R

Wc

ASJa

b

c

d

e

f

h

•••••

a

ARRAA

KFPVVWE

F

0h

Behavioural Brain Research 263 (2014) 70–79

Contents lists available at ScienceDirect

Behavioural Brain Research

jou rn al hom epage: www.elsev ier .com/ locate /bbr

esearch report

orking memory and visuospatial deficits correlate with oculomotorontrol in children with fetal alcohol spectrum disorder

ngelina Paolozzaa, Carmen Rasmussenb, Jacqueline Peib, Ana Hanlon-Dearmanc,arah M. Nikkeld, Gail Andrewe, Audrey McFarlanef, Dawa Samdupa,ames N. Reynoldsa,∗

Centre for Neuroscience Studies, Queens University, Kingston, ON, CanadaDepartment of Pediatrics, University of Alberta, Edmonton, AB, CanadaManitoba FASD Centre, University of Manitoba, Winnipeg, MB, CanadaDepartment of Genetics, Children’s Hospital of Eastern Ontario, Ottawa, ON, CanadaGlenrose Rehabilitation Hospital, Edmonton, AB, CanadaLakeland Centre for FASD, Cold Lake, AB, Canada

i g h l i g h t s

Eye movement tasks can be used to understand brain damage caused by PAE.FASD group performed worse than controls working memory and visuospatial measures.FASD working memory deficits in psychometric and eye movement tests were related.FASD visuospatial deficits were in psychometric and eye movement tests were related.Eye movement control tasks can help with early identification of those with FASD.

r t i c l e i n f o

rticle history:eceived 21 October 2013eceived in revised form 9 January 2014ccepted 20 January 2014vailable online 29 January 2014

eywords:etal alcohol spectrum disorderrenatal alcohol exposureisuospatial skillsisual-spatial skillsorking memory

ye movements

a b s t r a c t

Previous studies have demonstrated that children with Fetal Alcohol Spectrum Disorder (FASD) exhibitdeficits in measures of eye movement control that probe aspects of visuospatial processing and workingmemory. The goal of the present study was to examine, in a large cohort of children with FASD, pre-natal alcohol exposure (PAE) but not FASD, and typically developing control children, the relationshipbetween performance in eye movement tasks and standardized psychometric tests that assess visuospa-tial processing and working memory. Participants for this dataset were drawn from a large, multi-siteinvestigation, and included children and adolescents aged 5–17 years diagnosed with an FASD (n = 71),those with PAE but no clinical FASD diagnosis (n = 20), and typically developing controls (n = 111). Par-ticipants completed a neurobehavioral test battery and a series of saccadic eye movement tasks. TheFASD group performed worse than controls on the psychometric and eye movement measures of work-ing memory and visuospatial skills. Within the FASD group, digit recall, block recall, and animal sortingwere negatively correlated with sequence errors on the memory-guided task, and arrows was nega-tively correlated with prosaccade endpoint error. There were no significant correlations in the control

group. These data suggest that psychometric tests and eye movement control tasks may assess sim-ilar domains of cognitive function, and these assessment tools may be measuring overlapping brainregions damaged due to prenatal alcohol exposure. The results of this study demonstrate that eye move-ment control tasks directly relate to outcome measures obtained with psychometric tests and are ableto assess multiple domains of cognition simultaneously, thereby allowing for an efficient and accurateassessment.

Abbreviations: FASD, Fetal Alcohol Spectrum Disorder; PAE, Prenatal Alcohol ExposureP, fixation point; SES, socioeconomic status; ANOVA, one-way analysis of variance.∗ Corresponding author at: Botterell Hall, 18 Stuart Street, Queen’s University Kingston

E-mail address: jnr@queensu.ca (J.N. Reynolds).

166-4328/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.bbr.2014.01.024

© 2014 Elsevier B.V. All rights reserved.

; NEPSYII, Neuropsychological Assessment; WMTB, Working Memory Test Battery;

, Ontario, Canada, K7L 3N6. Tel.: +61 3 533 6946; fax: +61 3 533 6840.

l Brain

1

etFAgaadtsniaaicsw

ocmthmadIsms

ucrcnmiar[hc[

epalvarFc

wctPoom

A. Paolozza et al. / Behavioura

. Introduction

The full spectrum of adverse effects induced by prenatal alcoholxposure, which includes several diagnostic subgroups, is collec-ively referred to as Fetal Alcohol Spectrum Disorder (FASD) [1].ASD is estimated to occur in at least 1% of the population in Northmerica [2]. The developing brain is the principal target organ forestational alcohol exposure and this type of injury can manifests intellectual, neurological and behavioural abnormalities. Thesebnormalities can produce a wide variety of mild to severe brainysfunctions in processes such as learning and memory, execu-ive function, social communication, attention, and sensory-motorkills [3–5], which may contribute to the negative behavioural,europsychiatric, and maladaptive outcomes commonly observed

n this population [6]. This study focused on working memorynd visuospatial deficits, which are prevalent problems for thoseffected by FASD. Poor working memory and/or visuospatial skillsn children are associated with poor academic performance espe-ially in reading and mathematics [7], and thus it is likely that theseame deficits contribute to poor academic outcomes in childrenith FASD.

Saccades are rapid eye movements that bring new visual targetsnto the fovea of the retina. The circuitry responsible for the effi-ient and accurate execution of saccadic eye movements involvesultiple cortical and subcortical brain regions, and the roles that

hese brain regions play in controlling eye movement behaviorsave been extensively investigated [8–10]. Measurement of eyeovement control is a powerful tool for assessing sensory, motor

nd cognitive function in healthy participants, as well as multipleiseases and disorders [10], including children with FASD [11–13].

n particular, children with FASD have been shown to have pooraccade accuracy both when looking to visual targets and whenaking saccades to the remembered locations of previously pre-

ented targets [13].Working memory is the ability to temporarily store and manip-

late information to complete complex tasks such as languageomprehension, learning, and reasoning. According to Baddeley’sevised model [14], working memory consists of four elements: theentral executive is responsible for the control of attention and isecessary for preserved immediate recall and integration of infor-ation; the phonological loop provides temporary storage of verbal

nformation; the visuospatial sketchpad provides temporary stor-ge of visual and spatial representations; and the episodic buffer isesponsible for integrating different types of information together14]. Previous research has found that children with prenatal alco-ol exposure show deficits on tests that assess the function of theentral executive, phonological loop, and visuospatial sketchpad15–18].

Visuospatial ability involves a wide variety of skills that canntail visual working memory. These visuospatial abilities includerocesses such as spatial orientation (imagining how an image willppear from another perspective), spatial visualization (manipu-ating and encoding spatial forms), figural flexibility (imagining aariety of ways to involve a spatial problem), closure speed (initi-ting an apparent disparate perception into a unified concept), andeference memory (assesses spatial memory) [19]. Children withASD have been reported to exhibit poorer visuospatial skills whenompared to controls on a variety of tasks [20–23].

Previously, we have shown that visuospatial processing andorking memory are impaired on eye movement control tasks in

hildren with FASD [13]. The aim of this study was to examine theseasks in a larger sample of children with FASD as well as those with

AE but not FASD. In a previous study, we showed that measuresf response inhibition in children with prenatal alcohol exposurebtained from standardized psychometric tests are correlated toeasures of response inhibition obtained from eye movement

Research 263 (2014) 70–79 71

control tasks [24]. Saccadic eye movement tasks can be designed toassess multiple domains of CNS function (sensory, motor, cognitive)simultaneously. Given the extensive knowledge concerning theneural circuitry underlying eye movement control [8–10], exami-nation of the domains of CNS function that are impaired during theperformance of eye movement control tasks may be used to helpidentify brain regions that have been injured by prenatal alcoholexposure. Based on overlapping brain areas believed to be engagedduring tests of working memory and visuospatial processing andeye movement control tasks, we hypothesized that a relation-ship should exist between these measures. More specifically, wehypothesized that the frequency of sequence errors, a measure ofworking memory, in the memory-guided saccade task will corre-spond to lower scores in specific psychometric tests that also probeworking memory. Additionally, we predicted that an increase in thesaccade endpoint error, a measure of visuospatial processing, inthe prosaccade task will correspond to lower scores in psychome-tric tests that also assess visuospatial processing skills. Therefore,we correlated the performance of children with FASD in specificeye movement control tasks to scores in separate tests of workingmemory and visuospatial skills.

2. Methods

2.1. Participants

All experimental procedures were reviewed and approved bythe Human Research Ethics Board at Queen’s University, Universityof Alberta, Children’s Hospital of Eastern Ontario, and University ofManitoba. Informed consent was obtained from the parent beforethe protocol, and the children also completed an assent form. Chil-dren with FASD (FAS, pFAS, ARND; n = 71) were assessed accordingto the Canadian Guidelines [25] and recruited through referralsfrom clinicians in diagnostic clinics from Kingston, ON, Ottawa,ON, Edmonton, AB, Cold Lake, AB, and Winnipeg, MB, as part ofa large, multi-site study [26]. Children who had confirmed prena-tal alcohol exposure from a credible source (PAE; n = 20) but didnot have a diagnosis of FASD because they do not meet all of thediagnostic criteria were also recruited. Typically developing chil-dren (n = 111) were recruited from the same geographical areas andwere excluded if they had any neurological or psychiatric disorder,or visual disturbance, other than requiring corrective lenses. Allgroups were matched as closely as possible for age and sex. Due toconstraints such as age restrictions, quality control, and geographicarea not all children completed both psychometric testing and eyemovement control experiments. Participants were asked to with-hold any medications (Table 1) typically taken on the day of testingto minimize any confounding effects of stimulant drugs, which haverelatively short half-lives. Socioeconomic status (SES) was calcu-lated using Hollingshead’s Four-Factor Index of Social Status forthe FASD, PAE, and control groups and analyzed for group differ-ences [27]. Participants received gift cards for the 2-h test session.We have previously reported data from this cohort showing cor-relations between measures of response inhibition obtained fromeye movement control tasks and psychometric tests [24].

2.1.1. Saccadic eye movement recordingsParticipants were seated comfortably in a dark, quiet room on a

stable chair. Eye position was recorded using the Eyelink 1000 (SRResearch, Kanata, ON). The 17′′ LCD monitor and mounted infraredcamera were at a distance of 58–64 cm from the left eye. The

position of the left pupil was digitized in both the vertical and hori-zontal axes at a sampling rate of 500 Hz. Performance was assessedin three tasks: the prosaccade, antisaccade and the memory-guided saccade tasks. Before each task the eye movements of each

72 A. Paolozza et al. / Behavioural Brain Research 263 (2014) 70–79

Table 1Demographic information for control, FASD and PAE group.

Diagnosis n

Fetal alcohol syndrome 8Partial fetal alcohol syndrome 14Alcohol related neurodevelopmental disorder 50Prenatal alcohol exposure 22

Comorbidities FASD n (%) PAE n (%) Control n (%) Chi-square (p-value)

Attention hyperactivity deficit disorder 45 (63)* 12 (57)* 0 (0) <0.0001Anxiety disorder 9 (13)* 2 (10)* 0 (0) 0.0009Oppositional defiant disorder 10 (14)* 2 (10)* 0 (0) 0.0004Language processing disorders 10 (14)* 1 (5) 0 (0) 0.0002Depression 4 (6)* 3 (14)* 0 (0) 0.0026Other neurological disorder 10 (14)* 5 (24)* 0 (0) <0.0001

Medication FASD n (%) PAE n (%) Control n (%) Chi square (p-value)

Stimulants 32 (44)* 7 (33)* 0 (0) <0.0001Antipsychotics 15 (21)* 2 (10)* 0 (0) <0.0001Antidepressants 7 (10)* 2 (10)* 0 (0) 0.0038Other 3 (4)† 5 (24)* 0 (0) <0.0001

Other FASD (average ± SD) PAE (average ± SD) Control (average ± SD) ANOVA (p-value)

Age 11.4 ± 3 years* 10.9 ± 2 years 10.1 ± 3 years 0.019Socioeconomic status 40.8 ± 14* 40.8 ± 11* 46.8 ± 8 0.0015

Ethnicity FASD n (%) PAE n (%) Control n (%) Chi square (p-value)

First Nations/Metis 44 (61)*† 18 (86)* 1 (1) <0.0001Caucasian 25 (35)* 3 (14)* 106 (96) <0.0001Other 3 (4) 0 (0) 4 (3) ns

S der grF roup

ptaceniw

2

iFptptca

ietlatpvbattbd

D = standard deviation; Con = Control group; FASD = Fetal Alcohol Spectrum Disorisher’s exact or Tukey’s post-test (where appropriate); †p < 0.05 compared to PAE g

articipant were calibrated using 9 on screen targets (8 aroundhe periphery and 1 central). The targets were flashed sequentiallyround the screen and the participant fixated on each one. Afteralibration, the process was repeated to validate that the averagerror between fixation and target was less than 2 degrees and thato loss of eye tracking occurred. This also ensured that the partic-

pants had no visual disturbances (e.g., nystagmus, diplopia) thatould impair task performance.

.1.2. Behavioral tasksIn the prosaccade and antisaccade task, each trial started with

llumination of a central fixation point (FP) for 800–1200 ms. TheP then disappeared and, after a delay of 200 ms (gap period), aeripheral target appeared randomly at 10◦ to the left or right ofhe central FP. Participants were given 1000 ms to initiate and com-lete a saccade to the correct location and were instructed to lookowards the target (prosaccade) or away from the target (antisac-ade). No error feedback was given. One block each of 60 prosaccadend 60 antisaccade trials were obtained from each participant.

In the memory-guided saccade task, participants werenstructed to maintain fixation at the central FP while two periph-ral targets appeared sequentially, and after the FP disappeared,hey were required to make sequential saccades to the rememberedocations of the peripheral targets in the same order they originallyppeared. The screen was divided into four quadrants in whichhe peripheral targets could appear. Each quadrant consisted of 9otential target locations in an invisible 3 × 3 grid centered at a 10◦

isual angle from the FP. The FP was illuminated for 200–1000 msefore the appearance of the two targets. The two targets thenppeared briefly in immediate succession for 100 ms each, within

wo of the four quadrants of the screen. Participants were requiredo fixate for an additional random time of 0, 600, 1200, or 1800 msetween the disappearance of the second peripheral target and theisappearance of the FP. They were instructed to remember the

oup; PAE = Prenatal Alcohol Exposure group; *p < 0.05 compared to controls usingusing Fisher’s exact or Tukey’s post-test (where appropriate).

order and spatial location of the peripheral targets, and to maketwo saccades as accurately as possible to these locations in the samesequence but only after the disappearance of the FP. A single blockof 72 trials was collected for this task.

2.1.3. Data analysis of eye tracking measuresData were analyzed using custom software developed in MAT-

LAB (R2009b, The Mathworks, Inc, Natick, Massachusetts). Saccadeswere defined as having a speed of greater than 2.5 times the stan-dard deviation of the background noise (measured during fixation)for at least 5 sample points. Only trials where the participant wasfixating on the FP at the appropriate time were used. If the partic-ipant broke fixation inappropriately (i.e. not to a target location oraway from the screen) the trial was discarded from analysis. Anytrials where eye tracking was lost were removed. To be includedin the analysis, each participant had to achieve greater than 50%viable trials in each of the tasks.

Due to the large control sample size obtained in this study wewere able to standardize the eye movement control scores by age.For all three saccadic eye movement tasks, there was a significantimprovement in task performance with increasing age. Therefore,the data were age corrected by calculating a standard score (t-score)for each age. Scores for the FASD and PAE group were then calcu-lated using the age-dependent t-score equation obtained from thecontrol group. Differences between groups were analyzed using aone-way ANOVA coupled with Tukey’s Multiple Comparison Test.Effect sizes were also calculated for the dependent variables usingCohen’s d scores. Data are expressed as mean ± s.e.m. for children inthe FASD (n = 69; average age 11.6 years, 38 males), control (n = 111;average age 10.0 years, 52 males), and PAE (n = 13; average age 10.9

years, 6 males) groups.

Saccade measures for all viable trials for the three tasks wereassessed by examining the working memory and visuospatial out-come measures. For the prosaccade and antisaccade tasks the error

l Brain

ipamwbst

2

iTma(5wvoeoc

22seiafAca

2wsTiac

2oaintn

22potT

2

sddsma

A. Paolozza et al. / Behavioura

n the saccade endpoint was defined as the angle between the idealath and the trajectory of the first saccade toward the goal (drawns a straight line from the beginning to the end of the saccade) (forore details see [13]). In the memory-guided task, individual trialsere assigned as either correct, timing errors (saccades initiated

efore 90 ms after the go signal), and/or sequence errors (initialaccade made closer to the second peripheral target location thano the first target, therefore in the wrong sequence).

.1.4. Psychometric testsThe Neuropsychological Assessment II (NEPSY-II) is a standard-

zed psychometric test battery for children 3–16 years of age [28].his battery assesses multiple domains of executive functioning,emory, sensorimotor functioning, social perception, language,

nd visuospatial processing. The Working Memory Test BatteryWMTB) is a standardized psychometric test battery for children–15 years of age [29]. This battery assesses working memoryith subtests measuring central executive, phonological loop and

isuospatial sketchpad components of working memory basedn Baddeley’s model [14]. Each subtest used included a teachingxample and a practice round where the experimenter had thepportunity to ensure the child understood the instructions andorrect the child’s mistakes.

.1.5. Working memory psychometric subtests:

.1.5.1. Animal sorting. This subtest is a card-sorting task that mea-ures concept formation, ability to apply concepts, and set shifting,ach of which requires a functioning working memory. The childs shown eight cards and is asked to look for ways that the cardsre the same and different. The child must then come up with dif-erent, self-initiated ways to sort the cards into two groups of four.n error occurs when the child completes a card sort that is not aorrect sort or the child repeats a card sort. The test is discontinuedfter 360 s of cumulative sort time.

.1.5.2. Digit recall. This subtest measures verbal/phonologicalorking memory. The child listens to a string of numbers (1 per

econd) and attempts to repeat these numbers in the same order.his is presented in a span paradigm where the number of digitsncrease incrementally until the child makes three errors in a row,t which time the task is discontinued. The number and span oforrect trials are recorded.

.1.5.3. Block recall. This subtest measures spatial working mem-ry. The child watches the experimenter tap a sequence of blocks in

three-dimensional array and attempts to reproduce that sequencen the same order. This is presented in a span paradigm where theumber of blocks is increased incrementally until the child makeshree errors in a row, at which time the task is discontinued. Theumber and span of correct trials are recorded.

.1.6. Visuospatial psychometric subtest

.1.6.1. Arrows (ages 5–16). This subtest measures the ability of thearticipant to assess line orientation. The child is shown an arrayf arrows around a target and asked to say which arrow(s) pointso the center of the target. The number of correct trials is recorded.he test is discontinued after five consecutive scores of 0.

.1.7. Data analysis of psychometric measuresAnimal sorting and arrows yield raw, scaled, and percentile

cores based on age. The mean scaled score is 10 with a standardeviation of 3. Digit recall and block recall also yield raw, stan-

ardized, and percentile scores based on age. The mean standardcore is 100 with a standard deviation of 10. Similar to the eyeovement scores, the raw scores improved with age and once

ge was removed by using the standard scores, the controls all

Research 263 (2014) 70–79 73

fluctuated around average (10 or 100, depending on the test) asexpected for a random population of typically developing controlchildren. Standard scores of correct trials were used for all analysesexcept animal sorting where the combined scaled score was usedbecause this takes into account both correct and error trials. Dif-ferences between groups were analyzed using an ANOVA, coupledwith Tukey’s Multiple Comparisons Test. Effect sizes were also cal-culated for the dependent variables using Cohen’s d scores. Data areexpressed as mean ± s.e.m. for children in the FASD (n = 72; averageage 11.5 ± 3 years, 38 males), control (n = 90; average age 10.0 ± 3years, 44 males) and PAE (n = 22; average age 10.9 ± 3 years, 10males) groups.

IQ was not obtained from any of the participants in this study,since IQ has not been consistently shown to affect performanceon psychometric tests [30], and matching groups on IQ can resultin mismatching on other important variables [31]. Several studiesin children and adolescents with FASD have found that executivefunctioning deficits remain even after controlling for IQ [32]. Thedegree to which controlling for IQ alters the association of prena-tal alcohol exposure to the outcome depends on how closely themeasure overlaps with cognitive aspects of IQ [33].

2.2. Correlation analysis

Hypothesis-driven Pearson correlations were used to identifywhether psychometric scores were associated with eye movementcontrol measures. The scores from each measure were chosen if theFASD group was significantly different from controls. Bonferronicorrection for multiple comparisons was applied to the correla-tional analyses. Children in the FASD (n = 64; average age 11.5 years,37 males) and control (n = 61; average age 10.1 years, 29 males)groups, who completed both psychometric testing and eye move-ment tasks were matched as closely as possible for age and sex. ThePAE group was excluded from the correlational analysis becausethe number of PAE participants that completed both psychometrictesting and eye movement tests was too small to run the group asa separate correlation, and they could not be combined with theFASD group because no deficit was found compared to controls.

3. Results

3.1. Demographic variables

To investigate the possible role of demographic variables weperformed several analyses on age, comorbidities, SES, groupsubtypes, and ethnicity (Table 1). The one-way ANOVA revealed asignificant main effect of age (F(2,205) = 4.0, p = 0.019); the FASDgroup was significantly older than the control group. As previouslystated, we controlled for age by performing age corrections onall of the outcome measures. Children with FASD with or withouta comorbid disorder of ADHD were compared on all measurespresented by first visually examining the data, followed by t-testsbetween those with or without a comorbid disorder of ADHD. Inagreement with a previous study [11], no differences were found.Due to the low frequency of other comorbid disorders we were onlyable to investigate ADHD. We next performed ANOVAs to examinesubgroup differences by dividing the FASD group into two sub-groups (FAS/pFAS and ARND), which were then compared to eachother and to the control and PAE group. No group differences werefound between the two FASD groups (FAS/pFAS and ARND) on anypsychometric or eye movement measure. Given that the FAS/pFAS

group is relatively small compared to our ARND group, differencesmay be detected between diagnostic subgroups with a largersample size. The SES was examined for group differences using anANOVA and the control group was found to have a significantly

7 l Brain

hptgsacTa(lfatTdcadg

3

tte(iSctAiF

3

ltbs1ctFs

3

cipagbsbTfse

4 A. Paolozza et al. / Behavioura

igher status than both the FASD and PAE group (F(2,173) = 6.7, = 0.0015). However, the SES showed no significant relationshipo any of the psychometric or eye movement measures in anyroup. This was analyzed by both visually examining the data andubsequently running Pearson’s correlations. Ethnicity was alsonalyzed since the majority of the control group (96%) was Cau-asian and the majority of our FASD group was First Nations (61%).his was accomplished by dividing the FASD group into those whore identified as First Nations and those with any other ethnicityprimarily Caucasian). The data was then examined visually toook for any trending disparities, and then t-tests were run to lookor significant differences. There were no significant differences onny of the psychometric and eye movement measures between thewo ethnic groupings represented among the FASD participants.herefore, comorbidities, diagnostic subgroup, SES, and ethnicityo not influence the data and did not need to be included asovariates. There was an imbalance in the ethnicity of the controlnd FASD groups, and while this does not appear to affect theata it will be important in the future to better match theseroups.

.2. Eye movement control

Children with FASD performed significantly worse than con-rols on the prosaccade, antisaccade, and memory-guided saccadeasks (see Table 2). The one-way ANOVA revealed a significantffect of group for saccade endpoint error in the prosaccade taskF(2,189) = 6.1, p = 0.0028) (Fig. 1A); the FASD group was signif-cantly less accurate than the control, but not the PAE group.imilarly, the FASD group displayed significantly less accurate sac-ade endpoints compared to the control group in the antisaccadeask (F(2, 177) = 10.24, p < 0.0001) (Fig. 1B). Finally, the one-wayNOVA revealed a significant effect of group for sequence errors

n the memory-guided task (F(2,162) = 6.4, p = 0.0022) (Fig. 1C); theASD group made significantly more errors than the control group.

.3. Psychometric testing

The FASD group achieved test scores that were significantlyower than controls on all psychometric tests previously men-ioned (see Table 2). The FASD group was significantly lower thanoth the control group and the PAE group for scores on animalorting (F(2,159) = 16.9, p < 0.0001) (Fig. 2A) and digit recall (F(2,75) = 25.8, p < 0.0001) (Fig. 2B). The one-way ANOVA was signifi-ant for the arrows subtest (F(2,179) = 4.5, p = 0.013) (Fig. 2C); onlyhe FASD group was significantly lower than the control group. TheASD group also was significantly lower than the control group forcores on block recall (F(2,174) = 13.3, p < 0.0001) (Fig. 2D).

.4. Correlations

The results of the psychometric and eye movement correlationsan be found in Table 3. As hypothesized, there were no signif-cant correlations among the control group for working memorysychometric and eye movement scores (Fig. 3), indicating thatge was the major factor influencing task performance in thisroup. In contrast, the FASD group did show a negative correlationetween sequence errors in the memory-guided task and animalorting (p = 0.012, r = −0.34), digit recall (p = 0.013, r = −0.33), andlock recall (p = 0.007, r = −0.35) in the psychometric testing (Fig. 3).

here were also no significant correlations among the control groupor visuospatial measures (Fig. 4). In contrast, the FASD grouphowed a negative correlation between the prosaccade endpointrror and performance of the arrows subtest (p = 0.008, r = −0.33),

Research 263 (2014) 70–79

but there was no relationship between antisaccade endpoint errorand arrows performance (Fig. 4).

4. Discussion

4.1. General comments

The objective of this study was to examine the relationshipbetween measures of working memory and visuospatial skills inchildren prenatally exposed to alcohol obtained using differentassessment tools. Eye movement control tasks have been usedextensively to probe sensory, motor and cognitive function acrossall stages of development, and there is a large body of literaturedescribing the brain structures that contribute to the differentaspects of eye movement control. Thus, the performance of eyemovement control tasks has been used to assess brain function ina number of clinical conditions such as schizophrenia, Parkinson’s,Huntington’s, stroke, etc. [10]. In the current study, children withFASD made significantly more sequence errors on the memory-guided saccade task (i.e., failing to recall the order of targetpresentation), which suggests a deficit in spatial working mem-ory. Similarly, children with FASD performed significantly worseon the psychometric tests that specifically assess working memory(digit recall, block recall), as well as a test that is at least partiallydependent on working memory (animal sorting). Within the FASDgroup, all three psychometric tests were found to be negativelycorrelated with sequence errors in the memory-guided saccadetask. In contrast, there were no significant correlations within thecontrol group among any of these outcome measures because theeffect of age, the major contributing factor to task performance,was removed by the use of age-corrected standard scores. Thatis, in healthy controls the age-corrected performance on the eyemovement and psychometric tests of working memory fluctuatesaround the mean for all ages. Therefore, the performance of thesetasks within the control group is quite homogeneous once age isadjusted and as a consequence no correlation was found betweenthese parameters. The fact that we still see a relationship in theFASD group indicates that something other than age is driving thiscorrelation. We hypothesize that this additional factor is the braininjury induced by prenatal alcohol exposure. We next looked atvisuospatial processing. The FASD group performed significantlypoorer than controls on arrows, the psychometric test of visuospa-tial processing. The FASD group also made greater endpoint errorsin the prosaccade and antisaccade tasks, which indicates poorervisuospatial processing. Among the FASD group, only prosaccadeendpoint error negatively correlated with arrows. There were againno significant correlations in the control group. The results in theFASD group indicates that common brain structures are impor-tant for regulating working memory and visuospatial processing,and that brain injury induced by prenatal alcohol exposure affectsperformance on both psychometric and eye movement controlmeasures. Additionally, the results of this study indicate that eyemovement control tasks can assess multiple domains of cognitivefunction simultaneously.

4.2. Working memory

Working memory is the limited capacity system allowing forthe temporary storage and manipulation of information [14]. Thefour component model has been theorized to consist of the cen-tral executive which is aided by two ‘subsidiary slave systems’

called the phonological loop and the visuospatial sketchpad [14].The memory-guided saccade task is assumed to test the visuospa-tial sketchpad as the participants were required to hold visualinformation about spatial location and the order of the flashed

A. Paolozza et al. / Behavioural Brain Research 263 (2014) 70–79 75

Table 2Means and effect size (Cohen’s d) statistics on cognitive tests.

Test FASD (mean ± SEM) PAE (mean ± SEM) Con (mean ± SEM) Con vs. FASD (d) Con vs. PAE (d) FASD vs. PAE (d)

Animal Sorting 6.7 ± 0.3 9.6 ± 0.7 9.3 ± 0.4 0.91* −0.082 −1.05†

Digit Recall 82.9 ± 1.6 93.8 ± 2.9 100.1 ± 1.8 1.15* 0.38 −0.84†

Block Recall 87.1 ± 2.0 94.8 ± 2.9 100.5 ± 1.8 0.82* 0.27 −0.65Arrows 8.8 ± 0.3 9.2 ± 0.7 10.2 ± 0.3 0.51* 0.20 −0.26Pro Endpoint 58.7 ± 24.7 51.2 ± 10.1 49.9 ± 9.5 −0.47* −0.13 0.40Anti Endpoint 65.1 ± 33.3 51.21 ± 14.1 49.8 ± 9.5 −0.63* −0.11 0.54Sequence Errors 58.1 ± 20.5 54.6 ± 10.7 49.6 ± 9.2 −0.86* −.50 0.21

SD = standard deviation; Con = Control group; FASD = Fetal Alcohol Spectrum Disorder group; PAE = Prenatal Alcohol Exposure group; d = Cohen’s d effect size; *p < 0.05compared to controls; †p < 0.05 compared to PAE group.

F are mg ry-guic

dswswoI

TP

C

ig. 1. Saccadic eye movement working memory and visuospatial measures. Data

roups. (A) Prosaccade endpoint error. (B) Antisaccade endpoint error. (C) Memoontrol subjects.

ots simultaneously in memory. Previously we published thatequence errors were not different between controls and childrenith FASD [13], however this result came from a smaller FASD

ample (n = 27) and the scores were not standardized for age ase did not have a large enough control group (n = 27). The visu-

spatial sketchpad is also being examined with block recall [29].nterestingly, the correlation between sequence errors and block

able 3earson correlations between psychometric and eye movement tests.

Correlational test FASD p-value

Animal Sorting & Sequence Errors 0.012 −Digit Recall & Sequence Errors 0.013 −Block Recall & Sequence Errors 0.0070 −Arrows & Prosaccade Endpoint 0.0076 −Arrows & Antisaccade Endpoint 0.65 −

on = Control group; FASD = Fetal Alcohol Spectrum Disorder group.

ean ± s.e.m. for subjects in the control (n = 111), FASD (n = 69) and the PAE (n = 13)ded sequence errors. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 compared with

recall displayed the strongest relationship of all the working mem-ory tests assessed. This would indicate that these tasks are themost similar and are most likely assessing the same underlying

construct of working memory. Digit recall assesses the phonologi-cal loop which is assumed to hold verbal and acoustic information[29]. The relationship still holds between this task and sequenceerrors, indicating a deficit in the central executive as it serves both

FASD Pearson r Con p-value Con Pearson r

0.34 0.23 0.190.33 0.33 −0.160.35 0.74 −0.0530.33 0.19 −0.170.057 0.31 −0.14

76 A. Paolozza et al. / Behavioural Brain Research 263 (2014) 70–79

Fig. 2. Psychometric working memory and visuospatial measures. Data are mean ± s.e.m. for subjects in the control (n = 88), FASD (n = 71) and the PAE (n = 20) groups. (A)Animal sorting subtest. (B) Digit recall subtest. (C) Block recall subtest. (D) Arrows subtest. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 compared with control subjects.

Fig. 3. Correlation of sequence errors from the memory-guided task and working memory subtests from the psychometric tests. Data are individual data points for subjectsin the control (n = 61) and FASD (n = 56) groups. (A) Correlation of control group sequence errors and digit recall subtest. (B) Correlation of FASD group sequence errors anddigit recall subtest. (C) Correlation of control group sequence errors and block recall subtest. (D) Correlation of FASD group sequence errors and block recall subtest.

A. Paolozza et al. / Behavioural Brain Research 263 (2014) 70–79 77

F tasks ap ntrol gp tisaccae

so(w

cmspSmcTtsmedf

4

twianatisbhil

ig. 4. Correlation of saccade endpoint errors from the prosaccade and antisaccade

oints for subjects in the control (n = 61), FASD (n = 64) groups. (A) Correlation of corosaccade endpoint error and arrows subtest. (C) Correlation of control group anndpoint error and arrows subtest.

ystems and a deficit in this component would affect working mem-ry overall [14]. This is further supported by Carmichael-Olsen et al.1998), who found deficits in the central executive in adolescentsith FASD.

Animal sorting assesses concept formation, the ability to applyoncepts, and set shifting, which requires a functioning workingemory to successfully complete the task. The participant must

imultaneously remember what sorts have already been com-leted, while also generating new sorts using the provided cards.imilar tasks have been hypothesized to assess abstract workingemory [14]. Animal sorting requires a functioning phonologi-

al loop and central executive as well as the episodic buffer [14].he episodic buffer is a limited-capacity temporary storage sys-em that is capable of integrating information from a variety ofources and serves as an interface between a range of workingemory systems. This component is also crucially dependent on

xecutive-based attention [14]. This poor working memory in chil-ren can contribute to poor academic performance, hyperactivity,orgetting, and disruptive classroom behaviours [7].

.3. Visuospatial processing

Visuospatial skills allow for objects to be perceived and spa-ial relationships among those objects to be assessed. In agreementith previous research, the saccade endpoint error was signif-

cantly greater in the FASD group for both the prosaccade andntisaccade tasks [13]. This deficit may be due to abnormal con-ections between the cerebellum and other cortical brain regions,s the cerebellum has indirect connections with the cerebral cor-ex, basal ganglia, and thalamus, and these pathways participaten the online correction of saccade trajectories [34–36]. Arrowsubtest assesses line orientation judgment, which involves similar

rain structures. For example, judgment of line orientation tasksave been found to be impaired in Parkinson’s disease patients,

ndicating a role of the basal ganglia [37]. Additionally, cerebel-ar patients were also observed to have poor line orientation

nd the visuospatial subtest from the psychometric testing. Data are individual dataroup prosaccade endpoint error and arrows subtest. (B) Correlation of FASD groupde endpoint error and arrows subtest. (D) Correlation of FASD group antisaccade

judgment, indicating a role of the cerebellum in this task [38].Importantly, children with FASD have been found to have bothcerebellar and basal ganglia abnormalities [39,40]. The relationshipbetween arrows performance and saccade endpoint error was onlypresent for the prosaccade task. This is because as with arrows, onlythe prosaccade task has a visual target to guide the participant’sresponse. Therefore, this relationship is specific to visually-guidedvisuospatial processing. Poor visuospatial processing in childrenwith FASD suggests that they may view materials in a disorganizedand disconnected manner that inhibits their ability to integrateinformation and attend to details in a meaningful way.

4.4. Linking working memory and visuospatial processing

Visual processing has generally been divided into dorsaland ventral streams. Both working memory and visuospatialprocessing take place in the dorsal stream. The dorsal streamis thought to give rise to three pathways; parieto-prefrontal,parieto-premotor, and parieto-medial temporal pathways [41].The parieto-prefrontal pathway mainly involves the lateral intra-parietal, middle temporal, and medial superior temporal areas,which links the occipito-parietal circuit with the caudal portionsof the prefrontal cortex [42,43]. This pathway is thought to controlspatial working memory and would therefore be involved with thememory-guided task and block recall [41]. Non-human primateresearch has implicated the prefrontal cortex in working memory[44], and structural imaging studies have found abnormalitiesin the prefrontal cortex in children with FASD [45]. Additionally,multiple studies have found that the dorsolateral prefrontal cortexis involved in both task switching and working memory [46,47].The parieto-premotor pathway has two parallel projections; oneoriginates in parietooccipital area and the other in the ventral

intraparietal area [48,49]. They both target the premotor cortexand control visually-guided saccades that are necessary to performthe prosaccade and arrows tasks [41]. Although working memoryand visuospatial processing are distinct they both have at least one

7 l Brain

osb

4

ctgswatippnatbnpe

5

ipapcpmiFcrctiiUwotpa

A

iioWdBe

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[

8 A. Paolozza et al. / Behavioura

verlapping area, the ventral intraparietal area, necessary for theuccessful completion of the tasks. This may help specify specificrain regions damaged due to prenatal alcohol exposure.

.5. Prenatal alcohol exposed group

Interestingly, the PAE group did not perform worse than theontrol group on all measures of working memory and visuospa-ial skills. However, in our previous paper, we found that this sameroup did perform poorer than controls on measures of attention,witching, and inhibition [24]. This indicates that in this populationorking memory and visuospatial skills may be CNS domains that

ppear relatively spared. This may, at least partially, explain whyhis group did not go on to receive a diagnosis. This also points to themportance of common diagnostic tools when assessing those withrenatal alcohol exposure due to the wide range of outcomes dis-layed by this group [25]. This group presents a unique opportunity,ot previously presented in the literature, to study psychometricnd eye movement measures of working memory and visuospa-ial skills in a subset of the population who do not have a diagnosisut have been exposed to alcohol prenatally. This also indicates theeed for future research to investigate if some CNS functions andatterns of brain damage are more vulnerable to prenatal alcoholxposure than others.

. Conclusions

The results of this study have important clinical and psycholog-cal implications. The psychometric tests and eye movement tasksoint to specific overlapping brain regions damaged by prenatallcohol exposure. Due to the extensive research and brain map-ing from previous research in eye movement control these tasksan be leveraged to better understand what is being assessed bysychometric testing. Additionally, it also indicates that eye move-ent tasks may be a viable option as a screening or adjunct tool

n the assessment of FASD. The process to receive a diagnosis of anASD is long, can be stressful on the child and family, requires spe-ialized training, has high costs, and may not be accessible to someural populations. Therefore, there is potential for eye movementontrol to be added as a first screening step in this process as thehree tasks presented in the study take less than one hour to admin-ster and can simultaneously measure multiple cognitive domainsncluding those presented here, as well as response inhibition [24].sing eye movement control tasks as a screening tool can also helpith early identification of those in need of a follow-up assessment

r those in need of early intervention while waiting for a diagnos-ic assessment. This could then lead to earlier interventions andotentially less adverse outcomes for those prenatally exposed tolcohol [50].

cknowledgements

We thank the participants and their families for taking partn the study. This work was supported by NeuroDevNet, whichs funded by the Networks of Centres of Excellence, a programf the federal government to advance science and technology.e also acknowledge the NeuroDevNet Neuroinformatics Core for

ata management system implementation and support and Donaldrein for his technical expertise in the collection and analysis of theye movement data.

eferences

[1] Chudley AE, Conry J, Cook JL, Loock C, Rosales T, LeBlanc N. Fetal alco-hol spectrum disorder: Canadian guidelines for diagnosis. CMAJ 2005;172(5Suppl):S1–21.

[

[

Research 263 (2014) 70–79

[2] May PA, Gossage JP, Kalberg WO, Robinson LK, Buckley D, Manning M, et al.Prevalence and epidemiologic characteristics of FASD from various researchmethods with an emphasis on recent in-school studies. Dev Disabil Res Rev2009;15(3):176–92.

[3] Kodituwakku PW. Neurocognitive profile in children with fetal alcohol spec-trum disorders. Dev Disabil Res Rev 2009;15(3):218–24.

[4] Mattson SN, Crocker N, Nguyen TT. Fetal alcohol spectrum disorders:neuropsychological and behavioral features. Neuropsychol Rev 2011;21(2):81–101.

[5] Rasmussen C. Executive functioning and working memory in fetal alcohol spec-trum disorder. Alcohol Clin Exp Res 2005;29(8):1359–67.

[6] Rasmussen C, Andrew G, Zwaigenbaum L, Tough S. Neurobehavioural outcomesof children with fetal alcohol spectrum disorders: A Canadian perspective. Pae-diatr Child Health 2008;13(3):185–91.

[7] Alloway TP, Gathercole SE, Kirkwood H, Elliott J. The cognitive and behav-ioral characteristics of children with low working memory. Child Dev2009;80(2):606–21.

[8] Munoz DP, Everling S. Look away: the anti-saccade task and the voluntarycontrol of eye movement. Nat Rev Neurosci 2004;5(3):218–28.

[9] Scudder CA, Kaneko CS, Fuchs AF. The brainstem burst generator for sac-cadic eye movements: a modern synthesis. Exp Brain Res 2002;142(4):439–62.

10] Ramat S, Leigh RJ, Zee DS, Optican LM. What clinical disorders tell usabout the neural control of saccadic eye movements. Brain 2007;130(Pt 1):10–35.

11] Green CR, Mihic AM, Brien DC, Armstrong IT, Nikkel SM, Stade BC, et al. Oculo-motor control in children with fetal alcohol spectrum disorders assessed usinga mobile eye-tracking laboratory. Eur J Neurosci 2009;29(6):1302–9.

12] Green CR, Munoz DP, Nikkel SM, Reynolds JN. Deficits in eye movement con-trol in children with fetal alcohol spectrum disorders. Alcohol Clin Exp Res2007;31(3):500–11.

13] Paolozza A, Titman R, Brien DC, Munoz DP, Reynolds JN. Altered Accuracy ofSaccadic Eye Movements in Children with Fetal Alcohol Spectrum Disorder.Alcohol Exp Res 2013;37(9):1491–8, http://dx.doi.org/10.1111/acer.12119.

14] Baddeley A. The episodic buffer: a new component of working memory? TrendsCogn Sci 2000;4(11):417–23.

15] Streissguth AP, Barr HM, Sampson PD. Moderate prenatal alcohol exposure:effects on child IQ and learning problems at age 7 1/2 years. Alcohol Clin ExpRes 1990;14(5):662–9.

16] Jacobson JL, Jacobson SW, Sokol RJ, Ager Jr JW. Relation of maternal age andpattern of pregnancy drinking to functionally significant cognitive deficit ininfancy. Alcohol Clin Exp Res 1998;22(2):345–51.

17] Uecker A, Nadel L. Spatial locations gone awry: object and spatial mem-ory deficits in children with fetal alcohol syndrome. Neuropsychologia1996;34(3):209–23.

18] Rasmussen C, Bisanz J. The relation between mathematics and working mem-ory in young children with Fetal Alcohol Spectrum Disorders. J Spec Educ2011;45:184–91.

19] Carroll JB. Human Cognitive Abilities. Cambridge: Cambridge University Press;1993.

20] Quattlebaum JL, O’Connor MJ. Higher functioning children with prenatal alco-hol exposure: Is there a specific neurocognitive profile? Child Neuropsychol2012.

21] Rasmussen C, Tamana S, Baugh L, Andrew G, Tough S, Zwaigenbaum L. Neu-ropsychological impairments on the NEPSY-II among children with FASD. ChildNeuropsychol 2012.

22] Korkman M, Kettunen S, Autti-Ramo I. Neurocognitive impairment in earlyadolescence following prenatal alcohol exposure of varying duration. ChildNeuropsychol 2003;9(2):117–28.

23] Pei J, Job J, Kully-Martens K, Rasmussen C. Executive function and mem-ory in children with Fetal Alcohol Spectrum Disorder. Child Neuropsychol2011;17(3):290–309.

24] Paolozza A, Rasmussen C, Pei J, Hanlon-Dearman A, Nikkel SM, Andrew G, et al.Deficits in response inhibition correlate with oculomotor control in childrenwith fetal alcohol spectrum disorder and prenatal alcohol exposure. BehavBrain Res 2013;259C::97–105.

25] Chudley AE, Conry J, Cook JL, Loock C, Rosales T, LeBlanc N. Fetal alco-hol spectrum disorder: Canadian guidelines for diagnosis. CMAJ 2005;172(5Suppl):S1–21.

26] Reynolds JN, Weinberg J, Clarren S, Beaulieu C, Rasmussen C, Kobor M, et al.Fetal alcohol spectrum disorders: gene-environment interactions, predictivebiomarkers, and the relationship between structural alterations in the brainand functional outcomes. Semin Pediatr Neurol 2011;18(1):49–55.

27] Hollingshead AA. Four-factor index of social status. 1975.28] Korkman M, Kirk U, Kemp S. NEPSY-II. Clinical and interpretation manual. 2007.

San Antonio, TX, Harcourt Assessment.29] Gathercole SE, Pickering SJ. Working Memory Test Battery for Children. London,

England: Psychological Corporation Europe; 2001.30] Dennis M, Francis DJ, Cirino PT, Schachar R, Barnes MA, Fletcher JM. Why IQ

is not a covariate in cognitive studies of neurodevelopmental disorders. J IntNeuropsychol Soc 2009;15(3):331–43.

31] Stigler JW, Miller KF. A good match is hard to find: Comment on Mayer, Tajika,and Stanley (1991). J Edu Psychol 1993;85:554–9.

32] Olson HC, Feldman JJ, Streissguth AP, Sampson PD, Bookstein FL. Neuropsycho-logical deficits in adolescents with fetal alcohol syndrome: clinical findings.Alcohol Clin Exp Res 1998;22(9):1998–2012.

l Brain

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[connections of the inferior parietal cortical convexity of the macaque monkey.

A. Paolozza et al. / Behavioura

33] Quattlebaum JL, O’Connor MJ. Higher functioning children with prenatal alco-hol exposure: Is there a specific neurocognitive profile? Child Neuropsychol2012.

34] Chen-Harris H, Joiner WM, Ethier V, Zee DS, Shadmehr R. Adaptive control ofsaccades via internal feedback. J Neurosci 2008;28(11):2804–13.

35] Glickstein M, Doron K. Cerebellum: connections and functions. Cerebellum2008;7(4):589–94.

36] Soetedjo R, Fuchs AF, Kojima Y. Subthreshold activation of the superior collicu-lus drives saccade motor learning. J Neurosci 2009;29(48):15213–22.

37] Muslimovic D, Post B, Speelman JD, Schmand B. Motor procedural learning inParkinson’s disease. Brain 2007;130(Pt 11):2887–97.

38] Lee TM, Liu HL, Hung KN, Pu J, Ng YB, Mak AK, et al. The cerebellum’s involve-ment in the judgment of spatial orientation: a functional magnetic resonanceimaging study. Neuropsychologia 2005;43(13):1870–7.

39] Mattson SN, Riley EP, Sowell ER, Jernigan TL, Sobel DF, Jones KL. A decrease inthe size of the basal ganglia in children with fetal alcohol syndrome. AlcoholClin Exp Res 1996;20(6):1088–93.

40] O’Hare ED, Kan E, Yoshii J, Mattson SN, Riley EP, Thompson PM, et al. Map-ping cerebellar vermal morphology and cognitive correlates in prenatal alcoholexposure. Neuroreport 2005;16(12):1285–90.

41] Kravitz DJ, Saleem KS, Baker CI, Mishkin M. A new neural framework for visu-ospatial processing. Nat Rev Neurosci 2011;12(4):217–30.

42] Schall JD, Morel A, King DJ, Bullier J. Topography of visual cortex connectionswith frontal eye field in macaque: convergence and segregation of processingstreams. J Neurosci 1995;15(6):4464–87.

[

Research 263 (2014) 70–79 79

43] Cavada C, Goldman-Rakic PS. Posterior parietal cortex in rhesus monkey: II.Evidence for segregated corticocortical networks linking sensory and limbicareas with the frontal lobe. J Comp Neurol 1989;287(4):422–45.

44] Goldman-Rakic PS. Architecture of the prefrontal cortex and the central exec-utive. Ann N Y Acad Sci 1995;769:71–83.

45] Sowell ER, Mattson SN, Kan E, Thompson PM, Riley EP, Toga AW.Abnormal cortical thickness and brain-behavior correlation patterns in indi-viduals with heavy prenatal alcohol exposure. Cereb Cortex 2008;18(1):136–44.

46] Moore TL, Schettler SP, Killiany RJ, Rosene DL, Moss MB. Effects on execu-tive function following damage to the prefrontal cortex in the rhesus monkey(Macaca mulatta). Behav Neurosci 2009;123(2):231–41.

47] Goldman PS, Rosvold HE. Localization of function within the dorsolat-eral prefrontal cortex of the rhesus monkey. Exp Neurol 1970;27(2):291–304.

48] Gamberini M, Passarelli L, Fattori P, Zucchelli M, Bakola S, Luppino G, et al. Cor-tical connections of the visuomotor parietooccipital area V6Ad of the macaquemonkey. J Comp Neurol 2009;513(6):622–42.

49] Rozzi S, Calzavara R, Belmalih A, Borra E, Gregoriou GG, Matelli M, et al. Cortical

Cereb Cortex 2006;16(10):1389–417.50] Streissguth AP, Bookstein FL, Barr HM, Sampson PD, O’Malley K, Young JK. Risk

factors for adverse life outcomes in fetal alcohol syndrome and fetal alcoholeffects. J Dev Behav Pediatr 2004;25(4):228–38.