h'wopsyholoy,u. Vol.30,No Printedm Great Britain
2. pp. 1235132.1992
c
002X-3932.92SS.OO+O.OO 1992 Pergamon Pressplc
COVERT ORIENTATION OF VISUAL ATTENTION CLOSED HEAD INJURY
AFTER
S. L. CREMONA-METEYARD, C. R. CLARK, M. J. WRIGHT and G. M. GEFFEN* Cognitive
Psychophysiology
Laboratory, (Receiced
Psychology
1 Auyust
Department,
1990: accepted
26
University
of Queensland,
Australia
4072
September 1991)
Abstract-The ability to orient visual attention covertly was studied in 11 patients who had suffered a moderate or severe closed head injury (CHI) at least 1 year previously. Their performance was compared to nine matched controls using a cued reaction time (RT) task which measured the RT benefit of valid directional cueing and the RT cost of miscueing. The CHI and control groups did not differ in overall RT. Relative to controls, the CHI group showed normal cost but hardly any benefit, indicating that the normal capacity to pre-align attention with a cued location was impaired.
DISORDERS OF ATTENTION are a common outcome of closed head injury (CHI) and may underlie the characteristic problems of learning, memory and problem solving that are frequently reported after significant head injuries [ 11. This study aimed to identify residual deficits in the ability to shift and focus attention covertly following CHI. Since we were interested in identifying persistent deficits, our subjects were tested l-5 years post injury when their recovery may be regarded as stable. Previous studies of attention have consistently identified slower information processing in subjects who have sustained CHI (e.g. Refs Cl], [21] and [22]). Most of these studies have used relatively complex tasks and found deficits in the decision-making phase of information processing. In a recent study, Rucc et al. [ 1S] found that response time on a simple RT task with auditory cues and imperative signals did not differentiate the CHI group from controls. Other studies of simple and choice visual RT have shown that head-injured subjects were slower than controls with the extent of the difference increased with the number of choices or complexity of the task [ 1,9, 17,223. Severe CHI subjects tested less than 1 year post injury showed an impairment in both the early identification and the response selection stages of processing in a visuo-spatial task [ 191. The early perceptual deficit appears to diminish with time, as patients tested more than 1 year post injury showed an impairment in only the response selection phase [ 193. The present study used a visual cued RT task involving covert orientation of attention [ 12, 143. This task is particularly useful for studying head-injured subjects as it requires one simple response yet is sensitive to covert changes of orientation of attention that are independent of muscular movement of the eyes or head [ 161. Expectancy is manipulated by
*Address Queensland
for reprints: Professor 4067, Australia.
G. Geffen,
Psychology
123
Department,
University
of Queensland,
St Lucia,
124
s. L. CKl.hlOL.A-MI
71 YAKI)
c’f cl/.
providing directional cuts that are either valid or not valid in informing the subject of whether a target will occur in the left or right visual field 1121. The benefits (in speed of response to the target) of directional engagement of attention and costs of subsequent shifting of attention are measured in relation to RT to a target after a neutral cue that provides no directional information about target location. Covert orienting involves the directional engagement of attention. whereas reorienting after an invalid CLK requires disengagement, shifting, and re-engagement of attention 1161. This model of attention is based on a spotlight metaphor, with the locus of attention being highlighted. While the metaphor has been criticized [3.4], empirical testing of its predictions has been successful in cases with circumscribed lesions [ 16. 241. Hence in this study the task was given to casts with more variable lesions and the extent to which the model can account for findings was examined. Brain lesion studies [I31 and single cell recording in alert monkeys [IO, I I. 251 have identified particuiar anatomical locations associated with a number of the mental operations involved in the covert orientation of visual attention. Specifically. the posterior parietal lobe has been linked to the disengagement of attention, and areas of the midbrain and thalamus to the covert movement and engagement of attention, respectively. Consequently, injury to these three brain areas causes deficits of different natures. Patients with unilateral parietal lesions show increased RTs to targets that are contralateral to the lesion and occur after an invalid cue (i.e. the side ipsilateral to the lesion was cued) 1161. This is consistent with a specific deficit in the patient’s ability to disengage from the current focus of attention and move in a direction contralateral to the lesion. A disengage deficit has also been reported in a patient with a right frontal lobe lesion 1151. suggccting that the frontal areas may also play an important role in the covert orientation of attention. In contrast, damage to the midbrain increases overall RT and lengthens the time taken to establish an efficient RT at the cued location. Damage to the thalamus produces a pattern of slowed RT to both cued and uncued targets on the side opposite the Icsion, suggesting an engage deficit. A similar deficit has been found in monkeys performing the task when chemical injections disrupt the performance of visual parts of the pulkinar [I !. IS]. Normal subjects injected with droperidol and clonidine to suppress central dopamine and noradrenaline transmission showed reductions in the cost of invalid cueing without a change in the benefit of valid cueing [2]. A similar result was found in patients with Parkinson’s disease 1243. From the results outlined above it is apparent that the simple act of shifting attention to a cued location is a complex operation involving a number of different anatomical locations. Several distinct computations must be orchestrated to allow the shifting of attention to a cued location. Given the variable and difluse damage resulting from head injury an impairment in the covert orienting of attention is expected, but the exact nature of the deficit is ditlicult to predict. While the CH! subjects differ in location and size of lesions. they may be regarded as homogeneous for a number of reasons that warrant studying them as a group. They all share the same etiology. they are mostly young males and they generally display similar deficits in terms of information processing. Common sites of injury are the frontal lobes and temporal poles. However. in more severe injuries, diffuse axonal injury may also be present 181. Although they differ in terms of the location of their focal brain damage. due to the nature of their injury, they are all likely to have undetected diffuse axonal injury. In addition to analysing the CHI subjects as a group, individual data are examined in order to determine the generality of group cKects.
VISUAL
ATTENTION
AFTER (‘LOSED
HEAL) INJLKY
125
METHOD Suhjrc~t.t Eleven closed head-injury (CHI) subjects (IO males, I female)and 9 controls (8 males. 1 female) completed the experiment as paid volunteers. Demographic and clinical information for the CHI subjects is shown in Table 1.All. except subject 1 I, had moderate or severe damage according to the duration of post traumatic amnesia (PTA) exceeding 24 hours. They were tested at least 1 year post injury. The control subjects were matched for sex. age (.M=29.3, i 10.5) years of education (.%4= 12.1. i3.1) and estimated full scale IQ (M= 112,+4X). According to self report. all control subjects were neurologically intact. Visual acuitj was equivalent to or better than 6,9 (corrected) for all subjects as assessed by the Snellen eye chart.
The tabk was an adaptation [.2] ofa covert orienting ofattention task orlginally devised by POSWK [I 21. Subjects were rcquircd to press a single key as quickly as poaslble with their preferred (left or right) index finger whenever a \~sual target stimulus was presented. Target stimuli were presented at eye lcvcl and 9 degrees laterally to the left or right ofa central fixation point. A visual cue vas presented at the central fixation point II00 msec before onxt of a target stimulus. On seven out of the X0 trials presented in a block the Fisual cue consIsted ofa plus sign to indicate that the target uas equally lIkeI to occur on the left or the right. This was the neutral cue. The cue on 66 of the remaimng trial:; consisted ofan arrow pointmg to the left or right indicating with X0”& probability the side on which the target would occur. An arrow WC which correctly predicted the side of the target represented a valid cue and one which mcorrcctl> predicted target side an invalid one. Fourteen out of the 66 directionally cued trials were invalid. Thcrc were an equal number of right and left arrows presented in each block. On the remaining se\en trials a Lertical line \\as presented requiring the subject to inhibit responding to the target. This fourth cue represented a no-go condition. In each block the number of targets presented to the left and right usual field was equal. The cut rcmalned on during target presentanon htth both cue and target stirnull being terminated 900 msec after tnrgct onset. To he accepted. responsca had to occur between 100 and 900 mscc following target onset. Immediately foIlwing response. the subject was giLen performance feedback on ;I computer monitor located hclo~ the stimulus display. This consisted of either the response time in milliseconds or an error message when responses were anticipations (RTI 100 msec) or too slw (RT> 900 msec) or Nhen subjects responded folIoming a no-go cue. An inter-trial intcr\al (I 100 1200 msec) randomI> v:iried under computer control follo\\ed feedback. Within each sewon. the order of presentation of valid. invalid, neutral and no-go trials was randomized by computer aith the overall occurrence oflcft and right side targets hcpt cquiprohahle. To pre\ent overt orienting following presentation ofany cue subjects wcrc instructed to fixate on the central cue until completion ofthc response. Eye movcmcnts uere monitored \ ii1 8 mm sllwzr silver chloride cup electrodes placed on the lnncr and outer canthi of the right orbit. Signals Mere dlgitzcd (sampling rate of 3 mscc) after filtering (0.01 30 hr. 3 dB dorm. 6 dB octa\e slope) and high gain amplificatton (IO’). Data acquisition &as controlled hy a PC AT microcomputer. The cqc-mo\emcnt monitoring ~a!, done az part of ongoing electrocnccphnlographic recording to dcrl\e e\ ent-related potcnilal (ERP) mcasurcs to cue and target s~pnals. These analysts will be described in ii separxtc paper.
Cue and target stimuli consisted of red light emlttlng diode (LED) displays (Hwlett-Packard HLMP 1340) mounted on a horizontal metal rod 35 mm long and 23 mm bide. A centrnllc located 23 x 23 mm sqwre contained 21 symmetrically arranged 3 mm LEDs. Neutral. directional and no-go cues were presented by illumination of .Lppropriate LEDb. Single 3 mm red target LEDs wcrc located at the left and right of the central displ;lq LED intemity during display uas fiwcd at I .7 voltage Input alld the occurrcncc. duration and randomization of trials wcrc controlled bq purpose MI-itten mirocomputor hoft\+arc.
All testing took place in an acoustically treated room with controlled temperature. humidlt! and dmlmed lights. Subjects were seated upright in a comfortable chair and responses were made on ;I single button response panel in front ofthcm and on the body midline. The liwtion point w;ib I m from the aubjccts nxsion The rwording appar,ltu\ was housed in a room adjacent to the cxperimentnl cuixclc. Each subject completed a training session of between 6 and 10 blocks of X0 trial, on the task to reach an asymptotic lc\cl of performance. The euperlmental session conducted 7 IO days later consisted ofan Initial unrm up block of 20 trials followed by secen block\ of X0 trials aith a I 2 min rest between blocks. Each block la\ted approximately 4 min. Before each session. subjects were ashed to fixate on the central cut locntlon durmg each trial from cut onbet to rc\ponse completion and reminded to keep their preferred (left or right) index tingcr placed in readiness o\er the response button. tu respond as quickly as they could to target occurrence hut not to anticipate it and to monitor
rccdback.
Age
I6 11 IO
113
100
99
98
103
123
107.0 1.6
38
30
27
31
22
43
30. I 9.4
6
7
x
9
IO
II
Mean SD
amnesra:
Enyinrrr
I6
110
41
5
PTA: Post traumattc
utwmplo~c~d Unemployed S,rstcms attal~sr Part-time systems analyst
10
104
30
4
Ed: Education:
11.7 2.2
I2
II
I2
II
workr.~
CT: C‘omputcd
Maintenance Nurse’s aid Hospitality student sc/100/ />O!. Casual musician
Sk/i
Labourer
Luhourrr
Engineer
Police otTicer on desk duties
Motor cyck policroficer
Student
Studrt~t
Wood machinist
Cusutrlluhourer
II3
IO
17
CUSUU/fiKtOr~~ worker Unemployed
110
10
Ed.
(yrd
25
IQ
103
(yrs)
Occupation pre-injllrj post-injury
22
Case
Premorbid
nh.:
53
14
tomography:
n.k. (coma 20 min)
6
I3
near the anterior
splenium
EEG: spike and wave acttrtty
contuston
hacmatoma
Right temporal lobe damage. over right temporal arca
left temporal
shearing
384
Posterior
Right frontal
of
Contusion in the left tcmporo-parietal region and a patchy area of hypodensity with peripheral haemorrhage in the cortex and generalized swelling of the left hemisphere but without gross ventricular displacement to the left. Further scanning 6 days later revealed some atrophy in the left temporal region with widening of the sulci in that arca No focal injury but an impression of rather small ventricles
Small left parietal haemorrhage the corpus callosum
Not hnoun.
recovery
Right temporo-parietal extradural haemotoma was evacuated. CT scan revealed a thin right-srded subdural collection and moderate hydroccphalus. The chronic subdural collection was drained two months later. Neurological recovery was slow Right occipital contusion which did not result in any visual held defects Subarachnoid blood around the right occipital pole. Initially experienced double vision when using both cycs but has since adapted and could read the mini Sncllen eye chart at the required level (6:‘6) Right basal ganglia intracerebral haematoma with compression of the right lateral ventrical: a small shearing haematoma with compression of the right partctal region adjacent to the corpus callosum. thickening of the left frontal tissues, and air m the medial aspect of the left orbit No abnormality detected
CT Scan and neurological
49
53
51
32
I4
I3
20
44
16
38
58
Injury-test (months)
I5
21
35
>42
63
PTA (days)
Table I. Profile of each closed head-injury subject. All were involved in motor vehrcle accidents. Subjects 2. 8 and I I were riding motor cycles. subjects 3 and 9 were pedeatrrans and the rest were ctthcr driving or passengers tn motor vehicles. Subjects I. X and 9 were left-handed. the rest wcrc right-handed. All subjects wcrc male except subject 10
c ‘c c
5 ‘2 2
2 2 0 $
r
Ln
r;:
VISUAL
ATTENTlON
AFTER CLOSED
HEAD INJURY
127
Analysis of variance [20] was used to examine the effects of Group (CHI, control), Cue Validity (neutral, valid, invalid) and Target Location (left visual field, right visual field) on RT and response errors. To satisfy the assumption of homogeneity of variance logarithmic (base 10) transformations were carried out on RT data since variability increased linearly with their frequency and magnitude. Sphericity was assessed for each analysis and found to be non-significant. Therefore no adjustments of the Fratios or P values were required. Tukey posr hoc tests (x ~0.05) were used where appropriate 173. Relationships between measures were obtained using Pearson product&moment correlation coefficients. Median response times were analysed in two planned comparisons to examine the hen~fit of being correctly cued (neutral vs valid) and the cost of being miscued (invalid vs neutral). Consistent with the logic of planned comparisons, an overall ANOVA (Group x Cue Validity x Target Location) was not performed 171.
RESULTS The percentage number of response errors made by both groups was small [Controls: anticipations, M= 3.1 k2.7; misses, M=0.9 i 0.6 and CHI: anticipations, M= 3.3 k4.2; misses, M = 0.7 f 11. No responses to no-go trials were made. Thus response errors occurred too infrequently to reveal systematic significant effects, and statistical analyses revealed no significant effects. A Group (2) x Target Location (left vs right visual field) ANOVA on median reaction times showed no significant main or interaction effects, (CHI: M=298 k70.6, Control: M= 254 1.33.1). Figure 1 shows a scatterplot of each individual’s RT collapsed over visual field and cue condition.
63
420 -
E E
193 370 -
B 8
320 -
!i
270 -
10 8
9 5 7 4
8
326 4 11 6
8
220 -
k
5 2
Fig. I. lndivtdual mean of median RT for each control (subject numbers l--Y) and CHI subject (numbers l-1 I) averaged across types of cue conditions and side of target presentation.
Figure 2 shows the mean of median RTs for each of the cue conditions for each group. The benefit analysis revealed significant main effects of Group, F (1, 18) = 3.65, PC 0.05, and Cue validity, F (1, 18) = 6.24, PC 0.05. However, these were modified by a significant Group X Cue validity interaction, F (1, 18) = 6.39, P < 0.05. This was due to a reduced benefit of valid cueing in the CHI group (M= 1 k25 msec) compared to the control group (M = 30 + 11 msec) (see Fig. 2). The mean benefit (30 msec) shown by the control group was larger than that reported by CLARK et al. [2] (14 msec) and POSNER et al. [14] (16 msec) but it
12x
Closed Head Injury
CUE CONDITION
BENEFIT
0
COST
control
q Closed Head Injury Fig. 2. 7bp: mean of median RT (msec) to onset of target following valid, neutral and invalid cuts III the control (n, 9) and closed head injury (K I I) groups.Bott~~m: mean benefit of valid cueing (msec) and mean cost of In\ulid cueing for the control and closed head injury group:;. Bcnelit represents the dilTerence in response time to targets betaeen neutral and valid cue conditions and cost the difference between the invalid and neutral cue conditiona. Vertical bars represent the standard error of the mean.
was of the same order of magnitude (< 50 msec). Eight of the 11 subjects with CH I showed a reduced benefit (see Table 2). Since most of the CHI subjects had focal as well as diffuse brain damage a post hoc analysis of the magnitude of benefit was examined for each visual field in every subject. All of the control sut#cts showed normal benefits and none had visual field asymmetries in benefit. The eight subjects showing a reduced benefit overall, exhibited this in each visual field. Of the three CHI subjects who showed a normal benefit, case I showed an increased benefit to right targets (51 msec), and a decreased benefit to left targets (8 msec), case 10 produced the opposite benefit pattern (left targets: 83 msec; right targets: - 8 msec), while case 5 showed a normal benefit with no visual field asymmetry. Thus only one of the 11 cases showed a normal benefit compared to control subjects. The cost analysis revealed a significant main effect of Cue validity, F (1, IS)= 57.81,
VISUAL
ATTENTION
AFTER
CLOSED
129
HEAD INJURY
Table 2. Median reaction times (RT) following valid, neutral and invalid trials and RT benefit and cost for each closed head injury (CHI) subject. The mean RTs and standard deviations for the CHI and Control group are also shown Case
3 4 5 6 7 8 9 IO 11 Mean SD Control Mean SD
Neutral
cost
Valid
Invalid
Benefit
349 200 303 352 318 218 225 264 I 90 387 210 279 68
319 235 311 361 280 292 222 271 218 350 200 278 55
531 266 366 416 324 316 250 298 227 461 222 339 106
30* -35 -8 -9 38* ~ I4 3 -7 -28 37* 10 I 25
182t
250 32
220 21
293 46
30 I1
43 25
*Benefit normal magnitude. tCost > 1 SD above control
66 63 l24t 6 38 25 34 37 74t I2 60 52
mean
P
2. pp. 1235132.1992
c
002X-3932.92SS.OO+O.OO 1992 Pergamon Pressplc
COVERT ORIENTATION OF VISUAL ATTENTION CLOSED HEAD INJURY
AFTER
S. L. CREMONA-METEYARD, C. R. CLARK, M. J. WRIGHT and G. M. GEFFEN* Cognitive
Psychophysiology
Laboratory, (Receiced
Psychology
1 Auyust
Department,
1990: accepted
26
University
of Queensland,
Australia
4072
September 1991)
Abstract-The ability to orient visual attention covertly was studied in 11 patients who had suffered a moderate or severe closed head injury (CHI) at least 1 year previously. Their performance was compared to nine matched controls using a cued reaction time (RT) task which measured the RT benefit of valid directional cueing and the RT cost of miscueing. The CHI and control groups did not differ in overall RT. Relative to controls, the CHI group showed normal cost but hardly any benefit, indicating that the normal capacity to pre-align attention with a cued location was impaired.
DISORDERS OF ATTENTION are a common outcome of closed head injury (CHI) and may underlie the characteristic problems of learning, memory and problem solving that are frequently reported after significant head injuries [ 11. This study aimed to identify residual deficits in the ability to shift and focus attention covertly following CHI. Since we were interested in identifying persistent deficits, our subjects were tested l-5 years post injury when their recovery may be regarded as stable. Previous studies of attention have consistently identified slower information processing in subjects who have sustained CHI (e.g. Refs Cl], [21] and [22]). Most of these studies have used relatively complex tasks and found deficits in the decision-making phase of information processing. In a recent study, Rucc et al. [ 1S] found that response time on a simple RT task with auditory cues and imperative signals did not differentiate the CHI group from controls. Other studies of simple and choice visual RT have shown that head-injured subjects were slower than controls with the extent of the difference increased with the number of choices or complexity of the task [ 1,9, 17,223. Severe CHI subjects tested less than 1 year post injury showed an impairment in both the early identification and the response selection stages of processing in a visuo-spatial task [ 191. The early perceptual deficit appears to diminish with time, as patients tested more than 1 year post injury showed an impairment in only the response selection phase [ 193. The present study used a visual cued RT task involving covert orientation of attention [ 12, 143. This task is particularly useful for studying head-injured subjects as it requires one simple response yet is sensitive to covert changes of orientation of attention that are independent of muscular movement of the eyes or head [ 161. Expectancy is manipulated by
*Address Queensland
for reprints: Professor 4067, Australia.
G. Geffen,
Psychology
123
Department,
University
of Queensland,
St Lucia,
124
s. L. CKl.hlOL.A-MI
71 YAKI)
c’f cl/.
providing directional cuts that are either valid or not valid in informing the subject of whether a target will occur in the left or right visual field 1121. The benefits (in speed of response to the target) of directional engagement of attention and costs of subsequent shifting of attention are measured in relation to RT to a target after a neutral cue that provides no directional information about target location. Covert orienting involves the directional engagement of attention. whereas reorienting after an invalid CLK requires disengagement, shifting, and re-engagement of attention 1161. This model of attention is based on a spotlight metaphor, with the locus of attention being highlighted. While the metaphor has been criticized [3.4], empirical testing of its predictions has been successful in cases with circumscribed lesions [ 16. 241. Hence in this study the task was given to casts with more variable lesions and the extent to which the model can account for findings was examined. Brain lesion studies [I31 and single cell recording in alert monkeys [IO, I I. 251 have identified particuiar anatomical locations associated with a number of the mental operations involved in the covert orientation of visual attention. Specifically. the posterior parietal lobe has been linked to the disengagement of attention, and areas of the midbrain and thalamus to the covert movement and engagement of attention, respectively. Consequently, injury to these three brain areas causes deficits of different natures. Patients with unilateral parietal lesions show increased RTs to targets that are contralateral to the lesion and occur after an invalid cue (i.e. the side ipsilateral to the lesion was cued) 1161. This is consistent with a specific deficit in the patient’s ability to disengage from the current focus of attention and move in a direction contralateral to the lesion. A disengage deficit has also been reported in a patient with a right frontal lobe lesion 1151. suggccting that the frontal areas may also play an important role in the covert orientation of attention. In contrast, damage to the midbrain increases overall RT and lengthens the time taken to establish an efficient RT at the cued location. Damage to the thalamus produces a pattern of slowed RT to both cued and uncued targets on the side opposite the Icsion, suggesting an engage deficit. A similar deficit has been found in monkeys performing the task when chemical injections disrupt the performance of visual parts of the pulkinar [I !. IS]. Normal subjects injected with droperidol and clonidine to suppress central dopamine and noradrenaline transmission showed reductions in the cost of invalid cueing without a change in the benefit of valid cueing [2]. A similar result was found in patients with Parkinson’s disease 1243. From the results outlined above it is apparent that the simple act of shifting attention to a cued location is a complex operation involving a number of different anatomical locations. Several distinct computations must be orchestrated to allow the shifting of attention to a cued location. Given the variable and difluse damage resulting from head injury an impairment in the covert orienting of attention is expected, but the exact nature of the deficit is ditlicult to predict. While the CH! subjects differ in location and size of lesions. they may be regarded as homogeneous for a number of reasons that warrant studying them as a group. They all share the same etiology. they are mostly young males and they generally display similar deficits in terms of information processing. Common sites of injury are the frontal lobes and temporal poles. However. in more severe injuries, diffuse axonal injury may also be present 181. Although they differ in terms of the location of their focal brain damage. due to the nature of their injury, they are all likely to have undetected diffuse axonal injury. In addition to analysing the CHI subjects as a group, individual data are examined in order to determine the generality of group cKects.
VISUAL
ATTENTION
AFTER (‘LOSED
HEAL) INJLKY
125
METHOD Suhjrc~t.t Eleven closed head-injury (CHI) subjects (IO males, I female)and 9 controls (8 males. 1 female) completed the experiment as paid volunteers. Demographic and clinical information for the CHI subjects is shown in Table 1.All. except subject 1 I, had moderate or severe damage according to the duration of post traumatic amnesia (PTA) exceeding 24 hours. They were tested at least 1 year post injury. The control subjects were matched for sex. age (.M=29.3, i 10.5) years of education (.%4= 12.1. i3.1) and estimated full scale IQ (M= 112,+4X). According to self report. all control subjects were neurologically intact. Visual acuitj was equivalent to or better than 6,9 (corrected) for all subjects as assessed by the Snellen eye chart.
The tabk was an adaptation [.2] ofa covert orienting ofattention task orlginally devised by POSWK [I 21. Subjects were rcquircd to press a single key as quickly as poaslble with their preferred (left or right) index finger whenever a \~sual target stimulus was presented. Target stimuli were presented at eye lcvcl and 9 degrees laterally to the left or right ofa central fixation point. A visual cue vas presented at the central fixation point II00 msec before onxt of a target stimulus. On seven out of the X0 trials presented in a block the Fisual cue consIsted ofa plus sign to indicate that the target uas equally lIkeI to occur on the left or the right. This was the neutral cue. The cue on 66 of the remaimng trial:; consisted ofan arrow pointmg to the left or right indicating with X0”& probability the side on which the target would occur. An arrow WC which correctly predicted the side of the target represented a valid cue and one which mcorrcctl> predicted target side an invalid one. Fourteen out of the 66 directionally cued trials were invalid. Thcrc were an equal number of right and left arrows presented in each block. On the remaining se\en trials a Lertical line \\as presented requiring the subject to inhibit responding to the target. This fourth cue represented a no-go condition. In each block the number of targets presented to the left and right usual field was equal. The cut rcmalned on during target presentanon htth both cue and target stirnull being terminated 900 msec after tnrgct onset. To he accepted. responsca had to occur between 100 and 900 mscc following target onset. Immediately foIlwing response. the subject was giLen performance feedback on ;I computer monitor located hclo~ the stimulus display. This consisted of either the response time in milliseconds or an error message when responses were anticipations (RTI 100 msec) or too slw (RT> 900 msec) or Nhen subjects responded folIoming a no-go cue. An inter-trial intcr\al (I 100 1200 msec) randomI> v:iried under computer control follo\\ed feedback. Within each sewon. the order of presentation of valid. invalid, neutral and no-go trials was randomized by computer aith the overall occurrence oflcft and right side targets hcpt cquiprohahle. To pre\ent overt orienting following presentation ofany cue subjects wcrc instructed to fixate on the central cue until completion ofthc response. Eye movcmcnts uere monitored \ ii1 8 mm sllwzr silver chloride cup electrodes placed on the lnncr and outer canthi of the right orbit. Signals Mere dlgitzcd (sampling rate of 3 mscc) after filtering (0.01 30 hr. 3 dB dorm. 6 dB octa\e slope) and high gain amplificatton (IO’). Data acquisition &as controlled hy a PC AT microcomputer. The cqc-mo\emcnt monitoring ~a!, done az part of ongoing electrocnccphnlographic recording to dcrl\e e\ ent-related potcnilal (ERP) mcasurcs to cue and target s~pnals. These analysts will be described in ii separxtc paper.
Cue and target stimuli consisted of red light emlttlng diode (LED) displays (Hwlett-Packard HLMP 1340) mounted on a horizontal metal rod 35 mm long and 23 mm bide. A centrnllc located 23 x 23 mm sqwre contained 21 symmetrically arranged 3 mm LEDs. Neutral. directional and no-go cues were presented by illumination of .Lppropriate LEDb. Single 3 mm red target LEDs wcrc located at the left and right of the central displ;lq LED intemity during display uas fiwcd at I .7 voltage Input alld the occurrcncc. duration and randomization of trials wcrc controlled bq purpose MI-itten mirocomputor hoft\+arc.
All testing took place in an acoustically treated room with controlled temperature. humidlt! and dmlmed lights. Subjects were seated upright in a comfortable chair and responses were made on ;I single button response panel in front ofthcm and on the body midline. The liwtion point w;ib I m from the aubjccts nxsion The rwording appar,ltu\ was housed in a room adjacent to the cxperimentnl cuixclc. Each subject completed a training session of between 6 and 10 blocks of X0 trial, on the task to reach an asymptotic lc\cl of performance. The euperlmental session conducted 7 IO days later consisted ofan Initial unrm up block of 20 trials followed by secen block\ of X0 trials aith a I 2 min rest between blocks. Each block la\ted approximately 4 min. Before each session. subjects were ashed to fixate on the central cut locntlon durmg each trial from cut onbet to rc\ponse completion and reminded to keep their preferred (left or right) index tingcr placed in readiness o\er the response button. tu respond as quickly as they could to target occurrence hut not to anticipate it and to monitor
rccdback.
Age
I6 11 IO
113
100
99
98
103
123
107.0 1.6
38
30
27
31
22
43
30. I 9.4
6
7
x
9
IO
II
Mean SD
amnesra:
Enyinrrr
I6
110
41
5
PTA: Post traumattc
utwmplo~c~d Unemployed S,rstcms attal~sr Part-time systems analyst
10
104
30
4
Ed: Education:
11.7 2.2
I2
II
I2
II
workr.~
CT: C‘omputcd
Maintenance Nurse’s aid Hospitality student sc/100/ />O!. Casual musician
Sk/i
Labourer
Luhourrr
Engineer
Police otTicer on desk duties
Motor cyck policroficer
Student
Studrt~t
Wood machinist
Cusutrlluhourer
II3
IO
17
CUSUU/fiKtOr~~ worker Unemployed
110
10
Ed.
(yrd
25
IQ
103
(yrs)
Occupation pre-injllrj post-injury
22
Case
Premorbid
nh.:
53
14
tomography:
n.k. (coma 20 min)
6
I3
near the anterior
splenium
EEG: spike and wave acttrtty
contuston
hacmatoma
Right temporal lobe damage. over right temporal arca
left temporal
shearing
384
Posterior
Right frontal
of
Contusion in the left tcmporo-parietal region and a patchy area of hypodensity with peripheral haemorrhage in the cortex and generalized swelling of the left hemisphere but without gross ventricular displacement to the left. Further scanning 6 days later revealed some atrophy in the left temporal region with widening of the sulci in that arca No focal injury but an impression of rather small ventricles
Small left parietal haemorrhage the corpus callosum
Not hnoun.
recovery
Right temporo-parietal extradural haemotoma was evacuated. CT scan revealed a thin right-srded subdural collection and moderate hydroccphalus. The chronic subdural collection was drained two months later. Neurological recovery was slow Right occipital contusion which did not result in any visual held defects Subarachnoid blood around the right occipital pole. Initially experienced double vision when using both cycs but has since adapted and could read the mini Sncllen eye chart at the required level (6:‘6) Right basal ganglia intracerebral haematoma with compression of the right lateral ventrical: a small shearing haematoma with compression of the right partctal region adjacent to the corpus callosum. thickening of the left frontal tissues, and air m the medial aspect of the left orbit No abnormality detected
CT Scan and neurological
49
53
51
32
I4
I3
20
44
16
38
58
Injury-test (months)
I5
21
35
>42
63
PTA (days)
Table I. Profile of each closed head-injury subject. All were involved in motor vehrcle accidents. Subjects 2. 8 and I I were riding motor cycles. subjects 3 and 9 were pedeatrrans and the rest were ctthcr driving or passengers tn motor vehicles. Subjects I. X and 9 were left-handed. the rest wcrc right-handed. All subjects wcrc male except subject 10
c ‘c c
5 ‘2 2
2 2 0 $
r
Ln
r;:
VISUAL
ATTENTlON
AFTER CLOSED
HEAD INJURY
127
Analysis of variance [20] was used to examine the effects of Group (CHI, control), Cue Validity (neutral, valid, invalid) and Target Location (left visual field, right visual field) on RT and response errors. To satisfy the assumption of homogeneity of variance logarithmic (base 10) transformations were carried out on RT data since variability increased linearly with their frequency and magnitude. Sphericity was assessed for each analysis and found to be non-significant. Therefore no adjustments of the Fratios or P values were required. Tukey posr hoc tests (x ~0.05) were used where appropriate 173. Relationships between measures were obtained using Pearson product&moment correlation coefficients. Median response times were analysed in two planned comparisons to examine the hen~fit of being correctly cued (neutral vs valid) and the cost of being miscued (invalid vs neutral). Consistent with the logic of planned comparisons, an overall ANOVA (Group x Cue Validity x Target Location) was not performed 171.
RESULTS The percentage number of response errors made by both groups was small [Controls: anticipations, M= 3.1 k2.7; misses, M=0.9 i 0.6 and CHI: anticipations, M= 3.3 k4.2; misses, M = 0.7 f 11. No responses to no-go trials were made. Thus response errors occurred too infrequently to reveal systematic significant effects, and statistical analyses revealed no significant effects. A Group (2) x Target Location (left vs right visual field) ANOVA on median reaction times showed no significant main or interaction effects, (CHI: M=298 k70.6, Control: M= 254 1.33.1). Figure 1 shows a scatterplot of each individual’s RT collapsed over visual field and cue condition.
63
420 -
E E
193 370 -
B 8
320 -
!i
270 -
10 8
9 5 7 4
8
326 4 11 6
8
220 -
k
5 2
Fig. I. lndivtdual mean of median RT for each control (subject numbers l--Y) and CHI subject (numbers l-1 I) averaged across types of cue conditions and side of target presentation.
Figure 2 shows the mean of median RTs for each of the cue conditions for each group. The benefit analysis revealed significant main effects of Group, F (1, 18) = 3.65, PC 0.05, and Cue validity, F (1, 18) = 6.24, PC 0.05. However, these were modified by a significant Group X Cue validity interaction, F (1, 18) = 6.39, P < 0.05. This was due to a reduced benefit of valid cueing in the CHI group (M= 1 k25 msec) compared to the control group (M = 30 + 11 msec) (see Fig. 2). The mean benefit (30 msec) shown by the control group was larger than that reported by CLARK et al. [2] (14 msec) and POSNER et al. [14] (16 msec) but it
12x
Closed Head Injury
CUE CONDITION
BENEFIT
0
COST
control
q Closed Head Injury Fig. 2. 7bp: mean of median RT (msec) to onset of target following valid, neutral and invalid cuts III the control (n, 9) and closed head injury (K I I) groups.Bott~~m: mean benefit of valid cueing (msec) and mean cost of In\ulid cueing for the control and closed head injury group:;. Bcnelit represents the dilTerence in response time to targets betaeen neutral and valid cue conditions and cost the difference between the invalid and neutral cue conditiona. Vertical bars represent the standard error of the mean.
was of the same order of magnitude (< 50 msec). Eight of the 11 subjects with CH I showed a reduced benefit (see Table 2). Since most of the CHI subjects had focal as well as diffuse brain damage a post hoc analysis of the magnitude of benefit was examined for each visual field in every subject. All of the control sut#cts showed normal benefits and none had visual field asymmetries in benefit. The eight subjects showing a reduced benefit overall, exhibited this in each visual field. Of the three CHI subjects who showed a normal benefit, case I showed an increased benefit to right targets (51 msec), and a decreased benefit to left targets (8 msec), case 10 produced the opposite benefit pattern (left targets: 83 msec; right targets: - 8 msec), while case 5 showed a normal benefit with no visual field asymmetry. Thus only one of the 11 cases showed a normal benefit compared to control subjects. The cost analysis revealed a significant main effect of Cue validity, F (1, IS)= 57.81,
VISUAL
ATTENTION
AFTER
CLOSED
129
HEAD INJURY
Table 2. Median reaction times (RT) following valid, neutral and invalid trials and RT benefit and cost for each closed head injury (CHI) subject. The mean RTs and standard deviations for the CHI and Control group are also shown Case
3 4 5 6 7 8 9 IO 11 Mean SD Control Mean SD
Neutral
cost
Valid
Invalid
Benefit
349 200 303 352 318 218 225 264 I 90 387 210 279 68
319 235 311 361 280 292 222 271 218 350 200 278 55
531 266 366 416 324 316 250 298 227 461 222 339 106
30* -35 -8 -9 38* ~ I4 3 -7 -28 37* 10 I 25
182t
250 32
220 21
293 46
30 I1
43 25
*Benefit normal magnitude. tCost > 1 SD above control
66 63 l24t 6 38 25 34 37 74t I2 60 52
mean
P
