (2005) SpeciWc motion processing pathway deWcit duringeye tracking in schizophrenia: a performance-matched func-tional magnetic resonance imaging study. Biol Psychiatr57:726–732

Katsanis J, Iacono WG, Harris M (1998) Development of oculo-motor functioning in preadolescence, adolescence, andadulthood. Psychophysiology 35:64–72

Kaufman SR, Abel LA (1986) The eVects of distraction on smoothpursuit in normal subjects. Acta Otolaryngol 102:57–64

Keedy SK, Ebens CL, Keshavan MS, Sweeney JA (2006) Func-tional magnetic resonance imaging studies of eye movements

in Wrst episode schizophrenia: Smooth pursuit, visually guid-ed saccades and the oculomotor delayed response task. Epub

Leigh JR, Zee DS (1991) The neurology of eye movements, 2ndedn. F.A. Davis Company, Philadelphia

Lencer R, Nagel M, Sprenger A, Heide W, Binkofski F (2005)Reduced neuronal activity in the V5 complex underliessmooth-pursuit deWcit in schizophrenia: evidence from anfMRI study. Neuroimage 24:1256–1259

Lencz T, Raine A, Scerbo A, Redmon M, Brodish S, Holt L, BirdL (1993) Impaired eye tracking in undergraduates withschizotypal personality disorder. Am J Psychiatr 150:152–154

Lenzenweger MF (1994) Psychometric high-risk paradigm, per-ceptual aberrations and schizotypy: an update. SchizophrBull 20:121–135

Levy DL, Holzman PS, Matthysse S, Mendell N (1993) Eye-tracking dysfunction and schizophrenia: a critical perspec-tive. Schizophr Bull 19:461–536

Matthysse S, Holzman PS, Lange K (1986) The genetic transmis-sion of schizophrenia: Application of mendelian latent struc-ture analysis to eye tracking dysfunctions in schizophreniaand aVective disorder. J Psychiatr Res 20:57–65

Meehl PE (1989) Schizotaxia revisited. Arch Gen Psychiatr46:935–944

Ntonias S, Karastergiou A, Manos N (1990) Standardization ofthe symptom checklist 90-R rating scale in a Greek popula-tion. Psychiatriki 2:42–48

O’Driscoll GA, Lenzenweger MF, Holzman PS (1998) Antisac-cades and smooth pursuit eye tracking and schizotypy. ArchGen Psychiatr 55:837–843

O’Driscoll GA, Benkelfat C, Florencio PS, WolV AL, Joober R,Lal S, Evans AC (1999) Neural correlates of eye trackingdeWcits in Wrst-degree relatives of schizophrenic patients: apositron emission tomography study. Arch Gen Psychiatr56:1127–1134

O’Driscoll GA, WolV AL, Benkelfat C, Florencio PS, Lal S,Evans AC (2000) Functional neuroanatomy of smooth pur-suit and predictive saccades. Neuroreport 11:1335–1340

Raine A (1991) The SPQ: a scale for the assessment of schizotyp-al personality in a non-clinical sample-the role of task de-mand. Schizophr Bull 17:555–564

Reynolds CA, Raine A, Mellingen K, Venables PH, MednickSA (2000) Three-factor model of schizotypal personality:Invariance across culture, gender, religious aYliation, fam-ily adversity, and psychopathology. Schizophr Bull 26:603–618

Siever LJ, Coursey RD, Alterman IA, Buchsbaum MS, MurphyDL (1984) Impaired smooth pursuit eye movement: vulner-ability marker for schizotypal personality disorder in a nor-mal volunteer population. Am J Psychiatr 141:1560–1566

Siever LJ, Keefe R, Bernstein DP, Coccaro EF, Klar HM, Ze-mishlany Z, Peterson AE, Davidson M, Mahon T, HovarthT, Mohs R (1990) Eye tracking impairment in clinically iden-tiWed patients with schizotypal personality disorder. Am JPsychiatr 147:740–745

Siever LJ, Friedman L, Moscowitz J, Mitropoulou V, Keefe R,Roitman SL, Merhige D, Trestman R, Silverman J, Mohs R(1994) Eye movement impairment and schizotypal psycho-pathology. Am J Psychiatr 151:1209–1215

Simons RF, Katkin W (1985) Smooth pursuit eye movements insubjects reporting phsycial nhedonia and perceptual aberra-tions. Psychiatr Res 14:275–289

Smyrnis N, Evdokimidis I, Stefanis NC, Avramopoulos D, Co-stantinidis TS, Stavropoulos A, Stefanis CN (2003) Antisac-cade performance of 1273 men: eVects of schizotypy, anxiety,and depression. J Abnorm Psychol 112:403–414

123

408 Exp Brain Res (2007) 179:397–408

Smyrnis N, Kattoulas E, Evdokimidis I, Stefanis NC, Avramopo-ulos D, Pantes G, Theleritis C, Stefanis CN (2004) Active eyeWxation performance in 940 young men: eVects of IQ, schizo-typy, anxiety and depression. Exp Brain Res 156:1–10

Stefanis NC, Smyrnis N, Avramopoulos D, Evdokimidis I, Tzou-fras I, Stefanis CN (2004) Factorial composition of self ratedschizotypal traits amongst young males undergoing militarytraining: The ASPIS. Schizophr Bull 30:335–350

Thaker GK, Cassady S, Adami H, Moran M, Ross DE (1996) Eyemovements in spectrum personality disorders: comparison ofcommunity subjects and relatives of schizophrenic patients.Am J Psychiatr 153:362–368

Tregellas JR, Tanabe JL, Miller DE, Ross RG, Olincy A, Freed-man R. (2004) Neurobiology of smooth pursuit eye move-ment deWcits in schizophrenia: an fMRI study. Am JPsychiatr 161:315–321

123