You are on page 1of 12

Child Neuropsychology 0929-7049/01/0702-072$16.

00
2001, Vol. 7, No. 2, pp. 72±83 # Swets & Zeitlinger

Sustained Attention and Unintentional Injury


Among Preschool-Aged Children
Laura Bennett Murphy1, Caroline E. Murphy2, and Cynthia Laurie Rose1
1
Department of Psychology, One Otterbein College, Westerville, OH, USA, and
2
Department of Psychology, Ohio University, Athens, OH, USA

ABSTRACT

This study examined vigilance in preschool-aged children and explored the relationship between vigilance
and unintentional injury. There were 28 participating children, aged 4 and 5 years, who completed a
computerized vigilance task for two 5-min sessions. The task generated measures of correct detections, false
alarms, reaction time, and the signal detection indices of d0 and c. Primary caregivers completed daily injury
phone journals for a 4-week period. Results indicated that age and signal probability affected vigilance.
Older children made more correct detections, had greater perceptual sensitivity, and performed in patterns
similar to adults. Performance was enhanced in the high signal probability condition. In addition, vigilance
indicators of perceptual sensitivity and response bias were predictive of injury, while age was not.
Speci®cally, children with lower perceptual sensitivity scores, and who were less responsive to the vigilance
task, experienced more unintentional injures over the course of the study.

The ability to sustain attention to stimuli over an Most research on children's vigilance has
extended period, facilitates the processing of focused on the identi®cation of attention pro-
information and thus forms a basis for a wide blems, differentiating clinical from non-clinical
array of cognitive functions (Parasuraman & populations. However, there is a dearth of norma-
Davies, 1977). Adult vigilance is of signi®cant tive developmental data on sustained attention.
interest in occupational settings where sustained This paucity of data makes assessment of de®cit
attention is key to safety, productivity, and com- or delay dif®cult (Burack & Enns, 1997). Further,
fort (Warm & Dember, 1998). Vigilance also is there are many implications for understanding the
relevant in children's lives. Whether one exam- development of children's vigilance beyond iden-
ines social interactions (Adamson & Russell, tifying clinical subgroups. For example, Edley
1999; Goldsmith & Rogoff, 1997), cognitive and Knopf (1987) suggest that vigilance may be
development (Bell, 1998; Choudhury & Gorman, an indicator for school readiness. Further, the
2000), or the early identi®cation of children at study of individual differences in vigilance may
risk (Bornstein, 1990; Fagan & Haiken-Vasen, be useful for theory construction. For instance, in
1997), attentional processes are central to expla- the adult literature some researchers have used
nations of individual behavior. ``Attention . . . is individual differences research to examine com-
conceded by everyone to be essential to an under- peting theories of vigilance (Berch & Kanter,
standing of behavior'' (Burack & Enns, 1997, 1984). Ballard (1996a) argued that to fully under-
p. 9). stand vigilance, one had to take into account

Address correspondence to: Laura Bennett Murphy, Ph.D., Department of Psychology, One Otterbein College, West-
erville, OH 43081, USA. Tel.: ‡ 1-614-823-1698. Fax: ‡ 1-614-823-3285. E-mail: LBennett-Murphy@otterbein.edu
Accepted for publication: October 17, 2001.
SUSTAINED ATTENTION AND INJURY 73

subject characteristics (here referred to as partic- Across subject groups, Kerns and Rondeau
ipant characteristics), task parameters of the vigil, described a developmental trend such that chil-
and the environment in which vigilance is being dren performed better on the task with age. While
assessed. All of these factors in¯uence sustained the authors made an important contribution to the
attention and may interact. One such participant literature and demonstrated that young children
characteristic is age. could successfully complete their continuous per-
Studies of sustained attention in children formance task, there were several limitations to
clearly document the dynamic nature of the the study. First, there was no systematic manip-
developmental course of attention. For example, ulation of task parameters known to affect vigi-
Levy (1979) demonstrated a distinct increase in lance performance in adults (e.g., event rate,
capacity for vigilance in normal children across stimulus duration, signal probability). Second,
the ages of 4±6 years. This development also while the study effectively differentiated clinical
correlates to an increase in ability for motor subgroups, there was no measure of functional
inhibition, which is essential for completing vig- utility for the typically developing children.
ilance tasks. Sarid and Breznitz (1997) found that Third, the authors did not examine changes in
the development of vigilance follows a linear attention over time. A common ®nding in such
trend until age 4 and then reaches a plateau tasks is the decline in performance over time,
around age 5, and that vigilance may develop as which is referred to as the vigilance decrement
a stage in a sequential development of different (Dember & Warm, 1979). Finally, Kerns and
aspects of attention. The results of a study con- Rondeau examined only correct detections and
ducted by Woody-Ramsey and Miller (1988) also errors of commission (false alarms). However,
give rise to the idea of sequential development: signal detection analyses often are applied in
The period between ages 3 and 6 was found to be adult vigilance studies and promise a fuller
a transition point in the development of selective assessment of participants' performance (Losier
attention skills. This dynamic character of devel- et al., 1996; Nuechterlein, 1983; O'Dougherty,
opment of attention and vigilance among the Nuechterlein, & Drew, 1984). The current study
preschool years presents a special challenge in adapts Kerns and Rondeau's task with children,
using a vigilance task to assess the level of atten- aged 4±5 years, and addresses the above four
tion of a preschool child. Since this period seems issues. Further, the current study examines how
to be a point of transition, it is likely that there age interacts with task variables.
exists wide variation in vigilance skills between Ballard (1996b) provides an extensive review
ages 3 and 5. However, beyond the studies men- of factors that affect performance on vigilance
tioned above, there remains limited information tasks, including participant characteristics (e.g.,
on the normal development of attention and appli- personality, level of arousal, demographics), task
cations of laboratory-based measures of attention parameters (e.g., event rate, stimulus duration,
to function. nature of the stimuli), and the testing environ-
While the majority of research with children ment. Unfortunately, many such factors and para-
has been conducted with school-aged clinical meters are ignored in vigilance studies with
samples (Huguenin, 1997; Losier, McGrath, & children, leading some researchers such as
Klein, 1996; Schachar, Logan, Wachsmuth, & Corkum and Siegel (1993) to question the util-
Chajczyk, 1988; Sykes, Douglas, & Morgenstern, ity of vigilance tasks in assessing attention in
1973), a few studies exist that have examined children.
sustained attention in younger children (Byrne, Signal probability is an example of a task
DeWolfe, & Bawden, 1998; Ruff, Capozzoli, & parameter that is virtually ignored in the child
Weissberg, 1998; Ruff, Lawson, Parrinello, & literature on vigilance. A powerful in¯uence on
Weissberg, 1990). Kerns and Rondeau (1998) adult response criterion, signal probability, is
developed a vigilance task for use with preschool de®ned by See, Howe, Warm, and Dember (1995)
children, aged 3±6 years, to differentiate children as the ratio of critical signals to the total number of
with atypical attention from their normal peers. events that occur in a given time period. Generally,
74 LAURA BENNETT MURPHY ET AL.

adults are more receptive to critical signals in The few studies that have attempted to mea-
higher signal probability conditions (Baddeley & sure change over time in children have success-
Colquhoun, 1969; Parasuraman & Davies, 1976). fully documented the vigilance decrement
Thus, an increase in signal probability typically (Halperin et al., 1991; Laurie Rose et al., 2001;
will elicit maximization of ef®ciency and accuracy McKay, Halperin, Schwartz, & Sharma, 1994).
of signal detection in adults (Jenkins, 1958; Warm The current study examines effects of prob-
& Alluisi, 1971). ability manipulations on preschool-aged chil-
Although signal probability manipulations dren's vigilance and examines vigilance over
have been made consistently in examinations time. Further, the current study considers the
of adult vigilance, only one study has made external validity of this task by examining the
such manipulations while examining vigilance relationship between vigilance and unintentional
in children (Laurie Rose, Bennett Murphy, injury. Unintentional injury remains the leading
Schickedantz, & Tucci, 2001). Laurie Rose et al. cause of death for Americans from birth to middle
(2001) found that children's performance was opti- adulthood (U.S. Senate, 1996). The U.S. Depart-
mal in higher probability conditions. However, ment of Health and Human Services identi®ed the
Losier and coworkers (1996) found that in studies reduction of unintentional injuries as a major
of children's vigilance, signal probability varies health goal for the United States (CDC, 1998).
from 8 to 30% of the total stimuli presented in a A non-fatal injury occurs every 2 s in the United
typical testing session. This range stands in stark States (Heubusch, 1998) and injury is the leading
contrast to the typical range presented in studies on cause of childhood morbidity and mortality in the
vigilance performance in adults, which Craig United States (CDC, 1998). Among children, the
(1984) identi®ed as between 2 and 5% of the total incidence of injury increases dramatically from
stimuli in a vigilance session. Clearly, a more birth to age 2, peaks between ages 2 and 4, and
consistent approach to signal probability and its then declines in the school years (Matheny, 1988).
manipulation in research on children's vigilance Because of the public health threat and the sig-
is necessary to gain a more complete picture of ni®cant cost in treating injuries (approximately
vigilance performance. $80 billion in 1994 [CDC]), this problem warrants
While inconsistent manipulation of task para- scienti®c attention (Peterson, Brown, Bartelstone,
meters has been one limitation of the literature on & Kern, 1996).
children's vigilance, failure to examine vigilance Identifying and eliminating risk factors for
over time is another. One of the most signi®cant injury has been a major goal of research and
®ndings in adult vigilance performance is the public health initiatives. Repeatedly, the presence
vigilance decrement. Known to appear within of attention de®cits has been identi®ed as a
the initial moments of a vigil, the vigilance positive risk factor for injury (Farmer & Peterson,
decrement is a decrease in accuracy of perfor- 1995; Rivara, 1995a). While it has been estab-
mance and increase in reaction time or response lished that children with attention de®cits do incur
latency to stimuli over time (See et al., 1995). unintentional injuries more often than their nor-
Participants usually demonstrate longer reaction mal peers (Farmer & Peterson, 1995), there are
times to presented stimuli, and correctly detect limitations in the literature. First, most studies on
fewer target signals. The vigilance decrement attention and injury have looked solely at school-
has not been explored consistently in studies with age children (Matheny, 1988; Peterson, 1994;
children (Corkum & Siegel, 1993). Most studies Rivara, 1995a, b), effectively ignoring children
of children's attention have calculated an average under 6 years. Further, there has been little
score over the entire vigilance session. Such examination of the relationship between attention
scores provide no information about how atten- and injury among typically developing children
tion changes over time. Halperin, Sharma, Green- with the notable exception of Pless and Taylor
blatt, and Schwartz (1991) argued that these (1995).
average scores actually may re¯ect selective, In a case-control study of injured children,
and not sustained, attention. aged 5±15, Pless and Taylor compared 286 children
SUSTAINED ATTENTION AND INJURY 75

who had been involved in traf®c injuries either as ing the relationship between visual attention and
a pedestrian or as a bicyclist. To be included in the the incidence of unintentional injury; and (3)
study, children could not have received severe establishing that a vigilance decrement can be
head trauma, upper limb damage, or sensory demonstrated in a 5-min vigil with young chil-
impairments. A total of 562 children served as dren. Further, the current study makes use of
controls. Children in the control group had experi- signal detection analyses. Signal detection ana-
enced injuries where their behavior was not a lysis are frequently utilized in the adult literature,
major factor in the injury, for example, as a and provide a ®ne-grained and sensitive analysis
passenger in a car. The study found that children, of performance. These analyses are described
who had been injured in part due to their own below.
behavior, made more errors of omission and Ballard (1996a) charged researchers to system-
commission on a continuous performance task atically examine participant, task, and environ-
of attention than the children in the control group. mental variables conjointly. The current study
The authors concluded that de®cits in vigilance accepted that charge by examining the participant
may be related to increased risk for injury in characteristic, age, and the task parameter, signal
typically developing children. probability, within the context of a constant test-
The Pless and Taylor study suggests that vari- ing environment. First, it was predicted that age
ability in attention may play a role in the occur- would affect vigilance. It was expected that older
rence of unintentional injury. However, little children (5-year-olds) would make more correct
remains known about the relationship between detections and fewer false alarms. In addition, it
attention and injury in preschool-age children was hypothesized that 5-year-olds would have
who experience the highest rates of injury. The faster reaction times, greater perceptual sensitiv-
current study further examines that relationship ity (d0 ), and greater conservatism (c). Based on
using a prospective design. ®ndings in the adult literature, it was predicted
As Pless and Taylor (1995) point out, prospec- that (1) probability would in¯uence performance
tive designs are almost impossible when studying with the low probability condition eliciting greater
serious injuries. One would have to follow more conservatism and slower response times than the
than 35,000 children for 1 year to yield 100 high probability condition; (2) over the duration of
relatively serious injuries (Pless & Taylor, 1995). the session, children would tend to adopt a more
However, minor unintentional injuries could serve conservative criterion (c), that is, children will be
as a useful dependent measure of attention. First, less willing to risk a false alarm and thus would
minor unintentional injuries affect all children and respond less frequently; and (3) a vigilance decre-
occur with some frequency. Second, there is wide ment would reveal itself by a decrement of hits, a
variability in the frequency of occurrence of injury. decline in perceptual sensitivity (d0 ), and an
Third, Peterson, Brown, and coworkers (1996) increase in reaction time. With respect to the
have demonstrated that minor injury can be reliably relationship between vigilance and injury, it was
measured and systematically studied. Finally, even hypothesized that those children with lower per-
minor injuries may cause distress to the involved ceptual sensitivity (d0 ) and a more lenient response
child and parent. For these reasons, injury was bias as indicated by a lower c value, would incur
selected as an index of external validity for the more unintentional injuries.
vigilance task.
The current study examines normal sustained
METHOD
attention with respect to the incidence of unin-
tentional injury in preschool children. This pro-
Participants
ject adapts an age-appropriate vigilance task for
The participants in this study were 28 children, aged
use with preschool-age children with normal 4 years, 0 months to 5 years, 11 months and their
attention patterns. The study builds on prior work respective primary caregivers. There were 12 young-
by: (1) manipulating a task parameter, signal sters aged 4; 16 were 5-years-old. The children, 17
probability, within the vigilance task; (2) examin- boys and 11 girls, had a mean age of 4.54 years and
76 LAURA BENNETT MURPHY ET AL.

all were Caucasian. All participants were recruited Injury Journals


from local preschools. The preschools were not Daily injury journals were administered to the
af®liated with the local college or university. As such, caregivers of participating children over a 4-week
participating children were representative of the period. These journals consisted of injury reports by
suburban area of a large Midwestern city from where the primary caregivers ± those persons most aware
they were drawn. Primary caregivers were the source of the children's activities during the past 24 hr.
of injury reports for the second part of the study. Reported injuries were coded and quanti®ed using
Since this study was concerned with normal the Minor Injury Severity Scale (MISS; Peterson,
patterns of attention, the Conners' Parent Rating Saldana, & Heiblum, 1996). The MISS categorizes
Scale ± Revised (Conners, 1997) was used to exclude injury into 21 types (e.g., cuts, burns). Injuries,
children with perceived atypical patterns of atten- then, are rated on a 7-point severity scale to quantify
tion. The Conners' is a standardized instrument used tissue damage. The measure has demonstrated
widely to assist in identifying inattention and acceptable reliability and stability (Peterson et al.,
overactivity. Each child's data was included only if 1996). The total number of injuries over the 4-week
the Conners' subscale scores indicated that the period was tabulated as the dependent measure
child's behavior was within the range speci®ed as because, unlike the severity ratings, the number of
normal (less than 1.5 SD above the M). All children injuries was both continuous and had acceptable
fell within the normal range, so no children were variability.
excluded.

Data Analysis
Measures For the vigilance task, a 2 (Probability: low ˆ.20,
high ˆ .30)2 (Age)5 (Periods of Watch) repeated
Vigilance Task measures ANOVA was utilized. The dependent
Visual attention was measured by a computer variables for sustained attention included the tradi-
vigilance task, which was adapted from Kerns and tional indices of percent hits, percent false alarms,
Rondeau (1998). Stimuli consisted of pictures of and reaction time for correct detections. In addition,
common animals: bird, dog, elephant, cat, duck, signal detection indices were used. Signal detection
rabbit, frog, ®sh, and pig. Pilot data determined that analysis yields two independent measures: percep-
children of this age range are familiar with these tual sensitivity, often indexed by d0 , and response
animals. Each picture was a simple black-and-white bias, indexed by c. Perceptual sensitivity scores
outline drawing of an animal, similar to those found assess an observer's ability to detect changes in
in children's coloring books. These pictures were stimuli, such as those between target and distracter
scanned and saved on a Toshiba Tectra 720 laptop stimuli. Response bias scores assess a subject's
computer as individual events for the task, which willingness to respond to stimuli. A high score
was programmed using SuperLab Pro software. indicates a conservative response criterion whereby
Children participated in two vigils during two the observer is quite cautious and unwilling to make
separate testing sessions: a low probability vigil and a false alarm in response to stimuli. A low score
a high probability vigil. In both vigils, the bird was shows a willingness to accept false alarms (See,
designated as the target stimulus and the remaining Warm, Dember, & Howe, 1997).
animals were designated distracters. Each of these To analyze the relationship between vigilance
vigils consisted of ®ve, 1-min continuous periods of and injury, data were collapsed across age groups
watch during which the children were asked to and a general linear model procedure was used. The
constantly attend to the computer screen. Probability dependent variable was the frequency of injury over
values of .2 and .3 were set for the low and the high 4 weeks. As described above, frequency of injury, as
probability conditions, respectively. Thus, in the low opposed to severity, was selected as the dependent
probability condition, each 1-min period of watch variable because it was continuous and had suf®cient
consisted of 40 stimuli, 8 of which were targets variability. The independent variables included the
among 32 distracters. The high probability condition signal detection indices of mean perceptual sensi-
also consisted of 40 stimuli, including 12 targets and tivity and mean response bias. These were selected
28 distracters for each period of watch. Prior to each for two reasons. First, due to the limited sample size,
vigil, participants were given a 30-s practice session. it was impossible to use all vigilance indices (e.g.,
The probability levels of the practice session were reaction time, correct detections, false alarms), and
consistent with the probability levels of the main perceptual sensitivity and response bias are inde-
vigil. All stimuli were presented for 1350 ms and the pendent from one another (See et al., 1997). Second,
inter-stimulus interval was 150 ms. signal detection scores utilize correct detection and
SUSTAINED ATTENTION AND INJURY 77

false alarm data to yield indices of behavioral style the screen. Because there were two experimental
toward a task. It was believed that these indices of conditions, high probability and low probability,
behavioral style would be more informative in researchers randomly administered the tasks on
explaining minor injuries. separate days. The low probability task was
Scores were recorded for the ®ve continuous 1- administered on 16 children on 1 day, and then the
min periods of watch of each vigilance condition for high probability task approximately 4 days later; for
each of the ®ve dependent variables. In order to the remaining 12 children, the high probability
correct for errors of anticipation, researchers con- condition was administered ®rst. All tasks were run
sidered a response as a false alarm if a participant in a small, quiet, well-lit room with a table and chair
pressed a key to the target stimulus during the scaled in size for preschool children. Following
200 ms following the onset of that particular completion of the vigilance task, caregivers com-
stimulus, a standard ®rst described by Nuechterlein pleted the injury journals Monday through Friday for
(1983) and O'Dougherty et al., (1984). Responses 4 weeks. Upon completion of the study, all families
also were recorded as false alarms if the participants were entered in a raf¯e for one of three $ 40 awards.
responded to a distracter. Responses recorded be-
tween 200 ms after the presentation of the target
stimulus and prior to the onset of the next stimulus RESULTS
were recorded as correct detections.
Age of participant affected vigilance perfor-
Procedure mance. The analysis of correct detections
Letters were sent home to parents from three local revealed a signi®cant main effect for Age (F (1,
preschool classes that served children aged 4 and 5.
Approximately 40 letters were sent home and 28 26) ˆ 5.38, p<.05). Five-year-olds made more
families completed the protocol. The 28 participat- correct detections than 4-year-olds (55% vs.
ing families were representative of the preschool 38%, respectively). An AgePeriods of Watch
classrooms from which they were recruited. Partic- interaction approached signi®cance for correct
ipating parents returned a written informed consent detections (F (4, 104)ˆ 2.349, p<.06). This trend
form to the school. Researcher assistants then towards an interaction, presented in Figure 1,
traveled to the preschools to administer the vigilance shows that the 5-year-olds tended to decline in
task; at that time the child's verbal assent was
obtained. The protocol established by Kerns and accuracy, while 4-year-olds demonstrated a ¯at
Rondeau (1998) was followed. Each child was asked function of low accuracy. Tests of simple effects
to help `catch' all the birds on the computer by show that age was signi®cant at the ®rst three
pressing the spacebar each time a bird appeared on periods of watch, with the older group performing

Fig. 1. Age GroupPeriod of Watch interaction on mean percent correct detections in 4- and 5-year-old children.
78 LAURA BENNETT MURPHY ET AL.

Fig. 2. Age GroupProbability interaction on mean reaction time in 4- and 5-year-old children.

signi®cantly better than the younger group, F (1, In terms of perceptual sensitivity (d0 ), 5-year-
26)ˆ 8.46, p<.01, F (1, 26)ˆ 6.26, p<.05, and F olds were signi®cantly more sensitive than 4-
(1, 26)ˆ 4.14, p<.05, for periods 1, 2, and 3, year-olds (F (1, 26)ˆ 4.68, p<.05). Mean d0 scores
respectively. Age did not affect commission of were 0.15 and 0.63 for 4- and 5-year-olds, respec-
false alarms. tively. Finally, with respect to the signal detection
With respect to reaction time, there was an index of response bias, there was a non-signi®cant
AgeProbability interaction (F (1, 26) ˆ 4.636, trend for an AgeProbability interaction (F (4,
p<.05). Five-year-old children performed signi®- 104) ˆ 2.349, p<.06). As illustrated in Figure 3,
cantly slower in the low probability condition. both groups became less conservative in the high
This interaction can be seen in Figure 2. Tests of probability condition, however, this pattern was
simple effects con®rmed the observation that age more pronounced for the older children.
was signi®cant only in the low probability condi- Results of the analyses of variance performed
tion (F (1, 26) ˆ 3.929, p<.05). on mean reaction times revealed a signi®cant

Fig. 3. ProbabilityPeriod of Watch interaction on mean reaction time in the low and high probability tasks.
SUSTAINED ATTENTION AND INJURY 79

ProbabilityPeriod of Watch interaction (F (4, the number of injuries also increased (b coef®-
104) ˆ 2.999, p<.05) as evidenced by a lengthen- cientˆ ÿ5.568). R2 values for mean perceptual
ing in reaction times across the vigilance task only sensitivity and mean response bias were .16 and
in the low probability condition (tests of simple .30, respectively.
effects: F (4, 104) ˆ 2.999, p<.05). Also, as
described above, probability interacted with age
group such that 5-year-olds had longer reaction DISCUSSION
times in the low probability condition.
There was a signi®cant main effect for Prob- This study measured preschool-aged children's
ability in predicting response bias (F (1, 26) ˆ vigilance and provided evidence for the func-
12.7, p<.001). Children became less conservative tional utility of such measures. The current
in responding in the high probability condition. study illustrates that within a 5-min vigil: (1)
Further, as described above, 5-year-olds were participant and task parameters critically affect
most sensitive to this manipulation. A trend vigilance, (2) a vigilance decrement might be
approaching signi®cance suggested that 5-year- likely, at least with 5-year-olds, and (3) signal
olds became less conservative in the high prob- detection analyses might capture individual dif-
ability condition. ferences in behavior. Finally, the study shows
The number of injuries each child incurred that perceptual sensitivity and response bias
within the 4-week period ranged from 2 to 11 accounted for almost half of the variability in
injuries, with a weekly mean of 1.41 injuries. The unintentional minor injuries over a month's time.
average severity rating for these injuries was 1.47 With respect to the study hypotheses, many
on a scale from 0 to 6, where 0 re¯ects no tissue were supported. First it was hypothesized that age
damage and 6 re¯ects permanent dis®gurement or would affect performance in all indices of vigi-
impairment. In general, the 1±2 range indicates a lance (correct detections, false alarms, reaction
minor injury with modest bleeding, swelling, or time, perceptual sensitivity, and response bias).
both. Average scores for perceptual sensitivity Indeed, older children made more correct detec-
and response bias for each probability condition tions and had greater perceptual sensitivity than
were entered into an univariate general linear 4-year-olds. As described below, age effects were
model to predict injury. Since age and gender modi®ed by probability condition in predicting
are factors known to in¯uence injury, a prelimin- reaction time and response bias. No age effects
ary analysis was conducted to determine whether were found with respect to false alarms.
these variables should be included as covariates in Beyond the age hypotheses, several predic-
the model. Results revealed that there was no tions were made regarding vigilance and most
relationship between age, gender, and injury in were supported. First, it was hypothesized that the
this study, so these variables were excluded. low probability condition would be associated
In the high probability task, neither of the with greater conservatism and longer reaction
variables reached signi®cance as predictors of times. The hypothesis was partially supported.
unintentional injury. In the low probability task, Five-year-olds demonstrated longer reaction
however, both mean perceptual sensitivity and times. Further, there was a trend for higher re-
mean response bias scores were found to be sig- sponse bias scores (greater conservatism) in the
ni®cant predictors of injury (F (2, 25)ˆ 6.7, p<.05, low probability condition. There was no support
Z2 ˆ .21 and F (1, 26)ˆ 18.656, p<.001, Z2 ˆ .43, for the hypothesis that children would become
respectively). The observed power for the model more conservative over time. This may be due, in
was .982 and the model accounted for 46% of the part, to the brevity of the task required by such
variability in injury. As mean perceptual sensitivity young participants. Finally, as a few others have
scores increased, indicating increased sensitivity demonstrated (Halperin et al., 1991; Laurie Rose
to changes in stimuli, the number of injuries et al., 2001; McKay et al., 1994), the current
incurred decreased (b coef®cientˆ ÿ1.422). As vigilance task was useful in eliciting a vigilance
mean response bias scores increased, however, decrement in 5-year-old children similar to those
80 LAURA BENNETT MURPHY ET AL.

observed in adult studies (Ballard, 1996b). This action. This hesitancy may prevent children from
decline in vigilance performance was manifested acting to avoid or escape injury.
in general by longer reaction times over time and Signal probability, notably lower signal prob-
a trend for decreased correct detections. These ability, emerges from the ®ndings as a critical
effects were modi®ed by age group and signal variable in predicting vigilance performance and
probability manipulations. unintentional injury. It is possible that low prob-
Thus, in general, 5-year-old children responded abilities, with the higher demand on vigilance,
to the vigilance task much like adults and older may produce greater discrimination between par-
children, in terms of responding to probability ticipants. The current study suggests that the
manipulations, and by evidencing a vigilance external validity of a vigilance task may depend
decrement (Baddeley & Colquhoun, 1969; Jenkins, upon the context created by task parameters.
1958; Laurie Rose et al., 2001; Parasuraman & Given the typically high probability levels of
Davies, 1976; See et al., 1997; Warm & Alluisi, children's tasks (Losier et al., 1996) and dearth
1971). Prior studies have described the period of data on signal probability manipulations,
between 3 and 6 years as a transition point in researchers must be increasingly attentive to sig-
the development of vigilance abilities (Sarid & nal probability. Clearly more work is needed to
Breznitz, 1997; Woody-Ramsey & Miller, 1988). understand the relationship between characteris-
The current study lends modest support to the tics of the task itself and its applications.
identi®cation of the preschool years as a period The ®nding that injury was predicted only in
of change. the low probability condition may have practical
Perhaps the most important outcome of this implications. For children under the age of 5
study is the relationship between laboratory mea- years, 79% of injuries occur in the home (Carter
surement of vigilance and unintentional injury in & Jones, 1993). One might speculate that the
the child's daily environment. Perceptual sensi- home, by virtue of its familiarity, has relatively
tivity and response bias in the low probability few critical signals for danger. This natural, low
condition accounted for 15.8 and 30.4% respec- probability condition may lead to less vigilance
tively, of the variability in injury. The inverse on the part of parents and children, thereby lead-
relationship between perceptual sensitivity and ing to more injury.
unintentional injury in the low probability task Finally, with respect to measurement issues,
supports the hypothesis that children who are less this study demonstrated the added sensitivity and
vigilant are injured more often than those who utility of signal detection analysis. Perceptual
maintain high vigilance. Participants' with low sensitivity (d0 ) and response bias (c) were in¯u-
perceptual sensitivity, that is, lower receptivity enced by age and signal probability manipulations
to changes between target and distracter, experi- in expected directions, and were related to beha-
ence a higher incidence of minor unintentional vior outside the laboratory: unintentional injury.
injuries. Perceptual sensitivity and response bias, which
The positive relationship between response incorporate traditional measures of attention, may
bias and incidence of injury was unexpected and be more sensitive and speci®c in describing vigi-
in the opposite direction than predicted. The lance. Given that these two factors accounted for
results indicate that children with a more con- almost half of the variability in unintentional
servative response bias experienced more injuries. injury, the functional utility of the indices appears
One explanation for this ®nding may be that an high. Obviously, more research is needed to
exceedingly conservative response bias may indi- examine the relationships between perceptual
cate a general lack of receptiveness to cues in the sensitivity, response bias, traditional markers of
environment. While more research clearly is attention, and behavioral applications such as
needed to understand the relationship between injury or school performance.
response bias and injury, it may be that children In conclusion, the task was successful in asses-
who are on the high end of conservatism demon- sing preschool children's level of attention and
strate a hesitancy to respond to critical signals for provided preliminary support for the external
SUSTAINED ATTENTION AND INJURY 81

validity of such tasks with children of normal vigilance performance. Journal of Clinical and
attention. Future studies may bene®t from Experimental Neuropsychology, 18, 843±863.
Bell, M.A. (1998). Frontal lobe function during
expanding the age range of participants to clarify infancy: Implications for the development of
the developmental picture of attention that begins cognition and attention. Cognitive neuroscience of
to emerge from this study. Further, future studies attention: A developmental perspective (pp. 287±
must systematically examine the effects of task 316). Mahwah, NJ: Lawrence Erlbaum Associates.
parameter manipulations. Signal probability is Berch, D.B., & Kanter, D.R. (1984). Individual
differences. In J.S. Warm (Ed.), Sustained attention
rarely manipulated, and yet diagnostic instru- in human performance (pp. 143±178). Chichester,
ments are being developed with little heed to UK: Wiley.
optimal signal probabilities. Further, as outlined Bornstein, M.H. (1990). Attention in infancy and the
by Ballard (1996a) a whole host of task para- prediction of cognitive capacities in childhood.
meters can serve to facilitate or impede the In J.T. Enns (Ed.), The development of attention:
Research and theory (pp. 3±19). Amsterdam: North-
maintenance of vigilance. In order to develop Holland.
appropriate diagnostic tools or computerized edu- Burack, J.A., & Enns, J.T. (1997). Attention, develop-
cational devices, a better understanding of what ment, and psychopathology. New York: Guilford
elements of the task promote sustained attention Press.
clearly is warranted. In terms of injury prevention, Byrne, J.M., DeWolfe, N.A., & Bawden, H.N. (1998).
Assessment of attention-de®cit hyperactivity dis-
more work is needed to apply the ®ndings to order in preschoolers. Child Neuropsychology, 4,
prevent injuries. Children lower in perceptual 49±66.
sensitivity and responsiveness may need more Carter, Y.H., & Jones, P.W. (1993). Accidents among
frequent `critical signals' to warn them of poten- children under ®ve years old: A general practice
tial hazards. This may be particularly true in based study in North Staffordshire. British Journal
of General Practice, 43, 159±163.
contexts low in stimulation. Future studies are Center for Disease Control and Prevention (CDC),
needed to examine the relationships between pre- (1998). Health, United States, 1998. Washington,
school and home environments, vigilance, and DC: U.S. Department of Health and Human
injury. Services.
Choudhury, N., & Gorman, K.S. (2000). The relation-
ship between sustained attention and cognitive
performance in 17±24 month old toddlers. Infant
ACKNOWLEDGEMENT and Child Development, 9, 127±146.
Conners, C.K. (1997). Conners Rating Scales ±
Appreciation is expressed to the Otterbein Col- Revised: Technical manual. North Tonawanda,
lege Student Research Fund. NY: Multi-Health Systems.
Corkum, P.V., & Siegel, L.S. (1993). Is the Continuous
Performance Task a valuable research tool for use
with children with Attention De®cit-Hyperactivity
REFERENCES Disorder? Journal of Child Psychology and Psy-
chiatry and Allied Disciplines, 34, 1217±1239.
Adamson, L.B., & Russell, C.L. (1999). Emotion Craig, A. (1984). Human engineering: The control of
regulation and the emergence of joint attention. vigilance. In J.S. Warm (Ed.), Sustained attention in
Early social cognition: Understanding others in the human performance (pp. 247±291). Chichester, UK:
®rst months of life (pp. 281±297). Mahwah, NJ: Wiley.
Lawrence Erlbaum Associates. Dember, W.N., & Warm, J.S. (1979). Psychology of
Baddeley, A.D., & Colquhoun, W.P. (1969). Signal perception (2nd ed.). New York: Holt, Rinehart and
probability and vigilance: A reappraisal of the Winston.
``signal rate'' effect. British Journal of Psychology, Edley, R.S., & Knopf, I.J. (1987). Sustained attention as
60, 165±178. a predictor of low academic readiness in a preschool
Ballard, J.C. (1996a). Computerized assessment of population. Journal of Psychoeducational Assess-
sustained attention: Interactive effects of task ment, 4, 340±352.
demand, noise, and anxiety. Journal of Clinical Fagan, J.F., & Haiken-Vasen, J. (1997). Selective
and Experimental Neuropsychology, 18, 864±882. attention to novelty as a measure of information
Ballard, J.C. (1996b). Computerized assessment of processing across the lifespan. In J.A. Burrack &
sustained attention: A review of factors affecting J.T. Enns (Eds.), Attention, development, and
82 LAURA BENNETT MURPHY ET AL.

psychopathology (pp. 55±73). New York: Guilford Parasuraman, R., & Davies, D.R. (1976). Decision
Press. theory analysis of response latencies in vigilance.
Farmer, J.E., & Peterson, L. (1995). Injury risk factors Journal of Experimental Psychology: Human Per-
in children with attention de®cit hyperactivity ception and Performance, 2, 578±590.
disorder. Health Psychology, 14, 325±332. Parasuraman, R., & Davies, D.R. (1977). A taxonomic
Goldsmith, D.F., & Rogoff, B. (1997). Mothers' and analysis of vigilance performance. In R.R. Mackie
toddlers' coordinated joint focus of attention: (Ed.), Vigilance: Theory, operational performance,
Variations with maternal dysphoric symptoms. and physiological correlates (pp. 559±574). New
Developmental Psychology, 33, 113±119. York: Plenum Press.
Halperin, J.M., Sharma, V., Greenblatt, E., & Schwartz, Peterson, L. (1994). Child injury and abuse-neglect:
S.T. (1991). Assessment of the continuous perfor- Common etiologies, challenges, and courses toward
mance test: Reliability and validity in a non-referred prevention. Current Directions in Psychological
sample. Psychological Assessment: A Journal of Science, 3, 116 ±120.
Consulting and Clinical Psychology, 3, 603±608. Peterson, L., Brown, D., Bartelstone, J., & Kern, T.
Heubusch, K. (1998). Crashes, ®res, and falls. Amer- (1996). Methodological considerations in partici-
ican Demographics, 20, 49±54. pant event monitoring of low-base-rate events in
Huguenin, N.H. (1997). Employing computer technol- health psychology: Children's injuries as a model.
ogy to assess visual attention in young children and Health Psychology, 15, 124±130.
adolescents with severe mental retardation. Journal Peterson, L., Saldana, L., & Heiblum, N. (1996).
of Experimental Child Psychology, 65, 141±170. Quantifying tissue damage from childhood injury:
Jenkins, H.M. (1958). The effects of signal rate on The minor injury severity scale. Journal of Pediatric
performance in visual monitoring. American Jour- Psychology, 21, 251±267.
nal of Psychology, 71, 647±651. Pless, I.B., & Taylor, H.G. (1995). The relationship
Kerns, K.A., & Rondeau, L.A. (1998). Development of between vigilance de®cits and traf®c injuries
a continuous performance test for preschool chil- involving children. Pediatrics, 95, 219±224.
dren. Journal of Attention Disorders, 2, 229±238. Rivara, F.P. (1995a). Developmental and behavioral
Laurie Rose, C., Bennett Murphy, L., Schickedantz, B., issues in childhood injury prevention. Developmen-
& Tucci, J. (2001). The effects of event rate and tal and Behavioral Pediatrics, 16, 362±370.
signal probability on children's vigilance. Journal of Rivara, F.P. (1995b). Crime, violence, and injuries in
Clinical and Experimental Neuropsychology, 23, children and adolescents: Common risk factors?
215±224. Criminal Behavior and Mental Health, 5, 367±385.
Levy, F. (1979). The development of sustained attention Ruff, H.A., Capozzoli, M., & Weissberg, R. (1998).
(vigilance) and inhibition in children: Some norma- Age, individuality, and context as factors in
tive data. Journal of Child Psychology and Psy- sustained visual attention during the preschool
chiatry, 21, 77±84. years. Developmental Psychology, 34, 454 ± 464.
Losier, B.J., McGrath, P.J., & Klein, R.M. (1996). Error Ruff, H.A., Lawson, K.R., Parrinello, R., & Weissberg,
patterns on the continuous performance test in non- R. (1990). Long-term stability of individual differ-
medicated and medicated samples of children with ences in sustained attention in the early years. Child
and without ADHD: A meta-analytic review. Development, 61, 60 ± 75.
Journal of Child Psychology and Psychiatry and Sarid, M., & Breznitz, Z. (1997). Developmental
Allied Sciences, 37, 971±987. aspects of sustained attention among 2- to 6-year
Matheny, A.P., Jr. (1988). Accidental injuries. In D.K. old children. International Journal of Behavioral
Routh (Ed.), Handbook of pediatric psychology (pp. Development, 21, 303±312.
108±133). New York: Guilford Press. Schachar, R., Logan, G., Wachsmuth, R., & Chajczyk,
McKay, K.E., Halperin, J.M., Schwartz, S.T., & D. (1988). Attaining and maintaining preparation: A
Sharma, V. (1994). Developmental analysis of three comparison of attention in hyperactive, normal, and
aspects of information processing: Sustained atten- disturbed control children. Journal of Abnormal
tion, selective attention, and response organization. Child Psychology, 16, 361±378.
Developmental Neuropsychology, 10, 121±132. See, J.E., Howe, S.R., Warm, J.S., & Dember, W.N.
Nuechterlein, K.H. (1983). Signal detection in vigi- (1995). Meta-analysis of the sensitivity decrement
lance tasks and behavioral attributes among off- in vigilance. Psychological Bulletin, 117, 230±249.
spring of schizophrenic mothers and among See, J.E., Warm, J.S., Dember, W.N., & Howe, S.R.
hyperactive children. Journal of Abnormal Psychol- (1997). Vigilance and signal detection theory: An
ogy, 92, 4 ± 28. empirical evaluation of ®ve measures of response
O'Dougherty, M., Nuechterlein, K.H., & Drew, B. bias. Human Factors, 39, 14±29.
(1984). Hyperactive and hypoxic children: Signal Sykes, D.H., Douglas, V.I., & Morgenstern, G. (1973).
detection, sustained attention, and behavior. Journal Sustained attention in hyperactive children. Journal
of Abnormal Psychology, 93, 178±191. of Child Psychology and Psychiatry, 14, 213±220.
SUSTAINED ATTENTION AND INJURY 83

U.S. Senate (1996). NIH revitalization act of 1996. Warm (Eds.), Viewing psychology as a whole: The
Washington, DC: GPO. integrative science of William N. Dember (pp. 87±
Warm, J.S., & Alluisi, E.A. (1971). In¯uence of temporal 112). Washington, DC: American Psychological
uncertainty and sensory modality of signals on watch- Association.
keeping performance. Journal of Experimental Psy- Woody-Ramsey, J., & Miller, P.H. (1988). The facil-
chology, 87, 303±308. itation of selective attention in preschoolers. Child
Warm, J.S., & Dember, W.N. (1998). Tests of vigilance Development, 59, 1497±1503.
taxonomy. In R.R. Hoffman, M.F. Sherrick, & J.S.

You might also like