the development of sustained attention in children: the effect of age and task load

19
Child Neuropsychology, 12: 205–221, 2006 Copyright © Taylor & Francis Group, LLC ISSN: 0929-7049 print / 1744-4136 online DOI: 10.1080/09297040500488522 205 THE DEVELOPMENT OF SUSTAINED ATTENTION IN CHILDREN: THE EFFECT OF AGE AND TASK LOAD Jennifer Betts, 1 Jenny Mckay, 1 Paul Maruff, 4 and Vicki Anderson 1,2,3 1 University of Melbourne, Melbourne, Australia, 2 Royal Children’s Hospital, Melbourne, Melbourne, Australia, 3 Murdoch Children’s Research Institute, Melbourne, Australia, and 4 CogState, Pty. Ltd, Melbourne, Australia This study explored how children’s sustained attention develops and the effect of manipulat- ing task parameters on sustained attention. The sample comprised 57 children (5–12 years) who completed CogState and Score! (Test of Everyday Attention for Children). Novel vari- ability and traditional indices indicated rapid development from 5–6 to 8–9 years on all measures and a developmental plateau from 8–9 to 11–12, with growth evident on some measures. Findings suggest that sustained attention improves to age 10, then plateaus with only minor improvements. Further, performance was generally poorer on high load tasks compared to low load, with the same developmental pattern uncovered. Keywords: sustained attention child development variability Sustained attention, or vigilance, refers to the ability to maintain attention over an extended period of time. Brain-behavior models propose that this component of attention is mediated to a large extent by the reticular formation and brain stem structures (Mirsky, Anthony, Duncan, Ahearn, & Kellam, 1991; Mirsky, 1996; van Zomeren & Brouwer, 1994), with some involvement of frontal regions (Stuss, Shallice, Alexander & Picton, 1995). The capacity to sustain attention plays a key role in children’s school performance, determining the child’s capacity to maintain concentration over long periods in order to understand and integrate large amounts of information (Catroppa & Anderson, 1999). Impairments in sustained attention, therefore, may potentially impact on the child’s ability to acquire and integrate new skills and knowledge. Despite this, only limited research has been conducted into the normal development of sustained attention. While there is a large body of knowledge concerning theoretical models of sustained attention and its underlying neural basis within the adult literature, the child literature is less developed. One of the earliest studies to address these issues was conducted by McKay, Halperin, Schwartz, and Sharman (1994). These researchers employed a cross-sectional design to describe developmental trajectories for healthy children aged 7 to 11 years, and they compared these to an adult sample, using a continuous performance paradigm (CPT) Address correspondence to Vicki Anderson Ph.D., Professor/Director, Psychology, Royal Children’s Hospital, Parkville Victoria 3052, Australia. Tel: +61 3 9345 5524. Fax: +61 3 9345 6002. E-mail: vaa@ unimelb.edu.au

Upload: independent

Post on 30-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Child Neuropsychology, 12: 205–221, 2006Copyright © Taylor & Francis Group, LLCISSN: 0929-7049 print / 1744-4136 onlineDOI: 10.1080/09297040500488522

205

THE DEVELOPMENT OF SUSTAINED ATTENTION IN CHILDREN: THE EFFECT OF AGE AND TASK LOAD

Jennifer Betts,1 Jenny Mckay,1 Paul Maruff,4 and Vicki Anderson1,2,3

1University of Melbourne, Melbourne, Australia, 2Royal Children’s Hospital,Melbourne, Melbourne, Australia, 3Murdoch Children’s Research Institute,Melbourne, Australia, and 4CogState, Pty. Ltd, Melbourne, Australia

This study explored how children’s sustained attention develops and the effect of manipulat-ing task parameters on sustained attention. The sample comprised 57 children (5–12 years)who completed CogState and Score! (Test of Everyday Attention for Children). Novel vari-ability and traditional indices indicated rapid development from 5–6 to 8–9 years on allmeasures and a developmental plateau from 8–9 to 11–12, with growth evident on somemeasures. Findings suggest that sustained attention improves to age 10, then plateaus withonly minor improvements. Further, performance was generally poorer on high load taskscompared to low load, with the same developmental pattern uncovered.

Keywords: sustained attention child development variability

Sustained attention, or vigilance, refers to the ability to maintain attention over anextended period of time. Brain-behavior models propose that this component of attentionis mediated to a large extent by the reticular formation and brain stem structures (Mirsky,Anthony, Duncan, Ahearn, & Kellam, 1991; Mirsky, 1996; van Zomeren & Brouwer,1994), with some involvement of frontal regions (Stuss, Shallice, Alexander & Picton,1995). The capacity to sustain attention plays a key role in children’s school performance,determining the child’s capacity to maintain concentration over long periods in order tounderstand and integrate large amounts of information (Catroppa & Anderson, 1999).Impairments in sustained attention, therefore, may potentially impact on the child’s abilityto acquire and integrate new skills and knowledge. Despite this, only limited research hasbeen conducted into the normal development of sustained attention.

While there is a large body of knowledge concerning theoretical models of sustainedattention and its underlying neural basis within the adult literature, the child literature is lessdeveloped. One of the earliest studies to address these issues was conducted by McKay,Halperin, Schwartz, and Sharman (1994). These researchers employed a cross-sectionaldesign to describe developmental trajectories for healthy children aged 7 to 11 years, andthey compared these to an adult sample, using a continuous performance paradigm (CPT)

Address correspondence to Vicki Anderson Ph.D., Professor/Director, Psychology, Royal Children’sHospital, Parkville Victoria 3052, Australia. Tel: +61 3 9345 5524. Fax: +61 3 9345 6002. E-mail: [email protected]

206 J. BETTS ET AL.

to measure sustained attention. As might be expected, CPT performances, or the capacityto sustain attention, deteriorated over time, regardless of age, suggesting lower levels ofefficiency with time on task. Results indicated stable but less efficient performances forchildren aged between 7 and 11 years in comparison to adults, with a significant improve-ment in children’s performances around 11 years of age, suggesting relatively little devel-opmental progress in sustained attention skills through middle childhood.

While McKay et al.’s (1994) research made an important contribution to the field,its methodology has been questioned. Using a rigorous longitudinal design, Rebok et al.(1997) reported on the development of attention skills in 435 children assessed at ages 8,10, and 13 years. A CPT paradigm was also employed in this study, with outcome mea-sures of reaction time (RT), accuracy (correct responses and correct omissions), and omis-sion errors. Rebok and colleagues detected significant age effects on all measures.Reaction times improved between 8 to 10 years and again from 10 to 13 years. Accuracydramatically improved from 8 to 10 years then showed gradual improvement from 10 to13 years. Errors declined by about half from 8 to 10 years with more gradual declines from10 to 13 years. Contrary to McKay et al. (1994), Rebok et al. (1997) concluded that sus-tained attention develops rapidly from 8 to 10 years then plateaus from 10 to 13 years,with only gradual improvements during that period. More recent cross-sectional researchhas supported Rebok et al.’s (1997) findings, suggesting that, regardless of the outcomemeasure employed, there is continued improvement in sustained attention skills from 8 to16 years, but with the magnitude of gains reducing from around 10 to 11 years (Klenberg,Korkman, & Lahti-Nuuttila, 2001; Manly et al., 2001). While these studies provide a bet-ter understanding of attentional development for children aged eight years and older, thereis ongoing debate regarding the relative sustained attention capacities of younger children,particularly those in their first years of schooling.

Sustained Attention and Load

A wealth of literature exists concerning the impact of variations in task parameters onsustained attention performance in adults. In contrast, very few studies have addressed thepresence of such effects on children’s sustained attention performance. In adults, instruc-tions to participants, feedback, task duration and complexity, stimulus presentation times,and interstimulus intervals have all been found to influence performance (Ballard, 1996b;van der Meere & Sergeant, 1988). Manipulating task parameters varies the processingdemands and level of sustained attention necessary to complete the task, and performancealters accordingly (Corkum & Siegel, 1993). When the requirements are greater, the task isconsidered one of high load and when requirements are simpler, the task is one of low load.

While similar patterns may emerge in children, given the rapid cognitive and neurologi-cal development occurring throughout childhood, it is inappropriate to simply extrapolate fromadult findings. For example, adult studies have traditionally found that performance ispoorer when event rate and signal probability are high (eg. Parasuraman, 1979; Warm,Howe, Fishbein, Dember, & Sprague, 1984), however, in a study of 7- and 8-year-olds,Laurie-Rose, Bennett-Murphy, Schickedantz, and Tucci (2001) found that children weresignificantly faster and more accurate when event rate and signal probability were high. Theyhypothesized that a low event rate might be “under-arousing” for children, placing greaterdemands on children’s sustained attention resources and resulting in poorer performance.

Using adult samples, researchers have consistently identified performancedifferences associated with task load or complexity. Noonan, Ash, Loeb, and Warm

DEVELOPMENT OF SUSTAINED ATTENTION 207

(1984) varied target criteria in four conditions of a CPT task that presented pairs of num-bers on a screen every three seconds for 1.5 seconds duration. Participants were asked torespond by pressing a button when target criterion was satisfied. In Condition 1, target cri-teria comprised a pair of numbers that differed by no more than one. Condition 2 addedthe requirement that the summed number pair fall within a specified range (eg. 4–14)placed above the pair. The complexity increased further in Condition 3, with two rangesand two number pairs presented, and continued to increase in Condition 4, where fournumber pairs and ranges were presented. In Conditions 2–4, the ranges changed at 30, 60,or 90 second intervals (averaging a change every 60 seconds). Sustained attention perfor-mance was measured via RT and correct responses at ten-minute intervals over the 50-minute task duration. As expected, correct responses decreased and RT increased as thetarget criteria became more complex, supporting the presence of “task load” effect.

Similar results were reported by Lysaght, Warm, Dember, and Loeb (1984) whoused just two conditions (Condition 1 and Condition 2 from the aforementioned study),which they nominated as the “low” and “high” load tasks respectively. Sustained attentionwas reflected in the percentage of correct responses, which was assessed at 20-minuteintervals over a 60-minute period. Again, participants were less accurate on the task withcomplex target criteria than the task with simple target criteria.

As far as we are aware, only a handful of studies have been conducted with childrento determine whether similar task complexity effects exist. These have generally beenrestricted to clinical populations, such as traumatic brain injury or Attention deficit/hyper-activity disorder (Catroppa & Anderson, 2003; van der Meere & Sergeant, 1988).

The Measurement of Sustained Attention

One of the challenges within the field of attention is the identification of appropriateand reliable assessment tools. A wide range of tasks and performance indices has beenemployed in the literature, leading to difficulties in comparisons across studies and proto-cols. Traditionally, sustained attention has been operationalized via tasks of long duration,such as signal detection tasks, that tap the continuity of performance. Performances onsuch tasks are usually characterized by a gradual fall-off with time-on-task. Signal detec-tion tasks have been used extensively in adult research but have also been employed inchildhood studies. In the child and adult literature, the length of tasks have varied greatlyfrom a few minutes to over an hour, with tasks of 10 to 15 minutes typically employed indevelopmental research. More recently, however, within the developmental literature, anumber of studies have demonstrated sustained attention deficits using much shorter pro-tocols (e.g. Catroppa & Anderson, 2003; Manly et al., 2001). The current research utilizesa computer-based visual continuous performance task paradigm of 3–4 minutes duration,dispersed amongst a battery of attention and executive function tasks. The duration of thetotal test protocol is approximately 20 minutes. In addition, a 6-minute auditory-basedcontinuous performance task was administered. Historically, a characteristic of tasks pur-porting to measure sustained attention has been the inclusion of monotonous stimuli, withminimal demands on participants, (e.g. signal detection), so as to tap and to exhaust inter-nal resources of attention and concentration. The tasks utilized in the current study areconsistent with these principles.

Many researchers endorse the use of multiple tests to improve the validity of conclusions(Fletcher, 1998; Shapiro, Morris, Morris, Flowers, & Jones, 1998). An extension of this princi-ple is the recording of multiple performance indices on multiple tests. Conventionally,

208 J. BETTS ET AL.

performance indices on tasks of sustained attention comprise mean RT, accuracy, anderrors (both inaccurate responses and omissions). It has been proposed that measuring per-formance variability, in addition to traditional indices, provides a more complete pictureof sustained attention (van Zomeren & Brouwer, 1994). Although variability has not beenapplied in this context before, it has advantages over some of the traditional measures.Psychological research primarily deals with latent phenomena, however, it is preferable toobserve and measure phenomena directly wherever possible (Castellanos & Tannock,2002). That is, it is better to infer sustained attention processes from variables such asmean RT or variability in RT, rather than from a missed response, as is the case in errorsof omission. Further, measures such as errors or accuracy often produce data that have askewed distribution and are subject to floor or ceiling effects. This is especially commonin normative samples and for traditional neuropsychological tests that typically have asmall number of trials (Ballard, 1996a). These types of data are more appropriately treatedas ordinal, or nonparametric, as the difference between an individual with two errors com-pared to four errors suggests that, while the individual with fewer errors is better, he/she isnot necessarily twice as good. The advent of computers has facilitated the use of RTs andvariability in RTs, which do not suffer the measurement problems associated with floor orceiling effects. They are normally distributed (or become normal with transformation) andare a true interval scale of measurement. In addition to these psychometric advantages,variability measurement has provided insight into the process of cognitive change (Miller,2002) and has been advocated in the fields of head trauma and focal brain lesions(Bleiberg, Garmoe, Halpern, Reeves, & Nadler, 1997; Stuss et al., 1989) and has also beenemployed to study ADHD (Hynd, Nieves, Connor, & Stone, 1989).

There are three main sources of variability (see Figure 1). First, when an individualresponds to an item in a task, the RT to that item is recorded. Over an entire task, an indi-vidual’s mean RT can be calculated. The fluctuations of RTs around the mean reflect the

Figure 1 A visual display of the sources of variability.

DEVELOPMENT OF SUSTAINED ATTENTION 209

participant’s intraindividual variability (Makdissi et al., 2001). Second, when a participantcompletes the same task several times, their performance on each task can be compared. Achange in a person’s intraindividual variability from one task to the next reflects their con-sistency (Hultsh, MacDonald, & Dixon, 2002). Third, when different groups of individu-als complete tasks, the average intraindividual variability and average consistency can becompared between the groups.

The use of variability as a tool for understanding sustained attention is in its infancy,with no established research tradition advocating specific statistical procedures. Two pos-sible techniques that may be used to measure variability include an established measure—standard deviation (SD), and a newer technique, the mean absolute deviation from themedian (MAD). The MAD involves identifying the median RT and computing the differ-ence between the RT of each item and the median. The mean of these difference scores isthe MAD. The MAD is argued to be a robust measure of variability, but unlike SD, outliersdo not unduly affect it (Garret & Nash, 2001). Both MAD and SD will be employed in thecurrent study.

The Current Study

The present study aimed to extend current understanding of the development of sus-tained attention skills in children aged 5–12 years, employing both traditional and novelperformance indices. With regard to the novel index, it was predicted that variability inperformance would be greater for younger participants. In line with previous research itwas expected that, regardless of age, participants’ performance over time would evidencegreater variability as sustained attention resources become depleted, with traditional indi-ces showing evidence of development of sustained attention throughout the 5–12-yearperiod, but with greatest progress occurring before age 10 years. It was predicted that asimilar developmental pattern would be demonstrated for the novel variability index.

The second aim of the study was to explore the effect of task load on children’s sus-tained attention performance using both novel and traditional indices. The task parametersof target complexity and display size were manipulated to create a task of low load and atask of high load. Although these parameters have not previously been explored with childsamples, in line with the adult literature, it was predicted that traditional indices woulddemonstrate poorer performance on the high load task compared to those for the low loadtask. It was expected that as the high load task places greater sustained attention demandson participants, their performance on the high load task would comprise greater variabilitythan on the low load task.

METHOD

Participants

The sample comprised 57 participants (28 males and 29 females) aged between 5 and 12years, recruited from public and private primary schools in metropolitan Melbourne,Australia. Availability sampling techniques were employed whereby all children whoreturned consent forms and met the inclusion criteria were tested. Inclusion criteria were:attending mainstream school with no additional educational support; no history of diag-nosed ADHD (indicated by a score less than two standard deviations from the age-meanon the Rowe Behavioural Rating Inventory: RBRI: Rowe & Rowe, 1995, a parent-based

210 J. BETTS ET AL.

measure of attentional function); and no diagnosed neurological, developmental or psychi-atric disorder, based on parental responses provided via an information questionnaire.

Participants were divided into three age groups: 5–6 years (n = 19), 8–9 years (n = 19),11–12 years (n = 19). Group characteristics are presented in Table 1. The socio-economicstatus (SES) of each child’s family was determined using Daniel’s Scale of OccupationalPrestige (Daniel, 1983), which rates parental occupation from 1 to 7, with 1 reflecting highSES and 7 representing low. There was no significant difference in SES between thegroups (F (2,54) = 0.163, p = .85).

Materials

Demographic and Background Information. Parents were required to com-plete a questionnaire eliciting demographic details, children’s medical and educationalhistory, and parental occupation. They also completed the Rowe Behavioural RatingInventory (RBRI: Rowe & Rowe, 1995), a 16-item scale measuring children’s everydaysymptoms of inattentiveness, normed on an Australian sample. Parents responded tobehavioral statements, such as “restless; fidgety; can’t sit still” versus “relaxed; can sitstill,” by marking a five-point ordinal scale. A total score was calculated by summingresults on the attentive-inattentive and restless-settled dimensions, as described in the testmanual.

(i) CogState, Version 2.1.0 (CogState Ltd, 2002). Participants completedCogState, a computer-administered battery of nine neuropsychological subtests designedto tap aspects of attention and information processing, and reported to be sensitive to sub-tle changes in performance. Although completing the entire battery, participant’s perfor-mances on five subtests were of interest in the current study (refer to Table 2).

Table 1 Demographic characteristics of sample

Age (months) GenderSocioeconomic

Status

Group Mean (SD) Range Males Females Mean (SD) Range

5–6 years 65.95 (4.52) 62–77 10 9 3.38 (0.98) 1.8–6.18–9 years 106.05 (3.34) 101–13 10 9 3.23 (0.68) 1.8–411–12 years 141.84 (4.21) 134–48 8 11 3.24 (1.11) 1.8–5.3

Table 2 CogState: Task composition, order of presentation, duataion of task, and variable labels

Subtest Order TaskApproximate

DurationAnalysed in

Current StudyVariable

Label

1 Simple Reaction Time task 2 minutes ✓ SRT12 Choice Reaction Time task 2 minutes3 Congruent Reaction Time task 2 minutes ✓ Low Load4 Dynamic Monitoring task 2 minutes5 Working Memory task 2 minutes6 Simple Reaction Time task 2 minutes ✓ SRT27 Matching Task 3 minutes ✓ High Load8 Memory task 4 minutes9 Simple Reaction Time task 2 minutes ✓ SRT3

DEVELOPMENT OF SUSTAINED ATTENTION 211

Attention Over Time. To capture sustained attentional processes, participantscompleted three identical Simple Reaction Time tasks presented as the first, sixth, andninth tasks in the battery. Participants were asked to monitor the presence of a playingcard and to press a response key on the keyboard as soon as the card turned face up. Theinterstimulus interval was varied randomly between 1000 and 1500 msec, with each stim-ulus present for a maximum of five seconds.

These tasks required participants to call on internal resources to maintain concentra-tion on the monotonous stimuli. The repeated administration of the task at three intervalsduring the test battery captured any change in participant’s performance over time, anessential component of any measure of sustained attention. For the purposes of the presentstudy, the three Simple Reaction Time tasks will be referred to as SRT1, SRT2, and SRT3,respectively. Performance was measured by speed (mean RTs), errors (number ofresponses occurring before a card’s face was displayed), accuracy (number of correctresponses), variability of RT (SD and MAD), and max outs (failure to respond to an itemwithin five seconds).

(ii) Score! (Manly et al., 1999). Participants completed Score!, a subtest fromthe Test of Everyday Attention for Children. Participants were asked to listen to a taperecording of 10 “games” and silently count the scoring sounds (without using their fin-gers). Sounds were presented at varied intervals, with between 9 and 15 per game. At theend of each game the participant’s response represented the number of scoring soundsthey counted. Taking approximately six minutes to complete, Score! reflects the subjects’ability to maintain alertness over time in order to monitor and count the sounds. Numberof correct responses was the outcome measure used in analysis.

Attention Under Differing Loads. This study also examined participants’ per-formance on two other subtests from the CogState battery, the Congruent Reaction TimeTask, and the Matching Task, administered as the third and seventh tasks. Each of thesetasks employs the same basic task characteristics (i.e. computer presentation, playing cardstimulus) as the simple reaction time tasks described above. The outcome measures onboth tasks included speed (mean log10 RT), errors (number of responses that reportedmatching cards when they were incongruent, or nonmatching cards when they were con-gruent), variability (MAD log10 RT and SD log10 RT), and max outs. The CongruentReaction Time Task asks participants to judge whether two playing cards are the samecolor, thus requiring them to scan two cards and focus on color as the important feature.With its simple target criterion (color) and a small display size, the Congruent ReactionTime Task is the designated “low load” task and will herein be referred to as such. TheMatching Task asks participants to determine if playing cards match, but it broadens thetarget criteria to include suit and number as well as color (refer to Figure 2). Further,instead of examining individual cards, the target is a pair of cards and six possible matchesare presented. Hence, the Matching Task has been designated the “high load” task. Again,the interstimulus interval was randomly varied from between 1000 and 1500 msec, with amaximum stimulus duration of five seconds.

Both of these tasks capture participants’ sustained attention capabilities by requiringthem to maintain concentration and monitor the playing card stimuli over time, makingdecisions regarding the match between playing cards. The low and high load tasks there-fore involve participants engaging in analogous mental processes, however the high loadtask has greater sustained attention demands due to the larger display size and the com-plexity of target criteria.

212 J. BETTS ET AL.

Procedure

Ethics approval was granted by the University of Melbourne’s Human Ethics Com-mittee. The cooperation of primary schools was sought via letter. Upon gaining permis-sion to conduct research within a school, information packages were forwarded to allparents, inviting participation. Written consent was obtained from both parents and chil-dren. Parents completed the RBRI and an information questionnaire.

Children were tested in a quiet room at school, on an individual basis. Each childwas seen for one testing session, taking approximately one hour. Tasks were administeredwithout breaks, in a set order to control for any learning effects common in neuropsycho-logical research.

CogState was administered on colored IBM laptops. The researcher gave a briefdescription of CogState and prior to each task, specific written instructions appeared onthe screen. These were read aloud and elaborated by the researcher if necessary. Partici-pants had to demonstrate that they comprehended the task and that their perceptual andmotor skills were adequate by completing practice trials. The assessment began once threesuccessive practice trials were successfully completed.

Score! was administered using a portable cassette player. Verbal instructions weregiven according to the administration manual. Before beginning the test, participants wereasked to demonstrate that they could count to fifteen and were required to respond cor-rectly (by accurately counting the stimuli sounds) to the two practice trials. If practice tri-als were failed, instructions were explained further and the practice trials repeated untilcomprehension was demonstrated.

Figure 2 The Matching Task (the keyboard and instructions disappear after the practice trials).

DEVELOPMENT OF SUSTAINED ATTENTION 213

Statistical Analysis

The first independent variable was age, with three groups comprising 5–6-year-olds, 8–9-year-olds, and 11–12-year-olds. The dependent variable was sustainedattention as measured by speed (mean RT), errors (total number incorrect), variabil-ity (SD and MAD), and accuracy (total number correct). The second independentvariable was load (high or low), with the dependent variable, sustained attention,measured by speed (mean RT), errors (percent incorrect), max outs, and variability(SD and MAD).

All statistical analyses were performed with Statistical Packages for SocialSciences (SPSS) Version 9.05. Exploratory data analysis was conducted, with alphalevel set at .01. Boxplots and matrix scatterplots illustrated score distributions. As theRTs were skewed, a log10 transformation was performed on each participant’s RTbefore calculating descriptive statistics. These transformations helped to reduce theinfluence of outliers, which may have affected variability statistics (Luce, 1986).Box’s M test was employed to identify problems with homogeneity of the variance-covariance matrices, influential outliers were uncovered by Cook’s distance, andsphericity was examined using Mauchly’s test.

When test assumptions were satisfied, repeated measures multivariate ANOVAswere performed on i) age vs. time and ii) age vs. load, with an alpha level set at .05 forthese inferential tests. When a significant age vs. time interaction occurred, planned t-testswere undertaken to compare adjacent levels of the time variable for each age, with a Bon-ferroni adjusted alpha level being employed to control Type 1 error. Variables that wereskewed to an extent that transformation was inappropriate were analyzed using nonparametricstatistics (Kruskal Wallace, Mann-Whitney U and Chi-square). There were some missingdata due to computer malfunction. Only participants with complete data for the hypothesisunder investigation were included in the analysis, e.g., only participants with completedata for the Simple Reaction Time tasks (SRT1, SRT2, SRT3) were included when exam-ining performance change over time.

RESULTS

Development of Sustained Attention

Speed. A significant interaction was identified between group (5–6, 8–9, and 11–12-year-olds) and performance on the repeated administration of the Simple ReactionTime task (SRT1, SRT2, SRT3), F (4,92) = 2.74, p = .03 (see Figure 3).

Planned comparisons revealed that the responses of the 5–6-year olds became sig-nificantly slower from SRT1 to SRT2, t (12) = −5.05, p < .001, then plateaued, with nosignificant change from SRT2 to SRT3, t (12) = .73, p = 0.48. The 8–9 and 11–12-year-olds’ performance was unaffected by time, with no significant change between SRT1 andSRT2, t (18) = −2.40, p = .03, and t (17) = .40, p = .69 respectively, or SRT2 and SRT3, t(18) = .77, p = .45, and t (17) = .73, p = .48, respectively.

Errors. There were no systematic differences in the number of errors occurringover time, x2 (2, N = 50) = 2.11, p = .35. The observed slowing of RTs over the duration ofthe tasks could therefore not be explained in terms of a speed-accuracy trade-off, as errorrates remained constant. There were significant differences in median errors committed byeach group, x2 (2, N = 150) = 16.2, p < .001. The 5–6-year-olds (Mdn = 2, Range 0–27)committed more errors than 8–9-year-olds (Mdn = 1, Range = 1–8), U = 757.50, p = .01,

214 J. BETTS ET AL.

and 8–9-year-olds committed more errors than 11–12-year-olds (Mdn = 0, Range = 0–6),U = 1090.00, p = .01. The 5–6-year-olds were therefore slower and less accurate than 8–9-year-olds, who in turn were slower and less accurate than 11–12-year-olds.

Variability—SD. There was no significant change in consistency over time, F(2,46) = .36, p = .70. However, a significant difference across age groups was detected, F(2,47) = 9.82, p < .001. The 8–9-year-olds were not significantly less variable than 5–6-year-olds, t (30) = 1.52, p = .14. The 8–9-year olds were significantly more variable than the 11–12-year-old participants, t (35) = 3.25, p = .003. Refer to Table 3 for descriptive data.

Variability—MAD. A significant difference in consistency, F (2,46) = 6.73, p <.001 and a significant group difference, F (2,47) = 4.90, p = .01 were identified. Overall,

Figure 3 The mean log10 reaction time across age and time.

0

0.5

1

1.5

2

2.5

3

3.5

SRT1 SRT2 SRT3

Time

Mea

n L

og

10 R

eact

ion

Tim

e(m

illis

eco

nd

s)

5-6 years

8-9 years

11-12 years

Table 3 Variability across age groups for the three simple reaction time tasks (SRT1, SRT2, SRT3)

SRT1 SRT2 SRT3

SD MAD SD MAD SD MAD

Age M SD M SD M SD M SD M SD M SD

5–6 0.20 7.290E-02 0.11 4.772E-02 0.22 6.526E-02 0.16 7.503E-02 0.19 8.645E-02 0.17 7.735E-028–9 0.17 8.701E-02 0.11 5.024E-02 0.18 4.917E-02 0.14 7.160E-02 0.19 5.917E-02 0.13 6.961E-0211–12 0.14 5.761E-02 0.08 2.925E-02 0.14 4.787E-02 0.1 3.690E-02 0.14 6.061E-02 0.11 6.119E-02

Note: M = mean and SD = standard deviation, MAD = mean absolute deviation.

DEVELOPMENT OF SUSTAINED ATTENTION 215

the intraindividual variability significantly increased between SRT1 and SRT2, t (49) =–3.22, p < = .001, then plateaued with no significant change present between SRT2 andSRT3, t (49) = .02, p = .98. The 5–6-and 8–9-year-olds were not significantly different, t(94) = 1.54, p = .13, while the 11–12-year-olds were significantly less variable, t (109) =2.65, p = .01. Refer to Table 3 for descriptive data.

Accuracy. For Score!, a significant group difference for number of correctresponses was detected, Kruskal-Wallis x2 (2, N = 50) = 17.88, p < .001. The 5–6-year-olds (Mdn = 6, Range = 1–8) were significantly less accurate than 8–9-year-olds (Mdn = 9,Range = 4–10), U = 47, p < .001. The accuracy of the 8–9- and 11–12-year-olds (Mdn =10, Range = 6–10) was not significantly different, U = 113.5, p = .07.

Sustained Attention and Load

Speed. There was a significant interaction between group and load, F (2,51) =4.37, p = .02, refer to Figure 4.

The high load task yielded significantly slower RT performance than the low loadtask for each of the age groups: 5–6-year-olds, t (16) = –3.90, p < .001; 8–9-year-olds, t(18) = –13.66, p < .001; 11–12-year-olds, t (17) = –10.14, p < .001. On the low load task,5–6-year-olds and 8–9-year-olds and 8–9-year-olds and 11–12-year-olds were signifi-cantly different, t (34) = 3.02, p = .01 and, t (36) = 2.11, p = .04. On the high load task, 5–6-and 8–9-year-olds’ performances were similar, t (34) = .100, p = .33. The 11–12-year-olds recorded significantly faster RTs than the other groups, t (36) = 2.54, p < .001.

Errors. Load was found to have a significant effect on percentage of errors, Fried-man x2 (1, N = 54) = 28.70, p < .001. The 5–6-and 8–9-year-olds committed more errors

Figure 4 Mean log10 reaction time across age and task load.

0

0.5

1

1.5

2

2.5

3

3.5

4

Low Load Task High Load Task

Load

Mea

n L

og

10 R

eact

ion

Tim

e (m

illis

eco

nd

s)

5-6 years

8-9 years11-12 years

216 J. BETTS ET AL.

on the high (Mdn = 61.11%, Range = 23.53%–92.86% and Mdn = 28.57%, Range =4.76%–66.67% respectively) than the low load task (Mdn = 15.79%, Range = 0–38.89%and Mdn = 11.11%, Range = 0–43.75% respectively), t (16) = 10.23, p < .001 and t (18)4.16, p < .001. The 11–12-year-olds did not differ significantly in number of errors com-mitted across the low (Mdn = 9.76%, Range = 0–28.57%) and high load (Mdn = 14.29%,Range = 0–42.1%) tasks, t (17) = 2.44, p = .03. On the low load task, there were no agedifferences for errors, Kruskal Wallace x2 (2, N = 54) = 2.73, p = .26. The observed differ-ences in speed on the low load task were therefore not translated into differences in accu-racy. On the high load task, there were significant age differences in errors, Kruskal-Wallis, x2 (2, N = 54) = 25.22, p < .001. The 5–6-year-olds committed significantly moreerrors than the 8–9-year-olds, U = 52.00, p < .001. The 8–9-year-olds and 11–12-year-oldswere not significantly different, U = 113.50, p = .08. Thus, 11–12-year-olds were the fast-est and most accurate, 5–6-year-olds were the slowest and least accurate, while 8–9-year-olds were as accurate as 11–12-year-olds but as slow as 5–6-year-olds.

Max Outs. On the low load task, each age group had a median number of max outsof 0 with a range of 0–0. On the high load task, group was found to have a significant effecton max outs, Kruskal-Wallis x2 (2, N = 54) = 29.88, p < .001. The 5–6-year-olds (Mdn = 8,Range = 3–11) committed significantly more max outs than the 8–9-year-olds (Mdn = 2,Range = 0–9), U = 33, p < .001. The 8–9-year-olds were significantly different from the 11–12-year-olds (Mdn = 0, Range = 0–7) in number of max outs, U = 95.5, p = .016.

Variability—SD. There was no effect of load as intraindividual variability did notchange significantly between tasks, F = 3.03, p = .09. There was a significant group effect,F (2,51) = 5.94, p = .01, with 5–6-year-olds having significantly greater variability than 8–9-year-olds, t (34) = 2.83, p = .008. The variability of 8–9-year-olds and 11–12-year oldsdid not differ significantly, t (35) = .30, p = .75. Refer to Table 4 for descriptive data.

Variability—MAD. Intraindividual consistency did not change from the low tohigh load task, Friedman x2 (1, N = 54) = .07, p = .79. A significant effect of group waspresent, Kruskal-Wallis x2 (2, N = 54) = 8.35, p = .02. This difference however did notoccur between the 5–6-and 8–9-year-olds, U = 112, p = .117, nor the 8–9-and 11–12-year-olds, U = 110, p = .064. Refer to Table 4 for descriptive data.

DISCUSSION

The aim of this study was to employ both traditional and novel indices to investigatethe development of sustained attention and the effect of task load on children’s perfor-mance. As predicted, sustained attention was found to develop throughout childhood.

Table 4 Variability data across age group fro low and high load tasks.

Low Load High Load

Standard DeviationMean Absolute

Deviation Standard DeviationMean Absolute

Deviation

Age M SD M SD M SD M SD

5–6 0.1776 5.802E-02 0.1682 5.829E-02 0.1366 7.758E-02 0.1162 8.927E-028–9 0.1257 4.109E-02 0.1189 4.793E-02 0.1170 5.724E-02 0.1229 6.353E-0211–12 0.1238 6.965E-02 0.1114 0.1142 0.1119 2.581E-02 0.1080 3.680E-02

Note: M = mean and SD = standard deviation.

DEVELOPMENT OF SUSTAINED ATTENTION 217

Rapid growth occurred from 5–6 to 8–9 years, and then a developmental plateau was evi-dent from 8–9 to 11–12 years with only minor improvement. As expected, the task param-eters of target complexity and display size influenced children’s sustained attention, withperformance poorer on the high load task than the low load task. The same developmentalpattern was evident on the load tasks.

As predicted, sustained attention showed development from 5–6 to 11–12 years. Forevery index (speed, errors, accuracy, and variability), performance was in the expecteddirection with increasing age associated with improved performance. On some indices,evidence for a plateau in development was present. For example, on indices of speed andaccuracy, the 8–9-year-olds’ performance was significantly better than that of 5–6-year-olds,but similar to that of 11–12-year-olds. In contrast, more consistent growth of sustainedattention across the age range was identified on other indices. For example, the 5–6-year-olds committed significantly more errors than the 8–9-year-olds who committed signifi-cantly more errors than the 11–12-year-olds. Both measures of variability — SD and MAD— showed the 5–6-year-olds to be significantly more variable than the 8–9-year-olds, whoin turn, were significantly more variable than the 11–12-year-olds. These differential find-ings suggest that the skills underpinning performance on measures of sustained attentionmay display varying developmental trajectories. Specifically, the 8–9-year-old grouprecorded response speeds as short as the 11–12-year-old group, but with less accuracy.This pattern of less accurate performance may reflect emerging but not yet establishedsustained attention skills in the 8–9-year-old children.

Although MAD indicated a change in consistency over time, this change was notidentified by SD. According to MAD, performances initially became more variable beforeplateuing and remaining consistent. As SD is vulnerable to the influence of outlying data,the conflicting results produced using the MAD statistic versus the SD statistic suggest thatin this case, outliers may have been present that masked the change in variability, possiblywhen concentration lapsed creating unusually slow responses (Garret & Nash, 2001).

The mixed developmental evidence after 8–9 years is consistent with the literatureon development of other cognitive skills (Anderson, Anderson, & Lajoie, 1996; Kirk,1985). Kirk (1985) found that when skills were approaching a new level of competencethere were clear differences on some indices but not others, due to a “reorganization offunction.” The mixed evidence is also consistent with the 8–9-year-olds having one or twoyears of rapid development left before they reach 10, the age at which the plateau is evi-dent in most studies (Klenberg et al., 2001; Manly et al., 2001; Rebok et al., 1997). Resultsfrom the present study are consistent with previous research showing that sustained atten-tion develops throughout childhood, with rapid improvements to age 10, then gradualimprovements thereafter (Klenberg et al., 2001; Manly et al., 2001; Rebok et al., 1997).

While Rebok et al. (1997) and Klenberg et al. (2001) found the decline in errors toplateau at age 10, the current study did not support this developmental pattern. Reboket al. (1997) divided errors into two types—errors of omission and commission, andKlenberg et al. (2001) worked with a performance “total score” involving subtractingerrors from correct responses. As errors rates were very low, the current study examinederrors of omission, commission, and anticipation together. A more detailed analysis ofeach type of error may have revealed the previously described developmental plateau.

As expected, performance was poorer on the high load task than the low load task,regardless of age. The parameters of target criteria and display size were successfullymanipulated to increase the sustained attention resources required to complete the tasks.When participants matched a pair of cards for color, suit, and number with one of six pairs

218 J. BETTS ET AL.

of cards (high load), their speed was slower and more errors and max outs were committedthan when they matched a single card for color with one other card (low load). The resultssuggest that participants were not engaging in a speed-accuracy trade-off because botherrors and speed increased. Variability, as measured by both SD and MAD, did not changefrom the low load to high load task, indicating that the slowing speed was not a function offluctuations in performance unduly influencing the mean or median, but it was due togreater sustained attention demands of the high load task.

The current findings are consistent with the adult literature. When target criteriawere manipulated, Noonan et al. (1984) and Lysaght et al. (1984) observed a decrease incorrect responses and an increase in RT on the high load task. These studies demonstratethat when task demands become too great, participants are unable to cope and perfor-mance deteriorates (Corkum & Siegel, 1993). This study has shown that increasing targetcriteria and display size places greater demands on children’s sustained attention system.Subsequently, these parameters may now be added to the small, but growing, list of fac-tors that affect children’s sustained attention, along with display time, target probability,and event rate (Chee et al., 1989; Laurie-Rose et al., 2001).

As these tasks place differing demands on participants’ sustained attention, the perfor-mance of each age group provides further insight into development. As expected, the develop-ment of sustained attention from 5–6 to 11–12 years was evident on all indices (speed,variability, errors, and max outs). In addition, the gradual improvements characterizing the pla-teau in development from 8–9 to 11–12 years was evident on some indices (speed-accuracytrade-off, variability), while continued improvements were evident on others (max outs). Con-tinued rapid growth was evident on the max out index, with groups not differing at low loadbut the 5–6-year-olds committing significantly more max outs than the 8–9-year-olds at highload, who in turn committed significantly more max outs than the 11–12-year-old group.

The plateau in development was evident on measures of speed, accuracy, and vari-ability. As a speed-accuracy trade-off was identified, these measures will be interpretedcollectively. At low load there were no age differences in speed or error rate. At high load,the speed of the 5–6-year-olds was not significantly different from the 8–9-year-olds,while the 8–9-year-olds were significantly slower than the 11–12-year-olds. The 5–6-year-olds committed significantly more errors than the 8–9-year-olds, while number oferrors did not differ for the two older groups. Thus, 8–9-year-old children were as accurateas 11–12-year-olds, but they took longer to produce these accurate responses.

With both MAD and SD, the variability of the 8–9-year-olds was similar to that dis-played by the 11–12-year-olds, providing evidence for the plateau in sustained attention.While the MAD found the 5–6- and 8–9-year-olds to be performing similarly, SD showedthem to be significantly different. The influence of outliers on SD is likely to be responsi-ble for this discrepancy (Garret & Nash, 2001). The 5–6-year-olds may have sufferedlapses in attention that resulted in unusually slow responses and increased their variability.

The current findings are consistent with the developmental literature that has investi-gated sustained attention with tasks of the same load. In accordance with Klenberg et al.(2001), Manly et al. (2001), and Rebok et al. (1997), sustained attention was found to developfrom 5–6 to 11–12 years, with a plateau in development after 8–9 years evident on some indi-ces. Finding a plateau on only some indices is consistent with Kirk’s (1985) “reorganization offunction” and is reasonable, considering the 8–9-year-olds still have one or two years of rapiddevelopment left before they reach the age at which the plateau becomes clearly evident.

Methodologically, the development of sustained attention was successfully capturedby traditional indices. While errors and accuracy are often subject to ceiling or floor

DEVELOPMENT OF SUSTAINED ATTENTION 219

effects, this was not the case in the present study as only 29% of participants respondedwithout error on Score!, 11% on the low load task and a single person on the high loadtask (Ballard, 1996a; Manly et al. 2001). Error and accuracy measures, however, producedskewed data that required nonparametric tests to be employed.

This research demonstrated that the novel index of variability can be successfullyapplied in the context of normative research. Variability was found to be a useful tool,providing a more direct observation of sustained attention from response performance,rather than from missed response performance (Castellanos & Tannock, 2002). The use ofthis index is in accordance with the principle of multiple measures, whereby utilizing arange of indices provides a more comprehensive picture. The similarities between resultsobtained in this study using variability and the more traditional indices, gives greaterweight to the validity of the findings. The variability data was not subject to floor or ceil-ing effects and, unlike the traditional indices, was normally distributed (Ballard, 1996a).These results suggest that variability should be included as an index in future sustainedattention research and, with its advantages over more traditional indices, may become oneof the primary indices of sustained attention.

The results obtained with SD and MAD were sometimes inconsistent, possibly dueto the sensitivity of these measures to nonsignificant outliers (Garret & Nash, 2001). In thefuture, studies of variability may need to employ a more lenient definition of an outlier toaddress this. Further research is also required to explain the underlying mechanismsresponsible for the performance variability. For example, the variability could be theresult of unusually slow responses caused by lapses in attention, unusually fast responsescaused by disinhibition and impulsiveness, or a combination of both.

This study was subject to the usual problems associated with testing attention,namely, that attention cannot be measured directly but must be inferred from task perfor-mance (van Zomeren & Brouwer, 1992). The current study therefore used two differenttests of attention (CogState and TEA-Ch), and several outcome indices (speed, errors,accuracy, and variability). This reduced the influence of extraneous factors and improvedthe validity of conclusions (Fletcher, 1998). For example, even though CogState requiresmotor responses and Score! verbal responses, both tests produced comparable results,indicating that method of response was not influencing the findings.

The current research has applications in educational settings, where the knowledge canbe employed to design schedules that most effectively use the limited hours in a school day.For example, the research indicates that younger students require more breaks and that 8–9-and 11–12-year-olds could follow a similar study-break schedule. In addition, the researchprovides insight into how to present information most efficiently to children. This is especiallyrelevant as computers and games are now common classroom learning tools. This study hasalso demonstrated that children can attend best when small amounts of information aredisplayed. These findings have important implications in clinical settings where a sound under-standing of normal sustained attention and its developmental trajectory may aid in the diagno-sis and treatment of attentional problems, such as Attention deficit/hyperactivity disorder.

REFERENCES

Anderson, P., Anderson, V., & Lajoie, G. (1996). Standardization of the Tower of London Test. TheClinical Neuropsychologist, 10, 54–65.

Ballard, J. C. (1996a). Computerized assessment of sustained attention: Interactive effects of taskdemand, noise, and anxiety. Journal of Clinical and Experimental Neuropsychology, 18(6), 864–882.

220 J. BETTS ET AL.

Ballard, J. C. (1996b). Computerised assessment of sustained attention: A review of factors affectingvigilance performance. Journal of Clinical and Experimental Neuropsychology, 18(6), 843–863.

Bleiberg, J., Garmoe, W. S., Halpern, E. L., Reevers, D. L., & Nadler, J. D. (1997). Consistency ofwithin-day and across-day performance after mild brain injury. Neuropsychiatry, Neuropsychol-ogy and Behavioural Neurology, 10(4), 247–253.

Castellanos, F. X., & Tannock, R. (2002). Neuroscience of attention-deficit/hyperactivity disorder:The search for endophenotypes. Nature Reviews Neuroscience, 3, 617–628.

Catroppa, C., & Anderson, V. (1999). Attentional skills in the acute phase following pediatric trau-matic brain injury. Child Neuropsychology, 5(4), 251–264.

Catroppa, C., & Anderson, V. (2003). Children’s attentional skills two years post-traumatic braininjury. Developmental Neuropsychology, 23(3), 359–373.

Chee, P. Logan, G., Schachar, R., Lindsay, P., & Wachsmuth, R. (1989). Effects of event rate anddisplay time on sustained attention in hyperactive, normal and control children. Journal of Abnor-mal Psychology, 17(4), 371–391.

CogState Ltd. (2002). CogState. Version 2.1.0. Melbourne, Australia.Corkum, P. V., & Siegel, L. S. (1993). Is the continuous performance task a valuable research tool

for use with children with attention-deficit-hyperactivity disorder? Journal of Child Psychologyand Psychiatry, 34(7), 1217–1239.

Daniel, A. (1983). Power, privilege and prestige: Occupations in Australia. Melbourne, Australia:Longman-Chesire.

Fletcher, J. M. (1998). Attention in children: Conceptual and methodological issues. Child Neurop-sychology, 4(1), 81–86.

Garret, L., & Nash, J. C. (2001). Issues in teaching the comparison of variability to non-statisticsstudents. Journal of Statistics Education, 9(2), 1–16.

Hultsch, D. F., MacDonald, S. W. S., & Dixon, R. A. (2002).Variability in reaction time perfor-mance of younger and older adults. Journal of Gerontology, 57(2), 101–115.

Hynd, G. W., Nieves, N., Connor, R. T., & Stone, P. (1989). Attention deficit disorder with andwithout hyperactivity: Reaction time and speed of cognitive processing. Journal of Learning Dis-abilities, 22(9), 573–580.

Kirk, U. (1985). Hemispheric contributions to the development of graphic skills. In C. Best (ed.),Hemispheric function and collaboration in the child (pp. 193–228). New York: AcademicPress.

Klenberg, L., Korkman, M., & Lahti-Nuuttila, P. (2001). Differential development of attention andexecutive functions in 3- to 12-year-old Finnish children. Developmental Neuropsychology,20(1), 407–428.

Laurie-Rose, C. L., Bennett-Murphy, L. B., Schickedantz, B., & Tucci, J. (2001). The effects ofevent rate and signal probability on children’s vigilance. Journal of Clinical and ExperimentalNeuropsychology, 23(2), 215–224.

Luce, R. D. (1986). Response times: Their role in inferring elementary mental organization. NewYork: Oxford University Press.

Lysaght, R. J., Warm, J. S., Dember, W. N., & Loeb, M. (1984). Effects of noise and information-processing demands on vigilance performance in men and women. In A. Mital (ed.), Trends inergonomics/human factors 1 (pp. 27–32). Amsterdam: Elsevier Science Publishers.

Makdissi, M., Collie, A., Maruff, P., Darby, D. G., Bush, A., McCrory, P., & Bennell, K. (2001).Computerised cognitive assessment of concussed Australian Rules footballers. British Journal ofSports Medicine, 35(5), 354–360.

Manly, T., Anderson, V., Nimmo-Smith, I., Turner, A., Watson, P., & Robertson, I.H. (2001). Thedifferential assessment of children’s attention: The test of everyday attention for children(TEA-Ch), normative sample and ADHD performance. Journal of Child Psychology, 42(8),1065–1081.

Manly, T., Robertson, I. H., Anderson, V., & Nimmo-Smith, I. (1999). The test of everyday attentionfor children. United Kingdom: Thames Valley Test Company.

DEVELOPMENT OF SUSTAINED ATTENTION 221

McKay, K. E., Halperin, J. M. Schwartz, S. T., & Sharma, V. (1994). Developmental analysis ofthree aspects of information processing: Sustained attention, selective attention and responseorganization. Developmental Neuropsychology, 10(2), 121–132.

Miller, P. J. (2002). Order in variability, variability in order: Why it matters for theories of develop-ment. Human Development, 45, 161–166.

Mirsky, A. F. (1996). Disorders of attention: A neurospychological perspective. In G. R. Lyon &N. A. Krasnegor (eds.), Attention, memory and executive function. Baltimore, Maryland: PaulH. Brookes Publishing Co., Inc.

Mirsky, A. F., Anthony, B. J., Duncan, C. C., Ahearn, M. B., & Kellam, S. G. (1991). Analysis of theelements of attention: A neuropsychological approach. Neuropsychology Review, 2(2), 109–145.

Noonan, T. K., Ash, D., Loeb, M., & Warm, J. S. (1984). Task complexity, noise and cognitive vig-ilance performance. In A. Mital (ed.), Trends in ergonomics/human factors 1 (pp. 33–38).Amsterdam: Elsevier Science Publishers.

Parasuraman, R. (1979). Memory load and event rate control sensitivity decrements in sustainedattention. Science, 205, 924–927.

Rebok, G. W., Smith, C. B., Pascualvaca, D. M., Mirsky, A. F., Anthony, B. J., & Kellam, S. G.(1997). Developmental changes in attentional performance in urban children from eight to thir-teen years. Child Neuropsychology, 3(1), 28–46.

Rowe, K. J. & Rowe, K. S. (1995). Rowe Behavioural Rating Inventory profile user’s guide: Inter-active software for the assessment and monitoring of child/student externalising behaviours athome and at school. Melbourne, Australia: Centre of Applied Educational Research and Depart-ment of Paediatrics, University of Melbourne.

Shapiro, M. B., Morris, R. D., Morris, M. K., Flowers, C., & Jones, R. W. (1998). A neuropsycho-logically based assessment model of the structure of attention in children. Developmental Neu-ropsychology, 14(4), 657–677.

Stuss, D., Shallice, T., Alexander, M., & Picton, T. (1995). A multidisciplinary approach to anteriorfunctions. Annals of the New York Academy of Sciences, 769, 191–211.

Stuss, D. T., Stethem, L. L., Hugenholtz, H., Picton, T., Pivik, J., & Richard, M. T. (1989). Reactiontime after head injury: Fatigue, divided and focused attention, and consistency of performance.Journal of Neurology, Neurosurgery and Psychiatry, 52, 742–748.

van der Meere, J. J., & Sergeant, J. A. (1988). Controlled processing and vigilance in hyperactivity:Time will tell. Journal of Abnormal Child Psychology, 16, 641–655.

van Zomeren, A., & Brouwer, W. H. (1992). Assessment of attention. In J. Crawford, M. Parker, &W. McKinlay (eds.), A handbook of neuropsychological assessment. (pp. 241–266). Hove, UK:Lawrence Erlbaum Associates.

Warm, J. S., Howe, S. R., Fishbein, H. D., Dember., W. N., & Sprague, R. L. (1984). Cognitivedemand and the vigilance decrement. In A. Mital (ed.), Trends in ergonomics/human factors 1(pp. 27–32). Amsterdam: Elsevier Science Publishers.