\

Neuropsycholo`ia\ Vol[ 24\ No[ 5\ pp[ 636Ð647\ 0886

Þ 0886 Elsevier Science Ltd[ All rights reserved
Pergamon PII] S9917Ð2821"86#99904Ð7 Printed in Great Britain
9917Ð2821:86 ,06[99¦9[99

{Oops;|] Performance correlates of everyday

attentional failures in traumatic brain injured and
normal subjects
IAN H[ ROBERTSON\ TOM MANLY\ JACKIE ANDRADE\ BART T[ BADDELEY and
JENNY YIEND
MRC Applied Psychology Unit\ Cambridge\ U[K[

"Received 18 September 0885^ accepted 13 December 0885#

Abstract*Insu.cient attention to tasks can result in slips of action as automatic\ unintended action sequences are triggered
inappropriately[ Such slips arise in part from de_cits in sustained attention\ which are particularly likely to happen following frontal
lobe and white matter damage in traumatic brain injury "TBI#[ We present a reliable laboratory paradigm that elicits such slips of
action and demonstrates high correlations between the severity of brain damage and relative!reported everyday attention failures in
a group of 23 TBI patients[ We also demonstrate signi_cant correlations between self! and informant!reported everyday attentional
failures and performance on this paradigm in a group of 64 normal controls[ The paradigm "the Sustained Attention to Response
Task*SART# involves the withholding of key presses to rare "one in nine# targets[ Performance on the SART correlates signi_cantly
with performance on tests of sustained attention\ but not other types of attention\ supporting the view that this is indeed a measure
of sustained attention[ We also show that errors "false presses# on the SART can be predicted by a signi_cant shortening of reaction
times in the immediately preceding responses\ supporting the view that these errors are a result of {drift| of controlled processing
into automatic responding consequent on impaired sustained attention to task[ We also report a highly signi_cant correlation of
−9[47 between SART performance and Glasgow Coma Scale Scores in the TBI group[ Þ 0886 Elsevier Science Ltd[

Key Words] attention^ traumatic brain injury^ attentional failures in daily life^ sustained attention^ brain damage[

Introduction attention system of the brain described by Posner and

Peterson ð13Ł "0889# is important in such errors[ In the
Oops;*Pouring cream into a requested black co}ee or present paper\ we de_ne sustained attention as the ability
throwing away the vegetables while keeping their peelings to self!sustain mindful\ conscious processing of stimuli
are examples of action slips common in the everyday lives whose repetitive\ non!arousing qualities would otherwise
of normal people ð2\ 10Ł[ Such slips tend to happen when lead to habituation and distraction to other stimuli[ In
attention to task is degraded through such factors as short\ we distinguish between the capacity for endogen!
boredom\ worry or dividing attention between several ous modulation of alertness "self!sustained attention#
tasks simultaneously[ There is a considerable normal with exogenously controlled alertness\ which is governed
variation in action!error!proneness ð2Ł\ and brain dam! by factors such as novelty\ salience and stimulus change[
age*particularly to the frontal lobes of the brain* A possible link between slips of action on the one hand
increases the likelihood that individuals will stray from and sustained attention abilities on the other would be
intended goals and hence make errors ð22\ 23Ł[ important for two reasons]
While many di}erent cognitive processes may underlie
0[ The search for attentional performance measures that
such slips of action ð21Ł\ it is likely that the sustained correlate with everyday slips of action in the normal
population have yielded little success ð14Ł\ and this
*ÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐ may be because adequate measures of sustained atten!
 Address for correspondence] Ian H[ Robertson\ MRC tion were not obtained[
Applied Psychology Unit\ Rehabilitation Research Group\ Box
47\ Addenbrooke|s Hospital\ Cambridge CB1 1QQ\ U[K[^ tel[] 1[ We lack an adequate characterization of the attention
¦33 0112 244183\ ext[ 243^ fax] ¦33 0112 405529^ e!mail] de_cits shown by traumatically brain injured "TBI#
ian[robertsonÝmrc!apu[cam[ac[uk[ patients[ The clinical tests that have been shown to be

636
637 I[ H[ Robertson et al[:Everyday attentional failures

sensitive to TBI\ such as the Paced Auditory Serial degrade targets or load working memory in order to
Addition Test "PASAT# ð00\ 01Ł\ involve multiple cog! reduce high levels of performance ð11Ł[ In support of this
nitive operations\ and hence it is not possible to delin! view\ one study showed that vigilance decrements were
eate precisely the way in which patients fail on this test[ only observed in a task that required controlled process!
In fact\ most authors in this area interpret impaired ing\ but not in one where the responding relied on auto!
clinical performance on such tests as being due to matic processing ð8Ł[
reduced speed of processing ð00\ 01\ 31Ł\ rather than We proposed that sustained attention to task would be
in terms of any more speci_c attentional processes[ taxed more heavily "and therefore that a greater range of
As shall be seen below\ there are strong grounds for performance would be seen in tasks of shorter duration#
believing that sustained attention may be particularly if the automatic response set could be transferred to the
compromised following TBI\ and hence an attempt non!targets[ In this case\ when rare targets occur\ active\
at a more theoretically coherent characterization of controlled processing must be triggered to overcome or
attentional failures following TBI\ partly in terms of out!compete the prepotent automatic response[ Hence\
impaired sustained attention\ seems warranted[ in the present study\ we used a continuous performance
paradigm involving key presses to frequently presented
Traumatic brain injury particularly a}ects the frontal non!targets\ but with the requirement to withhold motor
lobes ð27\ 32Ł and white matter ð03\ 27Ł of the brain[ responses to occasional targets "Sustained Attention to
White matter damage has been shown to a}ect sustained Response Task*SART#[ It was predicted that such a
attention particularly ð15\ 29Ł\ as have frontal lobe lesions task would require a high level of continuous attention
mainly of the right hemisphere ð5\ 6\ 12\ 18\ 35Ł[ Reported to response and be sensitive to transitory reduction in
problems of attention and concentration occur in the attention or {lapses|\ while keeping to a minimum
majority of severely traumatically brain!injured patients demands on other cognitive processes such as memory\
ð06\ 39Ł[ planning and general intellectual e}ort[
In the present paper\ we argue that the action slips of Taking the view that action lapses in both normals
the normal population show characteristics in common and TBI patients can be attributed in part to sustained
with the attentional failures of traumatically brain injured attention de_cits\ the following hypotheses were formu!
patients\ albeit in a less extreme form[ We argue that lated[
one signi_cant factor determining such slips are transient
lapses in attention to task indicative of faulty sustained
attention[ In contrast to early work suggesting that sus!
Hypothesis 0
tained attention or vigilance in normal humans only
shows decrements after several tens of minutes ð05Ł\ recent
It was hypothesized that there would be a signi_cant
research shows that right fronto!parietal systems are
positive correlation in a non!brain damaged sample
active over periods as short as 39 sec ð12Ł\ and perhaps
between sustained attention capacities "as measured by
even over briefer periods ð33Ł[
the SART error score# and self! and informant!reported
The vulnerability of frontal and white matter areas to
attentional slips in everyday life[
traumatic brain injury leads plausibly to the prediction
It was further hypothesized that there would be no
that such patients will display sustained attention de_cits\ signi_cant relationship between performance on a more
and such a prediction is also reinforced by the nature of conventional perceptual detection!based test of sustained
the attention problems reported by relatives of traumatic attention "Triplets test# and these questionnaire
brain injured patients[ Detection of such de_cits using measures\ because of the proposed additional sensitivity
conventional vigilance!based perceptual detection para! of the SART to mild attentional de_cits[
digms has yielded mixed results\ however ð3\ 04\ 11\ 34Ł\
and some authors have even denied that traumatic brain
injury results in attentional problems over and above
di.culties presented by general mental slowing ð31Ł[ Hypothesis 1
One reason for the di.culty experienced to date in
_nding consistent performance correlates of sustained We predicted that the TBI group would make sig!
attention de_cits reported by brain injured people\ may ni_cantly more errors than a matched control group on
well lie in the sustained attention paradigms employed[ the SART sustained attention measure than on a con!
Typically\ continuous performance tests will require par! ventional sustained attention detection!based paradigm[
ticipants to monitor long sequences of stimuli and
respond on detecting infrequent targets[ Such paradigms
are\ arguably\ highly vulnerable to rapid automatization Hypothesis 2
in the sense of Schneider and Shi}rin|s distinction
between automatic and controlled\ e}ortful processing We predicted that\ within the TBI group\ pathology
ð20Ł[ Certainly\ such tasks have problems with ceiling severity measured by Glasgow Coma Scale "GCS# scores
e}ects\ which have led researchers to perceptually and post!traumatic amnesia duration "PTA# would be
 I[ H[ Robertson et al[:Everyday attentional failures 638

strongly related to SART!assessed sustained attention tive of the response being triggered by the anticipation
performance[ of the stimuli rather than as a result of an evaluation
of its relevance to response[

Hypothesis 3
Experiment 0
Traumatic brain injury and frontal lobe damage are
often associated with impoverished awareness of the Relationship between SART measures and everyday
extent of problems ð06\ 26Ł[ A number of factors may attentional lapses and other {co`nitive failures| amon`
contribute towards this\ including reduced sensitivity to normal controls
feedback and reduced attention to errors ð09Ł[ It was
therefore predicted that\ whereas SART measures may The purpose of this study was to test the hypothesis
show a relationship with self!reported attentional failures that SART performance would correlate with everyday
in a brain!injured group\ the reports of informants who attentional failures in a normal population[
are familiar with the patient may show the strongest
predictive relationship[
Method

Subjects[ A group of 64 control subjects "12 male\ 41

Hypothesis 4
female#\ ranging in age from 07 to 54 "mean 23[9^ S[D[
00[9#\ were recruited from the MRC Applied Psychology
Performance on the SART clearly requires the ability
Unit Subject Panel[
to withhold a response[ Response inhibition in classic
Procedure[ Subjects were assessed in a 0!hr session[
go:no!go paradigms has been shown to be impaired par!
Apparatus and materials[ The following tests were
ticularly after medial frontal lesions ð02Ł\ and given the
given]
likely location of damage in traumatic brain injury\ such
a task is liable to be sensitive to subtle e}ects of damage[
However\ we have argued that this continuous per!
Sustained attention to response test "SART#
formance task will be sensitive to the ability to endogen!
ously sustain attention[ Arbitrating between the relative
Reliability was tested by administering the procedure
contributions of an ine.ciency in response inhibition per
to a sub!group of 14 normal subjects ð04 women and 09
se and a failure to inhibit responses due to a lack of
men\ mean age 25[9 "S[D[ 7[9#^ mean SART errors 3[45
continuous attention to response is\ of course\ di.cult
"S[D[ 3[77#Ł on two occasions over a period of 0 week[
and indeed somewhat circular within this task[ However\
The Pearson correlation in the error score "false presses#
assuming that these phenomena are separable and that
between these two occasions was 9[65\ showing that per!
both may contribute to a poor performance\ it is possible
formance on this test is stable over time[
to draw support for the claim that SART is sensitive to
In the SART procedure\ 114 single digits "14 of each
sustained attention by predicting]
of the nine digits# were presented visually over a 3[2!min
0[ The SART measures would show a stronger relation! period[ Each digit was presented for 149 msec\ followed
ship to other validated measures of sustained attention by a 899!msec mask[ Subjects responded with a key press
"which have no such obvious response inhibition to each digit\ except 14 occasions when the digit 2
characteristics# than to measures of other attentional appeared\ when they had to withhold a response[ Subjects
capacities "including one with an arguably strong used their preferred hand[ The target digit was distributed
response inhibitory component#[ throughout the 114 trials in a pre!_xed quasi!random
1[ The occurrence of errors in the task can be predicted fashion[ The period from digit onset to digit onset was
by monitoring ~uctuations in the timing of accurate 0049 msec[ Subjects were asked to give equal importance
performance[ In other words\ an error can be seen not to accuracy and speed in doing the task[
simply as an isolated failure in withholding a response The digits were presented in one of _ve randomly allo!
but as the consequence of a failure in maintaining an cated font sizes to enhance the demands for processing
optimum approach to the task over time[ The SART the numerical value\ rather than simply setting a search
stimuli are highly predictable and rhythmic\ which template for some peripheral feature of the no!response
allows the frequent responses to non!targets to target[ These font sizes were 37 point\ 61 point\ 83 point\
become automatic and attentionally undemanding[ 099 point and 019 point\ respectively "Symbol font#\ cor!
E}ective sustained attention to the task would act to responding to a height varying between 01 mm and
counter this e}ect so that the response to an infrequent 18 mm[
stimulus could be withheld[ We would propose\ there! The mask following each digit consisted of a ring with
fore\ that an absence of such attention would be a diagonal cross in the middle[ The total diameter of the
revealed in a speeding of responses to stimuli\ sugges! circular mask was 18 mm[ Both digits and mask were
649 I[ H[ Robertson et al[:Everyday attentional failures

presented centrally in white against a black background Results

of the computer screen[ The screen "104 mm×024 mm]
Macintosh 069 PowerBook# was approximately 39 cm Hypothesis 0[ It was hypothesized that there would
from the subjects| eyes\ although no restrictions were be a signi_cant positive correlation in normals between
placed on the subjects| movements[ SART errors and reports "by both self and independent
Each session was preceded by a practice period con! informants# of attentional and other cognitive errors in
sisting of 07 presentations of digits\ two of which were everyday life[ There should be no such correlation
targets[ between a conventional measure of sustained attention*
the Triplets Test*and these questionnaire measures[
Table 0 shows a correlation matrix showing the inter!
Triplets test relationships of the above measures\ as well as their
associations with age and intelligence[ Only 59 out of the
One other experimental procedure was included for the 64 subjects completed CFQ questionnaires\ whereas 47
purposes of the present study[ This was a more con! out of 59 supplied informant questionnaires[
ventional continuous!performance!type test\ similar in Table 0 shows that\ in the normal sample\ the SART
form to the SART\ but requiring a response to infrequent measures were not sensitive to the e}ects of age or of
targets rather than a response to frequent non!targets di}erences in estimated intelligence[ There were sig!
and no response to a target[ Two hundred and twenty! ni_cant correlations between the SART and both self!
_ve digits were visually presented at an identical pacing and informant!reported cognitive failures as measured
to that used in the SART described above[ In the Triplets by the CFQ questionnaires for self and informants[ There
test\ however\ subjects had to respond whenever they were no signi_cant correlations between the Triplets Test
detected consecutive upward or downward runs of three and the CFQ measures[
digits*for example\ 4\ 5\ 6 or 3\ 2\ 1[ They responded to Hypothesis 0 was therefore supported for the SART\
these stimuli with a mouse key press[ As in the SART\ in that performance on the SART predicted self! and
there were 14 targets\ and the duration of the task was informant!reports of everyday attentional failures on the
also 3[2 min[ CFQ[ The results also indicate that the attempt to create
a measure that was sensitive to these forms of problem
but not to general intellectual level was\ at least for this
National adult reading test ð19Ł sample\ successful[

This reading test of irregularly spelled words gives an Experiment 1

estimate of intelligence[
Relationship between everyday attention failures\ SART
performance and brain dama`e severity amon` people with
Questionnaire measures of attentional failures in everyday traumatic brain injury
life
The aims of this experiment were to test hypotheses 1Ð4
Cognitive failures questionnaire ð2Ł[ This self!report as outlined above[ Each hypothesis is summarized below
questionnaire measures slips of action and of memory in prior to each analysis[
everyday life[
Co`nitive failures questionnaire for others ð2Ł[ This ques! Method
tionnaire is given to relatives or friends of the subject on
which they rate slips of action and of memory in everyday Subjects[ A consecutive sample of 23 traumatically
life[ brain injured patients who were between 8 and 07 months

Table 0[ Relationship between SART and Triplets performance and questionnaire self! and informant
reports of attentional failures in a group of normals "n59#
—
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
CFQ CFQ NART Triplets
self relative IQ Age "number correct#

SART −9[16 −9[18 n[s[ n[s[ 9[21

Triplets "number correct# n[s[ n[s[ n[s[ n[s[
NART IQ n[s[ n[s[ n[s[ n[s[
Age n[s[ n[s[
—
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
P³9[94[
CFQ self] Cognitive Failures Questionnaire ð2Ł[
CFQ relative] Cognitive Failures Questionnaire for others ð2Ł[
NART] National Adult Reading Test[
 I[ H[ Robertson et al[:Everyday attentional failures 640

post!injury and who had been admitted for at least 37 hr injury\ and one aim of the study was to compare the
to Addenbrooke|s hospital in Cambridge were assessed[ sensitivity to traumatic brain injury of this test to the
The following exclusions were made] response inhibition procedures[
0[ Resident outside of the East Anglia area[
1[ Pre!trauma history of epilepsy or other neurological
condition[
2[ History of drug or alcohol problems[ Triplets
3[ History of major psychiatric disorder[
4[ Reported hearing di.culties[ One other experimental procedure was included for the
purposes of the present study[ This was a more con!
The mean age of the sample was 23[7 "S[D[ 02[3#\ with ventional continuous!performance!type test\ similar in
13 males and 09 females[ The mean lowest Glasgow form to the SART\ but requiring a response to infrequent
Coma Scale "available for only 29 subjects# was 00[0 "S[D[ targets rather than a response to frequent non!targets
3[0#[ Post!traumatic amnesia duration "available for 21 and no response to a target[ Two hundred and twenty!
subjects# was used to classify subjects into the severity _ve digits were visually presented at an identical pacing
categories 0*mild "less than one hour#\ 1*moderate "0Ð to that used in the SART described above[ In the Triplets
13 hr#\ 2*severe "0Ð6 days#\ 3*very severe "6 Ð17 days# test\ however\ subjects had to respond whenever they
and extremely severe "more than 17 days#[ By this classi! detected consecutive upward or downward runs of three
_cation\ there were _ve mild\ six moderate\ _ve severe\ digits*for example 4\ 5\ 6 or 3\ 2\ 1[ They responded to
_ve very severe and 00 extremely severe cases\ respec! these stimuli with a mouse key press[ As in the SART\
tively[ The mean PASAT score "1!sec pacing# was 21[5 there were 14 targets\ and the duration of the task was
"S[D[ 00[6#\ and the mean number of categories obtained also 3[2 min[
on the Modi_ed Wisconsin Card Sorting Test "maximum
5# ð08Ł was 4[1 "S[D[ 0[4#[ They showed a mean total error
score of 6[6 "S[D[ 6[8# on this latter test\ of which a mean
of 06[1) "S[D[ 08[8# were perseverative[ On the Stroop
Test\ they showed a mean decrease in speed for the con! Measures of everyday attention failures
~ict over the control condition of 06[3 sec "S[D[ 6[2#[
Procedure[ Patients were assessed over two 1!hr The patient group and their relatives were also admin!
sessions\ having given informed consent to participating istered the same two rating scales "CFQ and CFQ for
in the study[ others# used with the normal group in Experiment 0
Apparatus[ All the measures\ with the exception of the above[ A total of 10 of the patients and relatives com!
National Adult Reading Test\ given to the controls in pleted these instruments[
Experiment 0 were also given to the subjects in the current
study[ In addition\ the following tests were given]

Comparison group for the brain injured subjects

SART "see Experiment 0 above#
Because the total sample of controls described in
Tests of sustained attention[ Experiment 0 above was not well matched in age\ sex and
Lottery subtest of the Test of Everyday Attention verbal intelligence with the brain injured sample\ for the
"TEA# ð16\ 17Ł[ purposes of testing Hypothesis 1\ a subset of both groups
Telephone Search with Counting Subtest of the TEA[ was selected so as to be matched on these variables[ The
Tests of attentional switchin`[ normal control group consisted of 06 subjects "six female\
Modi_ed Wisconsin Card Sorting Test ð08Ł[ 00 male^ mean age 28[7^ S[D[ 00[8^ mean percentile IQ
Visual Elevator Subtest of the TEA[ 70[6^ S[D[ 19[2# matched to a subsample of 11 of the
Tests of selective attention[ patient group "six female\ 05 male^ mean age 23[1^ S[D[
Stroop Test ð28Ł[ 01[0^ mean percentile IQ 66[5^ S[D[ 07[3# for age\ sex and
Telephone Search Test of the TEA[ estimated premorbid IQ[ The groups were compared on
the response inhibition and other measures[ Premorbid
In addition to the above test procedures\ the following intelligence was assessed either by the National Adult
were included] Reading Test ð19Ł or by the Spot!the!Word Test ð0Ł[ Per!
centile scores for estimated premorbid IQ were obtained\
and a patient subgroup was selected who matched the
Paced auditory serial addition test ð00\ 01Ł "1!sec pacing# control group on this as well as sex and age variables[
Matching was successful as there were no statistically
This test was included as it is one of the best established signi_cant di}erences between the two groups on age\
measures of attentional de_cit following traumatic brain sex!ratio or IQ[
641 I[ H[ Robertson et al[:Everyday attentional failures

Table 1[ Scores on SART and Triplets errors as well as SART reaction times\ for the patient and control
groups errors respectively
—
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
SART Triplets SART reaction time for the SART reaction time for the
errors omissions four presses prior to correctly four presses prior to
withheld respones non!withheld respones

Patients 6[5 "3[7# 2[8 "2[0# 261[7 "67[2# 299[3 "32[5#

Controls 3[9 "2[1# 4[9 "3[2# 286[0 "73[8# 294[8 "13[1#
F 6[9 1[6 9[60 9[03
P 9[90 n[s[ n[s[ n[s[
—––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

Results between Glasgow Coma Scale scores "GCS# and post!

traumatic amnesia duration on the one hand\ and atten!
Hypothesis 1] Do the SART measures distinguish con! tional measures on the other\ were calculated[ Figure
trols from brain injured subjects< Table 1 shows the scores 0 summarizes the signi_cant results[ GCS scores were
on the SART and Triplets errors[ available for only 29 subjects[
Table 1 shows that the SART signi_cantly dis! Figure 0 shows that the SART and PASAT scores
criminated between the two groups\ but the Triplets task were the best predictors of GCS scores among the tests
did not\ as was predicted in the introduction[ In fact\ the administered[ PTA grade was best predicted by PASAT
control group made slightly fewer errors on the Triplets and Triplets[ No other correlations were statistically sig!
task than did the patients[ ni_cant at the 4) level[
Hypothesis 2] Patholo`y severity measured by Glas`ow The hypothesis that SART measures would predict the
Coma Scale "GCS# scores and post!traumatic amnesia dur! severity of an injury is supported\ although the apparent
ation "PTA# will be stron`ly related to SART!assessed complexity of the relationships between di}erent cog!
sustained attention performance[ For the purposes of test! nitive tests and di}erent estimates of severity requires
ing this hypothesis\ the whole group of 23 brain injured further consideration[
patients was included in the analysis[ The correlations The relative contributions of PASAT and SART to

Fig[ 0[ Statistically signi_cant correlations between key attentional measures on the one hand\ and Glasgow Coma Scale scores and
Post Traumatic Amnesia durations\ respectively[
 I[ H[ Robertson et al[:Everyday attentional failures 642

Table 2[ Correlations between self!reports of attentional failures and test performance

"n10# "P³9[94^ $P³9[90#
—
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
INFORMANT
SELF Cognitive Cognitive Failures
failures "CQF# "CQF#

Tests of attentional switching

Wisconsin categories n[s[ n[s[
Visual elevator n[s[ −9[38
Selective attention tests
Stroop n[s[ 9[36
Telephone search "TEA# n[s[ n[s[
Sustained attention tests
Telephone search with counting "TEA# n[s[ n[s[
Lottery "TEA# n[s[ n[s[
SART
SART error n[s[ 9[33
Other tests
PASAT n[s[ −9[62$
Triplets n[s[ n[s[
GCS "Glasgow Coma Scale# n[s[ −9[40
PTA "post!traumatic amnesia duration# n[s[ 9[47$
—
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
P ³ 9[94[
$P ³ 9[90[

predicting GCS were therefore assessed using multiple in particular with tests where response inhibition is
regression[ PASAT and SART were entered as inde! important\ such as the Wisconsin Card Sorting Test\ the
pendent variables[ Only SART showed any signi_cant Visual Elevator Subtest of the Test of Everyday Attention
independent e}ect in the regression "t−1[92^ P9[94#[ and Stroop[
This was con_rmed by the fact that\ whereas the SART To test this hypothesis\ we carried out three stepwise
accounts for 22[6) of the variance in GCS on its own\ muliple regression analyses\ with SART as the dependent
adding in PASAT as an independent variable added only variable in each case[ Table 3 shows these three
a negligible 2) to the explained variance[ regressions[ In each case\ one of the three tests\ presumed
Hypothesis 3] Attentional test correlations ratin` scale to be sensitive to response inhibition\ was entered into the
measures of everyday problems of attention for brain regression _rst\ and then two tests of sustained attention
injured people[ Self ratings[ The relationship between "Lottery and Telephone Search with Counting Subtests
self! and informant!ratings of attentional failures on the of the Test of Everyday Attention# were each loaded in
CFQ scales was examined[ Table 2 shows the correlations turn into the regression to determine how much extra
between self!reports and informant!reports of attentional variance\ if any\ would be explained[
failures and test performance[ Rating scale data were In addition\ in each regression\ we also loaded in
available on only 10 of the brain injured subjects[ PASAT as the _nal independent variable\ in order to
Table 2 shows that zero out of 00 correlations between determine whether this sensitive yet complex benchmark
self!reports of attentional failures and attentional test test of TBI attention de_cits would add signi_cantly to
performance are statistically signi_cant\ in contrast to the explained variance in SART[
six out of 00 "43)# of informant!reports of attentional The _rst regression in Table 3 shows that Wisconsin
failures and test performance shown in Table 2[ perseverative errors are non!signi_cantly correlated with
Given the relatively low number of brain injured people SART[ The addition of the Lottery added almost 07)
on whom self and informant reports were available "10#\ of variance explained\ whereas the addition of Telephone
some caution is needed in drawing _rm conclusions from Search with Counting "TSC# added a further 01)\ giving
the correlational data[ However\ the results are certainly a total explained variance of 29)[ Adding PASAT\ how!
consistent with the proposal that\ due to problems with ever\ contributed a non!signi_cant extra 1) to the SART
insight and attention\ informants| reports would be more explained variance[
sensitive to cognitive problems experienced by the brain! The second regression in Table 3 showed near identical
injured people than self!reports[ results\ with the Visual Elevator "VE# test showing no
Hypothesis 4[ "a# Relationship of SART with other signi_cant relationship with SART^ yet the addition of
tests of attention[ We argued that SART is\ to a great two sustained attention tasks produced an explained vari!
extent\ a test of sustained attention\ and hence we predict ance of 21)[ Again the PASAT did not signi_cantly
a much stronger relationship between SART and tests of improve on this[
sustained attention than with other attentional tasks\ and The third regression produced similar results with
643 I[ H[ Robertson et al[:Everyday attentional failures

Table 3[ Multiple regressions of attentional measures on SART

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
—
SART Percentage
variance Additional P of added
R R1 explained variance variable

A[ Wisconsin 0[ WISC 9[95 9[992 9[2 * n[s[

1[ WISC¦Lottery 9[31 9[07 07 06[6 ³9[92
2[ WISC¦Lotter¦TSC 9[44 9[29 29 01 ³9[94
3[ WISC¦Lotter¦TSC¦PASAT 9[46 9[21 21 1 n[s[
B[ Visual elevator 0[ Visual elevator 9[02 9[91 1 * n[s[
1[ VE¦Lottery 9[31 9[07 07 05 ³9[92
2[ VE¦Lottery¦TSC 9[46 9[21 21 05 ³9[92
3[ VE¦Lottery¦TSC¦PASAT 9[59 9[25 25 3 n[s[
C[ Stroop 0[ Stroop "Increase in time of con~ict over no!con~ict
condition# 9[03 9[91 1 * n[s[
1[ Stroop¦Lottery 9[39 9[05 05 03 ³9[95
2[ Stroop¦Lottery¦TSC 9[41 9[16 16 00 ³9[95
3[ Stroop¦Lottery¦TSC¦PASAT 9[43 9[18 18 1 n[s[
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
—

Stroop\ which\ like Wisconsin and VE\ did not correlate preted in this task\ where the occurrence of stimuli is
signi_cantly with SART[ Again\ the explained variance highly predictable\ as indicating a lessening of active
only became signi_cant when sustained attention tasks attention[
were added\ and again the PASAT added no signi_cant There is also some indication of a group di}erence on
extra variance[ the e}ect of an error on response characteristics[ The
There were no statistically signi_cant correlations with control group showed a signi_cant increase in RT
any of the other attentional tests with the exception of between the four trials leading up to an error and the
Triplets*another measure of sustained attention* four trials following an error ðmean RT prior to error^
which correlated at a marginally signi_cant level\ 9[23 217[001 "S[D[ 42[3#\ mean RT following error 251[671
"P³9[95# with SART[ The hypothesis that SART is sen! "S[D[ 73[2#\ t−1[046\ P³9[94Ł\ suggesting clear e}ects
sitive to sustained attention and not simply to impaired on response style[ The patients tended not to show or
ability to inhibit a response per se is therefore supported[ maintain such error e}ects on response style ðmean RT
prior to error^ 234[0 "S[D[ 48[4#\ mean RT following error
237[8 "S[D[ 67[0#\ t−9[25^ n[s[Ł[
Predicting errors in SART on the basis of the timing of Figure 1 shows graphically the mean reaction times for
accurate responses the four responses preceding\ and the four responses
after\ correct and error trials\ respectively[ Figure 1 shows
In order to test whether an error on the SART these data for\ respectively\ the control group\ for the
"responding to a target# could be predicted on the basis TBI group whose error rate was less than 0 standard
of performance characteristics\ which may re~ect a less! deviation from the control group|s mean errors and for
ening of attention to the task\ we carried out the following the TBI group\ who made in excess of two standard
analysis[ Reaction times for each set of four correct deviations from the control group|s errors[
presses prior to correctly inhibited targets "i[e[ presenta! A disproportionate variability in RTs\ measured by
tions of the number 2\ which did not result in a response# within!subject variability\ has been noted in TBI patients\
were compared with each set of four correct presses prior interpreted as indicating a de_cit in sustaining consistent
to a mistakenly pressed target "i[e[ presentations of the performance ð27Ł[ In the present study\ analysis of vari!
number 2\ which did result in a response# in the patient ance revealed a statistically signi_cant e}ect of group
group[ The mean reaction time prior to correctly given on RTs "F11[48\ P³9[9990#\ with the TBI subjects
responses was 279[5 "S[D[ 53[3#\ whereas the mean reac! showing a greater variability "mean standard devi!
tion time in the trials prior to mistaken presses was 234[0 ation88[2\ S[D[ 33[7# than the control group "mean
"S[D[ 48[4#[ This was a statistically signi_cant di}erence standard deviation56[8\ S[D[ 08[2#[
"t−2[214^ P³9[90#[
A similar _nding was obtained for the controls[ The
mean pre!correct!trial RT for controls was 265[8 "S[D[
Discussion
43[6#\ whereas the pre!false!press mean RT was 217[0
"S[D[ 42[4#[ This di}erence remained statistically sig!
All _ve hypotheses received support from the data[ To
ni_cant "t−2[54^ P³9[914# after correction of sig!
summarize]
ni_cance level for multiple t!tests[ For both groups then\
errors may be predicted by a reduction in RTs\ inter! 0[ In normal controls\ SART performance signi_cantly
 I[ H[ Robertson et al[:Everyday attentional failures 644

Fig[ 1[ SART] Mean RTs for the four trials preceding and following correct or incorrect responses to target presentations for three
groups[

correlated with self!reports of attentional and other account of poor performance on this task than a sim!
{cognitive failures| in everyday life\ as well as with ple decrement over time[ These results support the
informant reports of such failures[ hypothesis that di.culty in maintaining continuous
1[ SART performance discriminated an unselected sam! attention to the task provides a more satisfactory
ple of brain injured subjects from age!\ sex! and IQ! account for failure than a simple di.culty in inhibiting
matched controls\ whereas a more conventional per! responses[
ceptually!based vigilance task "Triplets# did not[
2[ SART forms\ along with the PASAT\ the best pre! The hypotheses set out in the introduction were therefore
dictor of severity of brain damage as measured by broadly supported[ The SART appears to be sensitive to
lowest Glasgow Coma Scale scores of all the cognitive sustained attention de_cits and predicts self!reported and
measures administered[ Coma severity was the prin! informant!reported attentional failures in normals\ and
cipal determinant of poor SART performance[ informant!reported attentional failures in brain injured
3[ SART\ along with several other attentional measures\ participants[ Performance on the SART measures sig!
was strongly correlated with informant reports of ni_cantly discriminated an unselected sample of brain
daily life attentional failures in the TBI group[ No injured patients "with a wide variety of severity and post!
attentional measures were correlated with self! injury symptoms# from normal age! and premorbid IQ!
reported problems with attention in this group[ matched controls[ SART measures were as e}ective as
4[ Variance in SART performance was predicted by sus! the PASAT in predicting some measures of severity of
tained attention test performance and not by per! injury[
formance on tests presumed to be sensitive to response The fact that Triplets did not discriminate between the
inhibition[ Errors on the SART measure were pre! TBI patients and normal controls\ whereas the SART
dicted from participants| performance on correct non! did\ may be due to the greater sensitivity and lower auto!
target items preceding the occurrence of a target] sub! matizability of the SART\ as argued in the introduction[
jects show signi_cant speeding up of responding prior Though the Triplets is not as simple as some vigilance
to error responses[ TBI patients also show a sig! tasks where only single stimuli have to be detected\ we
ni_cantly reduced tendency to slow down responding argue that numerical sequences such as 2\ 3\ 4 are
after an error compared to the controls[ TBI patients su.ciently familiar that their detection does indeed
showed a signi_cantly greater variability in RTs to require less sustained attention to task than the require!
stimuli compared to controls[ As no signi_cant time! ment to inhibit a response in SART[ We do\ however\
on!task e}ects emerged for either group in terms of acknowledge the possibility that this _nding may be due
errors or RTs\ this _nding suggests that local ~uc! to the response inhibition aspects of the SART\ although
tuations in attention or {lapses| may provide a better the fact that SART correlates uniquely with sustained
645 I[ H[ Robertson et al[:Everyday attentional failures

attention and not other measures lends support to the ticular demands that the SART makes on the ability to
former argument[ sustain attention to response therefore may underlie this
As a relatively unselected group of TBI subjects of association[ PASAT is arguably a more complex task
mixed severity\ the fact that they were broadly within demanding greater cortical involvement and hence may
normal limits on conventional measures\ such as Wiscon! be less strongly linked to white matter damage and less
sin\ is perhaps not surprising[ As a group of above!aver! speci_cally related to coma severity[
age IQ\ this fact may have served to obscure executive Whereas it has been proposed that the SART is sen!
de_cits[ Attentional and frontal de_cits can only be sitive to variation in the ability to endogenously sustain
detected with di.culty with sophisticated experimental attention to task\ as outlined in the introduction\ a strong
methods in some studies with unselected populations ð30\ counter claim is that the test is sensitive to the ability to
31Ł\ and the relative insensitivity of these measures for inhibit a response\ a known impairment following frontal
such populations may explain\ in part\ the superior per! lesions\ and that additional accounts are redundant[
formance of SART in the present study[ Indeed\ impairments of the response selection stage of
The _nding that observed "by a close informant# atten! information processing following closed head injury have
tional slips in everyday life can be predicted by laboratory been reported in several Australian studies ð07\ 24\ 25Ł[
test performance is the _rst such _nding in the literature\ Simply suggesting that di.culties in inhibiting a
to our knowledge[ This result suggests a normal con! response\ whether in cognitive tests or in real life\ are the
tinuum of sustained attention capacity\ bearing strongly result of an absence of sustained attention to task moves
on everyday life performance[ This allows us to consider this no further[ One way of delineating the relative con!
the problems shown by TBI subjects in a similar light to tributions to performance would be to consider further
those shown by a proportion of the normal population[ the e}ects of time on task[ This has been the traditional
The results also emphasize the need\ in brain injured method in the search for sustained attention de_cits in
groups\ to consider the reports of informants who know this group ð34Ł[
the patient well and not simply to rely on self!reports in For reasons of producing a clinically useful measure\
considering cognitive di.culties[ the SART used in this study does not provide su.cient
The SART measures\ while being sensitive to vari! targets to perform this type of analysis reasonably in
ations in attentional performance within brain injured terms of error rates[ Whereas this is amenable to further
and normal populations\ also act as a powerful dis! experimental investigation\ it is not clear that such a view
criminator of group[ This _nding suggests that the mea! of an incremental decline over time is the best _t for the
sures may be a useful addition in clinical assessment\ in complaints of patients[ Such a design would not\ for
both predicting real life di.culties and in supporting example\ be sensitive to a pattern of attentional ~uc!
victim and family claims that an injury has led to impair! tuation\ of drifting o} and on task\ which may occur over
ment and disability[ periods of just a few seconds ð12\ 33Ł[
Other results also support the use of such measures in Another route to disambiguating the factors under!
this capacity[ Currently\ the PASAT is the key instrument lying SART failure is to consider its relationship to other
that is sensitive to the sometimes subtle processing tests[ It has been demonstrated that the SART shows
impairments\ which can result from traumatic brain stronger relationships with measures of sustained atten!
injury[ However\ as discussed\ this sensitivity must be tion than with other types of attention[ What we have
somewhat set against the di.culty in interpreting per! been unable to demonstrate is that it sits better with tests
formance due to the signi_cant contributions of arith! of sustained attention than it does with a {pure| measure
metical ability\ age and general intellectual resources ð1\ of response inhibition[ A problem with doing this\ from
4\ 7Ł\ not to mention the rather intimidating qualities of our perspective\ is that it is di.cult to conceive of such
the task[ At least in the samples tested\ the SART was measures that are not themselves vulnerable to a sus!
resistant to di}erences in age and estimates of IQ[ Con! tained attention to response argument or that are not
ceptually\ it is an easy task to pick up\ it has little in the contaminated with extraneous demands[
way of a memory load "there is only one target to keep A third source of evidence\ that we have suggested is
in mind# and it only requires identi_cation of single digits[ of relevance to this question\ is in considering the RTs to
It seems likely that another bene_t of the simplicity of non!targets that precede and follow the occurrence of
the SART will be its amenability to _ne!grained analysis targets[ It was proposed that\ because of the task charac!
through the manipulation of its few parameters[ teristics of simplicity\ rhythmicity and predictability\ that
Whereas the SART appears to be as strong a predictor waning attention to response would be characterized by
of some aspects of injury severity as PASAT\ SART was a speeding of RT to stimuli[ Subjectively\ setting up such
associated only with coma severity\ but not with post! a response pattern seems to be the least e}ortful and most
traumatic amnesia duration\ whereas PASAT was associ! errorful way of performing the task[ Certainly errors were
ated with both[ In considering these relationships\ it is of predictable by considering this factor alone[ It has also
note that coma severity\ as assessed by GCS\ is associated been claimed that the return to longer RTs that follow
with white matter damage ð36Ł\ which\ in turn\ has been errors\ at least in normal participants\ corresponds with
associated with sustained attention de_cits ð27Ł[ The par! a return of e}ortful sustained attention to the task[
 I[ H[ Robertson et al[:Everyday attentional failures 646

The slowing of response times following errors also sustained auditory discrimination[ Experimental
may arise because subjects adopt a more conservative Brain Research 81\ 054Ð061\ 0881[
response criterion\ or because they inhibit automatic 7[ Deary\ I[ J[\ Langan\ S[ J[\ Hepburn\ D[ A[ and
responses to a greater degree once an error has been Frier\ B[ M[ Which abilities does the PASAT test<
made[ Thus\ on the basis of the reaction time "RT# evi! Personality and Individual Differences 01\ 872Ð876\
0880[
dence alone\ one could not argue in favour of the sus!
8[ Fisk\ A[ D[ and Schneider\ W[ Controlled and auto!
tained attention hypothesis[ Taken together with the matic processing during tasks requiring sustained
regression analysis data from Table 3\ however\ the sus! attention] A new approach to vigilance[ Human Fac!
tained attention hypothesis explanation of the RT data tors 12\ 626Ð649\ 0870[
becomes slightly stronger\ although certainly not conclus! 09[ Freedman\ P[ E[\ Bleiberg\ J[ and Freedland\ K[
ive[ Anticipatory behaviour de_cits in closed head injury[
The convergence of evidence lends support to the view Journal of Neurolo`y\ Neurosur`ery\ and Psychiatry
that the SART may be particularly sensitive to the ability 49\ 287Ð390\ 0876[
to sustain attention to a dull but demanding task\ but 00[ Gronwall\ D[ M[ A[ and Wrightson\ P[ Delayed
is insu.cient to fully dismiss alternative accounts[ An recovery of intellectual function after minor head
advantage of the test|s sensitivity to variations in normal injury[ Lancet\ II "6763#\ 0341\ 0863[
01[ Gronwall\ D[ M[ A[ and Wrightson\ P[ Cumulative
populations is that such questions can be further pursued
e}ects of concussion[ Lancet 00\ 884Ð886\ 0864[
in such groups[ 02[ Leimkuhler\ M[ E[ and Mesulam\ M[ M[ Reversible
In summary\ we believe that we have developed a task goÐno go de_cits in a case of frontal lobe tumor[
that is sensitive to attentional de_cits in traumatic brain Annals of Neurolo`y 07\ 506Ð508\ 0874[
injury patients and that also may be sensitive to indi! 03[ Levin\ H[ S[\ High\ W[ M[\ Williams\ D[ H[\ Eisen!
vidual di}erences in everyday attention failures in normal berg\ H[ M[\ Amparo\ E[ G[\ Guinto\ F[ C[ and
controls[ The simplicity of the paradigm that shows such Ewert\ J[ Dichotic listening and manual performance
strong correlations with biological markers of severity in relation to magnetic resonance imaging after
of damage and everyday life performance means that closed head injury[ Journal of Neurolo`y\ Neuro!
considerable strides can be made in the future towards sur`ery\ and Psychiatry 41\ 0051Ð0058\ 0878[
further delineating the nature of the de_cits and ulti! 04[ Loken\ W[ L[\ Thornton\ A[ E[\ Otto\ R[ L[ and
mately providing adequate rehabilitation of these de_cits[ Long\ C[ L[ Sustained attention after closed head
injury[ Neuropsycholo`y 8\ 481Ð487\ 0884[
05[ Mackworth\ F[ J[ Vigilance\ arousal and habituation[
Psycholo`ical Review 64\ 297Ð211\ 0857[
06[ McKinlay\ W[ M[ The short!term outcome of severe
References blunt head injury as reported by relatives of the
injured persons[ Journal of Neurolo`y\ Neurosur`ery\
0[ Baddeley\ A[ D[\ Emslie\ H[ and Nimmo!Smith\ I[\ and Psychiatry 33\ 416Ð422\ 0870[
The Speed and Capacity of Lan`ua`e Processin` Test[ 07[ Murray\ R[\ Shum\ D[ K[ and McFarland\ K[ Atten!
Thames Valley Test Company\ Bury St Edmunds\ tional de_cits in head!injured children] An infor!
U[K[\ 0881[ mation!processing analysis[ Brain and Co`nition 07\
1[ Brittain\ J[ L[\ LaMarche\ J[ A[\ Reeder\ K[ P[\ Roth\ 88Ð004\ 0881[
D[ L[ and Boll\ T[ J[ E}ects of age and IQ on paced 08[ Nelson\ H[ A modi_ed card sorting test sensitive to
auditory serial addition task "PASAT# performance[ frontal lobe de_cits[ Cortex 01\ 202Ð213\ 0865[
The Clinical Neuropsycholo`ist 4\ 052Ð064\ 0880[ 19[ Nelson\ H[\ The National Adult Readin` Test "NART#
2[ Broadbent\ D[ B[\ Cooper\ P[ F[\ FitzGerald\ P[ and Test[ NFER\ Windsor\ 0871[
Parkes\ K[ R[ The Cognitive Failures Questionnaire 10[ Norman\ D[ A[ and Shallice\ T[\ Attention to action]
"CFQ# and its correlates[ British Journal of Clinical willed and automatic control of behaviour[ In Con!
Psycholo`y 10\ 0Ð05\ 0871[ sciousness and Self Re`ulation\ Vol[ 3\ ed[ R[ J[ Dav!
3[ Brouwer\ W[ H[ and Wol}elaar\ P[ C[ V[ Sustained idson\ G[ E[ Schwartz and D[ E[ Shapiro[ Plenum\
attention and sustained e}ort after closed head New York\ 0875\ pp[ 0Ð07[
injury] Detection and 9[09 Hz heart rate variability 11[ Parasuraman\ R[\ Mutter\ S[ A[ and Molloy\ R[ Sus!
in a low event rate vigilance task[ Cortex 10\ 000Ð tained attention following mild closed!head injury[
008\ 0874[ Journal of Clinical and Experimental Neuro!
4[ Chronicle\ E[ P[ and MacGregor\ N[ A[\ Are PASAT psycholo`y 02\ 678Ð700\ 0880[
scores related to mathematical ability< Neuro! 12[ Pardo\ J[ V[\ Fox\ P[ T[ and Raichle\ M[ E[ Local!
psycholo`ical Rehabilitation\ in press[ ization of a human system for sustained attention by
5[ Cohen\ R[ M[\ Semple\ W[ E[\ Gross\ M[\ Holcomb\ positron emission tomography[ Nature 238\ 50Ð53\
H[ J[\ Dowling\ S[ and Nordahl\ T[ E[ Functional 0880[
localization of sustained attention[ Neuropsychiatry\ 13[ Posner\ M[ I[ and Peterson\ S[ E[ The attention sys!
Neuropsycholo`y and Behavioural Neurolo`y 0\ 2Ð19\ tem of the human brain[ Annual Review of Neuro!
0877[ science 02\ 14Ð31\ 0889[
6[ Cohen\ R[ M[\ Semple\ W[ E[\ Gross\ M[\ King\ A[ 14[ Rabbitt\ P[ and Abson\ V[ {Lost and Found|] Some
C[ and Nordahl\ T[ E[ Metabolic brain pattern of logical and methodological limitations of self!report
647 I[ H[ Robertson et al[:Everyday attentional failures

questionnaires as tools to study cognitive ageing[ attentional processes] An information processing

British Journal of Psycholo`y 70\ 0Ð05\ 0889[ stage analysis[ Journal of Clinical and Experimental
15[ Rao\ S[ M[\ Leo\ G[ J[ and Haughton\ V[ M[\ Aubin! Neuropsycholo`y 01\ 136Ð153\ 0889[
Faubert\ P[ S[ and Bernardin\ L[ Correlation of mag! 26[ Stuss\ D[ and Benson\ D[ F[ Neuropsychological
netic resonance imaging with neuropsychological studies of the frontal lobes[ Psycholo`ical Bulletin 84\
testing in multiple sclerosis[ Neurolo`y 28\ 050Ð055\ 2Ð17\ 0872[
0878[ 27[ Stuss\ D[ and Gow\ A[ D[ Frontal dysfunction after
16[ Robertson\ I[ H[\ Ward\ A[\ Ridgeway\ V[ and traumatic brain injury[ Neuropsychiatry\ Neuro!
Nimmo!Smith\ I[\ Test of Everyday Attention[ psycholo`y and Behavior Neurolo`y 4\ 161Ð171\ 0881[
Thames Valley Test Company\ Bury St Edmunds\ 28[ Trenerry\ M[ R[\ Crosson\ B[\ DeBoe\ J[ and Leber\
U[K[\ 0883[ W[ R[\ Stroop Neuropsycholo`ical Screenin` Test[
17[ Robertson\ I[ H[\ Ward\ T[\ Ridgeway\ V[ and Psychological Assessment Resources\ Odessa\ FL\
Nimmo!Smith\ I[\ The structure of normal human 0878[
attention] The Test of Everyday Attention[ Journal of 39[ VanZomeren\ A[ H[ and Burg\ W[ V[ D[ Residual
the International Neuropsycholo`ical Society 1\ 414Ð complaints of patients two years after severe closed
423\ 0885[ head injury[ Journal of Neurolo`y\ Neurosur`ery\ and
18[ Rueckert\ L[ and Grafman\ J[ Sustained attention Psychiatry 37\ 10Ð17\ 0874[
de_cits in patients with right frontal lesions[ Neuro! 30[ VanZomeren\ A[ H[ and Deelman\ B[ G[ Di}erential
psycholo`ia 23\ 842Ð852\ 0885[ e}ects of simple and choice reaction after closed head
29[ Rueckert\ L[\ Sorensen\ L[ and Levy\ J[ Callosal injury[ Clinical Neurolo`y and Neurosur`ery 68\ 70Ð
e.ciency is related to sustained attention[ Neuro! 89\ 0865[
psycholo`ia 21\ 048Ð062\ 0883[ 31[ VanZomeren\ A[ H[ and Deelman\ B[ G[ Long term
20[ Schneider\ W[ and Shi}rin\ R[ M[ Controlled and recovery of visual reaction time after closed head
automatic human information processing] 0[ Detec! injury[ Journal of Neurolo`y\ Neurosur`ery\ and Psy!
tion\ search and attention[ Psycholo`ical Review 73\ chiatry 37\ 10Ð17\ 0867[
0Ð55\ 0866[ 32[ Varney\ N[ R[\ Bushnell\ D[ L[\ Nathan\ M[\ Kahn\
21[ Schwartz\ M[ F[\ Mayer\ N[ H[\ FitzpatrickDeSalme\ D[\ Roberts\ R[\ Rezai\ K[\ Walker\ W[ and Kirchner\
E[ J[ and Montgomery\ M[ W[ Cognitive theory and P[ NeuroSPECT correlates of disabling mild head
the study of everyday action disorders after brain injury] Preliminary _ndings[ Journal of Head Trauma
damage[ Journal of Head Trauma Rehabilitation 7\ Rehabilitation 09\ 07Ð17\ 0884[
48Ð61\ 0882[ 33[ Whitehead\ R[ Right hemisphere processing superi!
22[ Shallice\ T[ and Burgess\ P[ De_cit in strategy appli! ority during sustained visual attention[ Journal of
cation following frontal lobe damage in man[ Brain Co`nitive Neuroscience 2\ 218Ð224\ 0880[
003\ 616Ð630\ 0880[ 34[ Whyte\ J[\ Polansky\ M[\ Fleming\ M[\ Coslett\ H[ B[
23[ Shallice\ T[ and Burgess\ P[\ Higher!order cognitive and Cavallucci\ C[ Sustained arousal and attention
impairment and frontal lobe lesions in man[ In Fron! after traumatic brain injury[ Neuropsycholo`ia 22\
tal Lobe Function and Dysfunction\ ed[ H[ S[ Levin\ H[ 686Ð702\ 0884[
M[ Eisenberg and A[ L[ Benton\ Oxford University 35[ Wilkins\ A[ J[\ Shallice\ T[ and McCarthy\ R[ Frontal
Press\ New York\ 0880\ pp[ 014Ð027[ lesions and sustained attention[ Neuropsycholo`ia 14\
24[ Shum\ D[ H[ K[\ McFarland\ K[ and Bain\ J[ D[ 248Ð254\ 0876[
E}ects of closed head injury on attentional processes] 36[ Wilson\ J[ T[ L[\ Hadley\ D[ M[\ Wiedmann\ K[ D[
Generality of Sternberg|s additive factor method[ and Teasdale\ G[ M[ Neuropsychological conse!
Journal of Clinical and Experimental Neuro! quences of two patterns of brain damage shown by
psycholo`y 05\ 436Ð444\ 0883[ MRI in survivors of severe head injury[ Journal of
25[ Shum\ D[ H[ K[\ McFarland\ K[\ Bain\ J[ D[ and Neurolo`y\ Neurosur`ery\ and Psychiatry 48\ 217Ð
Humphreys\ M[ S[ E}ects of closed head injury on 220\ 0884[