Parallel and serial visual search after closed head injury: Electrophysiological evidence for perceptual dysfunctions

PARALLEL AND SERIAL VISUAL SEARCH AFTER CLOSED HEAD INJURY: ELECTROPHYSIOLOGICAL EVIDENCE FOR PERCEPTUAL DYSFUNCTIONS HANS-J• ~HEN HEINZE,” THOMAS FRANK M~JNTE,WOLFGANG GOBIET,? HENDRIK NIEMANN: and RONALD M. RUFF$ Department of Neurology, Medical School of Hannover. F.R.G.: F.R.G.: ZEpilepaiezentrum Bethel. F.R.G.: and $Department San Francisco.

tNeurologische of Psychiatry.

Klinik Hess&h Oldendorf. University

of California.

U.S.A.

Abstract -Event-related potentials (ERPs) were recorded from closed head injury (CHI) patients at least 2 years postmjury and From controls in order to assess their parallel and serial processing abilities during visual search. In Experiment I. stimuli consisted ofarrays ofeight triangles; halfofthe arrays contained a target item. In the “feature-present” condition, the target item was a triangle with an additional horizontal line that could be detected automatically and in parallel, while in the “feature-absent” condition all items except for the target triangle had an additional horizontal line, thus requiring a serial search. In Experiment 2. stimuli consisted ofeight solid bars (50?4). seven solid bars and a vertical open bar (25%). and seven solid bars and a horizontal open bar (25%): the array containing the horizontal bar serbed as a target. By recording ERPs to the arrays containing vertical open bars. which were similar to the target items. parallel processing of “pop-out” stimuli could be studied in the absence of any overt response. ERP data were compared with the results of neuropsychological and neuroimaging (MRI, CAT) exammation. Patient exhibited a decreased behavioral performance both in the parallel and in the serial processing mode. Furthermore. abnormalities of early and intermediate ERP components (PI. Nl. P2. N2) were found. whereas the late component (P3) was less alfected by CHI. The results were intcrprcted as an index of CHIinduced dysfunctions in perceptual proccsscs such as simple feature registration and early target discrimination. It was suggested that these dysfunctions contribute to impairments of parallel as well as serial processes in visual search.

INTRODUCTION

THE VIEW IS widely held that deficits in attention constitute a core problem after closed head injury (CHI) 14, 161. Cognitive impairments that are considered to reflect an attentional disturbance include a general slowness of reactions and thoughts, difficulties in concentrating on relevant and ignoring irrelevant events, an impairment in performing two tasks simultaneously and a reduced ability to focus on a problem over an extended period of time. It is generally agreed that a reduction of information processing speed is a basic factor underlying these disorders 191. From a clinical point of view. attentional deficits may present the strongest impediment to successful rehabilitation [3]. Despite the surge of interest in attention over the past few years, the neural mechanisms of attentional disorders associated with CHI are not well understood. Little is known about the

*Address correspondence to: H.-J. Heinze. Department Konstanty-Gutschow-Str. X. D-3000 Hannover 61, F.R.G. 495

of Neurology.

Medical

School

of Hannover.

4%

H.-.I. HI

IN/I

6’1

cr/

anatomy of traumatic brain lesions in relation to attentional problems, except for some aspects of visual selective attention [25]. Recent studies of magnetic resonance imaging (MRI) after CHI emphasize the correlation between the depth of cerebral lesions and neuropsychological impairments 117, 391. It has been proposed that the general cognitive slowness might reflect a decreased strength of association between conceptual nodes of neuronal networks due to diffuse lesions of white matter, whereas more specific cognitive deficits were related to focal cortical contusions and axonal damage of the rostra1 brain stem

C401~ Most theories of attention agree that a major distinction exists between processes that are automatic, parallel and with no capacity limits and processes that operate in a serial. capacity-limited mode. For example, it has been proposed that parallel processing occurs at an early stage of stimulus processing and performs the extraction of simple features. whereas serial processing is needed to localize or identify objects or the conjunction offeatures [14,2 I. 381. Behavioral studies in head-injured patients obtained so far provide evidence for a reduction of processing capacity as a basic mechanism underlying cognitive slowness after CHI [40]. Within the framework of parallel and serial processing, these results suggest that cognitive slowness after CHI basically reflects deficits on serial, capacity-limited stages of information processing. Electrophysiological studies in human subjects are playing an increasing role in the analysis of neuronal mechanisms mediating attcntional processes [IO, 121. In particular the recording of event-related brain potentials (ERPs) reveals neural events at successive stages of information processing independent of motor performance. ERPs arc scalp-recorded electrical potentials generated by neural activity and associated with specific sensory. cognitive and motor processes [7], Unlike beha\,ioral measures. they reflect proccsscs that intervene between stimulus and rcsponsc. thereby providing information about their time course and cerebral localization. To date. only a limited number of studies has investigated ERPs in head-injured patients. These studies mainly focused on the contingent negative variation (CNV) and the P300 [S, 6. 22, 24. 76. 30. 311. Most of these studies provided evidence that head-injured patients arc impaired on later stages of processing such as stimulus c\aluation and response preparation. RLJ(;G et trl. [-?O]. however. also found disturbances on earlier stages of processing as indcsed by an increased N2 component in an auditory oddball task. They argued that C’HI patients might need additional time to achic\e stimulus categorization even in ;I task as undemanding as auditory oddball detection. The present paper is aimed at delineating neural mechanisms of attentional disorders after CHI ivithin the context of parallel and serial processing. In particular. the question addressed was whether cogniti1.e slowness reflects deficits in capacity-limited stages of stimulus analysis or whether dysfunctions of automatic processes without capacity limits also play a role. ERPs were recorded in two visual search experiments. in which sub.jccts had to decide lvhether or not a target item was present in a number of distractor items. These paradigms resemble the natural situation that occurs when people arc forced to detect and rccognizo relevant information in a visual sccnc. According to the feature integration theory [34. 351. normal subjects register basic features such as colour, shape or orientation automatically and in parallel. while the conjunction of these features into unitary ob_jects is a serial. capacitylimited process. The purpose of Expcrimcnt 1 M;IS to ;MCSS mechanisms of parallel and serial search in head-injured patients compared to a control group. As shov, I1 by LCCK and HII.L.YAKI>[20] in an ERP study in normal sub.jects. mechanisms of par-;lElel and serial visual search arc

VISLAL

SFAKCH

4FTtR

(‘LOSLD

HE-AD ISJL’KY

497

reflected in the posterior N2 and P3 components of the visual ERPs. The posterior N2 was found both in the parallel and serial search condition indicating the presence of target discrimination on an early stage of stimulus analysis in both conditions, whereas the subsequent P3 significantly differed between these conditions, its latency and amplitude being a function of stimulus recognition processes and decision confidence [15, 231. If it is correct that processing capacity is reduced after CHI, one would expect differences between patients and normals to occur in the serial search condition. On the other hand, if CHI also impairs parallel feature detection or processes that are involved both in parallel and serial processing, patients might exhibit decreased performance and ERP abnormalities in the parallel and serial search condition. Evaluation of performance in head-injured patients is often difficult because of motor impairments. Therefore, Experiment 2 investigated automatic processing of simple visual features without overt responses. According to LUCK and HILLYARD [ 1S] and HILLYARD et al. [13], visual items that seem to “pop-out” from an array of distractor items elicit an early anterior P2 and an anterior N2 component. The authors showed that the anterior N2 was elicited, when any pop-out stimulus was present in the display, whereas the anterior P2 was more specifically elicited by target stimuli. They concluded that the anterior N2 reflects a more automatic form of feature analysis/mismatch detection than does the anterior P2 [13]. In Experiment 2 of the present study, automatic processing after CHI in the absence of a motor response was studied by presenting target and nontarget pop-out items within a number of distracters. It was reasoned that a difference between the P2 and N2 components to the nontarget pop-outs in the patients and controls would provide evidence for an impaired parallel processing of visual features after CHI.

METHODS The present study considers data from I I patients (two females) out of a group of 20 patients. For the remaining nine subjects no reliable ERPs could be obtained because of visual lield defects. inability to suppress eye movements and blinks or inability to maintain concentration over the time required for testing. All patients had suffered from a closed head injury at least 2 years before examination and fulfilled the following two criteria: (I I post-traumatic amnesia (PTA) greater than 48 hr. (2) Glasgow Coma Scale on admisston to hospital less than 8. All patients had normal or correct-to-normal vision. At the time of testing only two patients were gainfully employed. all others were unable to work. Pertinent subject characteristics are summarized in Table I.

Twelve control subjects were chosen to match the patient group. as closely as possible. with respect to age. sex. handedness and education. Patients as well as controls gave their informed consent to participate in the study.

Prior to ERP-recording an extensive neuropsychological test battery was administered to the patients (San Diego Neuropsychological Test Battery, SDNB [ZS]). This battery provides an examination according to cognitivespecific test measures including a number of standard test procedures (e.g. WAIS-R, REY complex figure test. Wechsler short stories. Stroop test, Wisconsin Card Sorting test) as well as additional tests to give an account of attention. intelligence. memory, visuoconstructive and psychomotor functions. If available, the German version of the test procedures was used (e.g. the German version of the WAIS-R f33]). For the other tests. German norms were established with regard to the specific clinical populatton. Selected test results. especially those addressing perception and attention (digit span. letter span. block span, 2 & 7 selective attention) and verbal memory (Wechsler short stories) as well as visuomotor performance (Grooved Pegboard test) and executtve functions (Rut? Figural Fluency test) are depicted in Table 2. The WAIR-R (the German version) was administered and scored according to the instruction manual [33]. The letter span and block span were used as a measure of immediate auditory and visuospatial attention and concentration and scaled according to the instruction manual of the SDNB. The 2 & 7 selective attention test addresses selective visual

GS. 1-s(‘r-K WA l?5-lC-K NA I-s-t~. th-I.
All values

I0 II

6 7 X 9

5

4

2 3

Patlent

arc r-scores.

except

for IQ

104 90 73 64 70 90 76 94 73 75 5X

110 IO0 122 78 X.5 X3 71 80 X6 5X 6X

109 9s I0I 69 76 x5 76

x5 79 62 60

IQ

Performance

10

Verbal

lC?

F~ullacale Vocabulary 0.2 ~ 1.0 0. I m-O.8 -0.x -1.5 ~ 1.7 - I.0 -0.7 -4.0 -2.7

-0.6 0.4 -0.7 -0.3 - I.6 -- I.5 ~ 1 .h -0.9 -2.0 - I .9 -2.6 -3.6 - I .6 -3.9 ~ 2.6 - 1.7 -1.8 -2.5 -2.X -3.9 -2.5

-2.0

Digit symbol

2. Ncuropsychological

Wechsler .short stories

Table

~ I.4 0.4 0.X -2.3 -0.5 2.0 -2.3 0.3 ~ 1.4 ~ 1.X -3.1

Digit span ~ 1.5 ~ I .o I .o ~ 1.9 ~ I .9 ~ 1.9 -0.5 -0.5 ~ I.0 -0.1 0.4

Block span

assessment

0.1

-0.3 -2.X ~ 1.2 - I .6 -2.0 ~ 7.0

-0.8 ~ I.2

I .3 -2.5 - I.6 -0.5 I .4 0.1 - I.9 -3.5 < ~ 4.0

0.0 0.9

~I2 -2.8

z & 7 test (automatic)

Letter span

0.0 I .4 -7.2 ~ I.6 -0.5 I .9 -0.0 -2.0 -3.6 < -4.0

~ 1.o

2 & 7 test (controlled)

I.2 -0.5 ~ 3.3 -0.4 ~ 0.7 -0.4 -0.2 0.6 ~ I.2 ~ 0.9 < ~ 4.0

-

Groo\cd Pegboard test

RCJl:F figural Iluency

H.-J. HI IV/I

Experiment

1’1trl

1

Feature absent

Feature present

6.50

i

8.20 Experiment Target

’

’

2

Pop-out

Non-target

’

I

I

‘0

I I

I

’

I I

0

I

Pop-out

’

I I 14O

\ISI:AL

SLAK(‘H

AFTEK C‘LOST.L, Ht41)

INJUKY

501

strict rejection criterion could not be employed. Trials with incorrect beha\ ioural responses were also excluded from the averages. The amplitudes of the early PI, N 1. P2 and anterior N2 components were assessed as mean boltage over the time ranges 60 120, 13&-190, 260-320 and I80 240 msec. respectively. with reference to the baseline voltage averaged over the 100 msec interval preceding stimulus onset. Mean amplitude measures were chosen because peak amplitude is reduced by latency jitter. whereas mean amplitude is not. The PI and NI peaks were measured at parietal, occipital and temporo-occipital sites. and the P2 peak was measured at frontal and central sites. In order to quantify changes of the posterior N2 component associated with the detection of visual targets within multi-element stimulus arrays [I I, 191, averages were performed based on target position (left or right half-field). The amplitude was assessed at parietal, occipital and temporo-occipital sites as mean voltage between 220 and 280 msec. P3 amplitude was measured as mean voltage between 350 and 750 msec post-stimulus. The analysis was restricted to this time range because of the eye blink artifacts that contaminated longer epochs in the patient group. although in the feature-absent condition P3 activity extended beyond 750 msec. For this reason. P3 latency was assessed onI) in the feature-present condition as the time required for the component to reach 50 ‘/u of Its mean amplitude over the 350 750 msec measurement window. This measurement was chosen instead of the usual peak latency measurement because the broad waveform in many patients did not allow for a clear peak identification. Behavioral performance was quantitied by measuring reaction time (RT) and calculattng % hit scores for dctectlon of targets. All ERP and behavioral measures were evaluated with repeated measures analyses of variance (ANOVAs). Since the pattern ofeffects was not known at the time ofstudy design and latency shifts ofthe late ERPcomponcnts in the patient group were a possibility, within-subject analyses were performed for patients and controls as well as between-subject analyses including the subject group as a factor. This was preferred to an overall design including group as a factor. because the within group analyses permit an easier interpretation of the effects within each group. The factors for ERP evaluation in Experiment I were: Hemisphere (left vs right). Trial type (standard vs target) and Search-task (feature-present vs feature-absent ). ANOVAs were performed separately for each electrode pair (frontal. central. parietal. occipital and temporo-occipital). This was done because the studies of LUCK and HILLYAKI) [IX, 19. 201 revealed a circumscribed scalp-distribution for the ERP-components (c.g. antrrior N2, posterior N2, P3), so the inclusion ofall channels would have led to interaction efl‘ects yielding no new information. Measures based on stimulus location were analysed with the same factors, except that the trial type factor was replaced by a target position factor (left vs right). Behavioral measures were analysed with tbo factors: Subject (patients vs controls) and Search-task (feature-present vs feature-absent), In Experiment 2. ERP and behavioral measures were analysed with the same factors. except for the trial type factor (standard. nontarget pop-out. target pop-out). adjusting for nonsphericity with the Greenhouse-Geisser epsilon coefficient.*

RESULTS E.uperinwnt

I

Brhur?orul nzeusures. Mean RTs and hit rates are presented in Table 3, showing a significant performance difference between patients and controls both in the feature-present and feature-absent condition (RT: P
separate analyses on the different channel pairs as being that in this may interaction effects providing This analysis strategy chosen for the present study we have followed the suggestions of ABT [I. 21. in the design and Interpretation of the study. He for this unduly increases the risk of type II error.

502

and and the the

H.-J. HI I\/,

c, II/

X00 msec. peaking at 410 msec. Ncgntivc trials produced similar early components PI N I kvhile eliciting a P3 component with smaller amplitude in the 300 X00 mscc range. In patient group (right side of Fig. 2). the amplitude of the PI component was reduced. and amplitude of the N I nave \vas strongly reduced compared to the controls. Phc 1’3 co~~~po~~c~~t to the positive trials showed ;I broader wawform and ;I delayed peak latency (SW Table 3). kvhereas the P3 to the negative trials exhibited only minor differences. The separation bet\vcen the P3 to the positive and negative trials started about 100 msec later in the patient than in the control group. In order to assess EKP dill‘erenccs between the patients and controls it is important that the variability of the waveforms in the two group4 is comparable. In Fig. 3. mean value and standard de\ iation (SD) of the dill’crcnce w;t\cs: target minus standard in the two groups arc presented. Ah can bc seen. patients and controls display about the same \ariahility of ERP ctl‘ects. Figure 3 shows the ERP efti’cts dcpcnding on the location of the target item within the arrays in the feature-present condition. Normal subjects showed a negativity (N2) over the contralateral hemisphere rclati\c to the ipsilatcrnl hcmispherc from about 230 to 2X0 mscc at the posterior leads. As can bc sew from the HEOG. the small cyc movements to the side of the target could not have produced this ctr‘ect because the contralatcral N2 was shorter in duration than the eye movement and rcstrictcd to the posterior scalp sites. The patients, on the other hand, did not exhibit as large ;I comparable contralatcral N2, although the qt‘ movcmcnts MCTCcvcn more pronounced than in the control group. In the fcaturc-absent condition. the controls displayed marked dilfercnces in the I:KP waveforms compared to the fcaturc-present condition. Thcsc dilTcrcnccs wcrc: in line with the lindings of the LL.C,K and HII.I.YAIW study [Xl, showing ;I dclaycd and broadcr Pi component to the positike trials that rcllectcd the longer duration and possibly an incrwscd \ariancc of the peak latencies. WIG patients’ F:RPs to the positive and negative trials at the posterior leads in the fcaturc-absent condition cxhibitcd ;I markedly dccreascd N I amplitude compared to the normals (Fig. 5 )_Also. the separation between the ER Ps to the positive and negative trials started about 100 IINX later in the patient than in the control group at the posterior Icads. Furthcrmorc. the P3 component to the positive trials did not display such :I strong ditl’ercnce bct\\ceen the feature-absent and fcaturc-present condition as sucn in the control group, possibly rcfcctinp an incrcascd variance in peak latency in both conditions.

503

CONTROLS F3/4

PATIENTS F3/4

_

-

Negative

-----

PosIt1”c

5PV +

I

:

900 / :

I

msec /

1

Fig. 7. Scalp distrihutlon of the grand average ERPs in patients (right) and controls (left) to the positive (target-prcscnt, dashed) and ncgati\e (target-ahscnt. solid) stimuli in the feature-present condition of Experiment I, averaged over left and right electrode pairs.

As in the feature-present condition, patients and controls displayed a comparable variability of the ERP waveforms in the feature-absent condition (Fig. 3). Table 4 summarizes some results of the statistical analysis. The ANOVAs confirmed a highly significant difference between the Nl amplitude in the patient and control group, while there was no significant PI amplitude difference. Because the early effects in the featureabsent condition were contaminated by baseline artifacts, a separate analysis for the PI component in the feature-present condition was performed that revealed a significant smaller amplitude (P < 0.03 ). whereas P 1 latency showed no significant differences. Furthermore, the separation between the P3 components to the positive and negative trials started earlier in the controls than in the patients, as assessed by mean amplitude difference measurements between 300. 400 and 400-500 msec at temporo-occipital leads. The group x stimulus x task interaction indicated a greater delay in the feature-absent condition for the patients. Consistent hemisphere effects were not observed and therefore not reported in Table 4. The contralaterality of the posterior N2 was significant at the occipital and temporo-occipital sites for the controls, but not for the patients. Because of the eye movement artifacts of the patients in the feature-absent condition that particularly contaminated difficult trials with an

H.-J. HI 1x71 r’l [ii

CONTROLS

PATIENTS

Feature

present

Feattire

Popou

eccentric amplitude the patient not in the

absent

t

target position, P3 latenq analysis was not performed. Measurement of P3 confirmed significant ditTcrences bctwecn positive and negative trials both ivithin and control group, while signiticant task effects were observed in the controls. but patients.

Bcl~tr~+or~r/IIJCWSIIWS. RTs and hit rates (Table 3) exhibited signiticant dift‘crcnccs bctwcen patients and controls (RT: P
VISLAL

SEARCH

CONTROLS

-

~LOStl~

HEAD

IhJLKY

505

PATIENTS

_ lPSl

(HEOG

-----

AFTER

left)

co”Cra (HEOG rlght)

5w +

I:

400 I

msec : I

Fig. 4. Grand average ERPs m patients (right) and controls (Icft) to the positive stimuli m the feature-present condition of Experiment I Trials were selectively averaged on the basis whether the target was contralateral (dashed) or ipsilateral (solid) to the side of the electrode at the frontal and temporo-occipital sites, collapsed over hemispheres. For the horizontal EOG (HEOG), trials were divided on the basis whether the target was in the right (solid) or left (dashed) visual field.

Some results of statistical analysis are given in Table 5. As in Experiment 1, the posterior PI component did not significantly differ between groups, whereas the N 1 exhibited a highly significant difference. The frontal P2 and the frontal and central N2 effects were significant in the control, but not in the patient group. The P3 wave showed a significant stimulus effect within both groups. the difference between groups not being significant. Finally. both target and nontarget pop-outs produced a significant enhancement of the posterior N2 over the recording site contralateral to the pop-out location in the control, but not in the patient group. P3 latencies to target stimuli at the parietal sites (Table 3) displayed a significant difference between patients and controls (P~O.001). An important question is to what extent ERP components can be differentially related to neurospychological impairments. The data from both experiments indicate that the early components exhibit marked differences between patients and controls. In order to explore early and late ERP components in relation to attentional performance, a cluster analysis was performed based on the following tests from the San Diego Test Battery: digit span, block span. letter span, 2 & 7 selective attention (parallel). 2 & 7 selective attention (serial), digit symbol. Various cluster methods (average linkage between groups, average linkage within groups, complete linkage) and distance measures (euclidean distance, squared euclidean

CONTROLS

-

Negative

-----

Pasltlve

PATIENTS

distance) (SPSS PC+ advanced statistics) pro~idcd cvidcnce for the the cxistencc of two subgroups: subgroup I (patients 2.1.6.7. X) and subgroup 2 (patients I. 4.5.9, IO, I I ). (Note that this separation based on behavioral data is similar. but not identical with a dichotomy on the basis of severity of injury. as measured by the combination of PTA and prcsence.absence of Lvhite matter Icsions: subject 9, e.g. has about the same length of PTA as subjects 2 and 3. and has no Lvhite matter Icsions, but exhibited ;I poor attentional performance.) The io\ver division of Eig. 7 illustrates the separation between subgroup I and 7 (average linkage between groups method \vith squared cuclidcan distance) with :I better attcntional pcrformancc of subgroup I (KC Table 2). As indicated by the results of the Grooved Pegboard test, the attention4 deficits in subgroup 3 wnnot . bc solely cxplaincd on the basis of impaired motor pcrformancc. Also note that the results of the Figural F3uency test for nonverbal fluency move about the same for both groups. The EK Ps to the bar stimuli 01 Experiment 2 in subgroups I and _’ are shown in the upper division of Fig. 7. Subgroup I than subgroup 2. whereas the displayed ;I larger anterior N 2 and posterior N I component parietal P3 MQVCshowed a similar waveform in both subgroups. Statistical analysis rcvcaled that the amplitude of the central NZ was significantly reduced in subgroup 2. (P
507 Table 4. Experiment

ERP Parielal PI (60 120) NI (130 190) A I (300 400) A2 (400 500) P3 (350 7.50) Occipital PI (60 120) Nl (130 190) Al (300 400) A? (400 SOO) P3 (350 750) Tcmp-OK PI (60 120) NI (130 190) Al (300 400) A3 (400 500) P3 (350 750)

I : P-valuesof ANOVA

comparisons for ERP mean amplitudea indicated latency range (msec)

S

Controls (within) i-

SxT

S

Patients (within) T SxT

over

Between group> G GxSxT

0.02 0.00 I 0.04

0.00 I

_

0.05 0.005

0.001 0.001 0.00 I

0.003 0.001

0.003

0.002 0.00 I

~

0.04 0.005

-

_

0.03 0.001

0.003 0.001 0.00 I

0.002 0.001 0.001

0.003 0.001 ~~

_

0.03

0.005

-

_

_

0.04 0.04

ERP Pariclal N2 (‘20 280) Occipital N2 (220 2X0) Temp-occ N3 (220 2X0)

_

0.03 0.07

ANOVA factors: C =target item location (contralateral G =group (controls vs patients): T= task (feature-present (positl\c trial\ vs negative trials).

vs ipsilateral to recording site): v\ feature-absent): S =stimulus

The question arises whether these ERP differences result from an increased variability of ERP components in subgroup 2 compared to subgroup I. Evidently, these subgroups are not homogeneous with respect to various psychometric and neurological characteristics. as can bc seen from Tables I and 2. However, the across-subject variability of the N2 component within each subgroup is relatively low. In Fig. 8, the ERPs to the bar stimuli of Experiment 2 are shown for each subject of subgroups 1 and 2 at the central leads. While all the subjects of subgroup 2 exhibited a distinct N2 component (and most of them an Nl and P2 component). none of the subjects of subgroup I displayed an N2 (and a P2) wave.

DISCUSSION The present paper is aimed at clarifying the electrophysiological mechanisms underlying attentional deficits after CHI. By studying processes involved in visual search, these experiments examined to what extent impairments of parallel and serial processing contribute to attentional dysfunctions. In a modification of two paradigms that investigated ERPs during parallel and serial visual search in normals [ 18,201 ERP correlates of (parallel)

50X

H.-J

CONTROLS

HIIVA :‘I
PATIENTS

feature detection ~rnd (serial) feature conjunction were studied in CHI patients and in a matched control group. Roth experiments in the present study closely replicated the findings of the L~JCK and HILLYAKD [ 18, 201 paradigms in the control group. For the patients, the data provide strong evidence that CHI causes deficits of automatic parallel processing during visual search. In particular. Experiment 2 demonstrated behavioral and ERP differences between patients and controls in the ER Ps to easily detectable targets. In the controls. the detection of pop-out features w;1s indexed by :tn early frontal P2 and frontal :tnd central N2 component. Furthermore. a posterior N2 was observed contralateral to the side of both target and nontarget pop-outs. indicating discrimination of visturl objects on ;rn early stage of processing. in the patients, the frontal P2 and N2 ;md the posterior contralateral N2 component were absent or greatly reduced. The only sign of feature detection w;ts a sm:rll central N2 effect that did not reach statistical significance. The hit rate was also significantly lower in the patients than in the controls. It is an important finding that the earliest (PI ) and the latest (P3) components were less affected than the strongly reduced intermediate N 1, P2 and N2 components. These different effects of traumatic brain damage on ERP components became particularly evident by

SOY

Table 5. Experiment

ERP

2: P-values of ANOVA comparisons for ERP mean amplitudes indicated latency range (msec) Patients (wlithin) H SxH

Control5 (within) H

SxH

S

0.007

0.006 0.007

0.008

~ ~

~

~

0.02

0.001

0.002

0.001

~

~

~

0.02

0.001

~

-

0.001

~

~~

0.001

~

~~

s

over

Between groups

G

GxS

Front31 P? (180 240) N2 (280 360) Central P2 (1X0 240) N’ (2X0 360) Parietal PI (60 120) Nl (130 190) P3 (350 750) Occipital PI (60 120) NI (130 190) P3 (350 750) Temp-occ PI (60 120) NI (130 190) P3 (350 750)

ERP Parietal N2 (220 280) Occipital N2 (220 280) Temp-occ N2 (220 280)

0.02

0.04 0.02 0.03

0.002

~ 0.001

C

-

P

~ 0.008

~

0.001

~

Contralaterality CXP c

0.006

~

of the posterior P CXP

-~

0.001

N2 G

GXC

~

0.0 I

_~

0.01 0.01

ANOVA factors: C= pop-out location (contralateral vs ipsilateral to recording site): G=group (controls vs patients); H=hemisphere of recording (left vs right): P=pop-out (target pop-out vs nontarget pop-out); S=stimulus (standard, nontarget pop-out. target popout).

dividing the patients into two subgroups according to their neuropsychological performance. The fact that a reduction of the intermediate components is paralleled by a poor performance in a number of attentional tests suggests that dysfunctions of intermediate processing stages play an important role in attentional deficits after CHI. These processes are responsible for transforming basic sensory information and passing it along to later, high level processes. Among these processes is the discrimination of relevant features that may be impaired in the patients as indexed by the lack of a contralateral posterior N2 component. As outlined in the introduction, Rucc et ul. [30] also found differences between CHI patients and controls in the N2 component to target stimuli in an auditory oddball task. They interpreted this result as evidence of an increase in the time needed to achieve stimulus categorization and suggested that patients can cope even with a simple discrimination task only by the allocation ofadditional cognitive effort. The changes ofthe fronto-central and the controlateral posterior N2 component during parallel search in the present study underline the assumption that CHI patients suffer from deficits of elementary feature discrimination. The fact that the present study yielded a decrease and not an increase of the N2 amplitude as observed by RUGG rt al. [30] might be due to the diflerent task modality and to differences

Subgroup

1

Subgroup

2

P3/4

P3/4

+tizq

+qp ?. ‘.;.J

/

.........J~

L../

Resealed Distance Cluster Combine 0

5

10

15

20

25

Seq 9 5 1 4 10 11

-I

8 2 7 6 3

between the patients groups. In any cast both studies provide evidence that the cognitive slowing exhibited by CHI patients on RT tasks is not entirely the result of inelticiency in late processcs such as higher stages of stimulus evaluation and response selection. These conclusions are based on the assumption that ERPs in head-injured patients rellect cognitive processes in the same way as they do in normal subjects. However. one could argue that CHI causes a ditT‘usc damage to the brain tissue thereby changing the physical conditions of ERP generation. Accordingly. ERP waveforms after head injury might not adequately index information processing deficits. However. the fact that intermediate ERP components are particularly impaired while the earliest and latest components are less affected provides evidence that there is not ;I diffuse reduction of ERPs. but rather effects of C‘HI on distinct stages of processing. The results obtained in the feature-present condition of Experiment ! support the and the assumption of perceptual dysfunctions. As in Experiment 2. the Nl amplitude

Subgroup

Subgroup

1

2

l--zgz+ .. ----..........’

Fg. 8. ERR

ran-target Target

Pop-outs

1J; ; :

1opv

~O~;~S7C,

POD-Out5

lo the target pop-out (dotted). nontarget pop-out (dashed) and standard (wlid) stirnull at central electrode SII~S (collapsed over hcmisphcrcs) from each member of subgroup I and 2.

o~“Experim~~t2 recorded

posterior N2 amplitude in the patient group were significantly reduced. The finding that the early P3 measure between 300~~400 msec exhibited a significant difference between the feature-present and the feature-absent task in the controls, but not in the patients. also suggests a relative impairment of automatic, parallel processing in the patients. Furthermore. there is some evidence that the Pl amplitude may also be reduced in CHI patients. PI amplitude differences between groups were significant in the feature-present condition of Experiment I, but not in Experiment 2. It might be that changes on a very early stage of visual processing also contribute to dysfunctions of automatic feature registration. llnlike the ERP findings, the correlation between anatomical (CAT, MRI) findings and neuropsychological performance is less clear. However, deep and subcortical white matter lesions are seen only in patients with a poor attentional performance, whereas cortical lesions and ventricular enlargement are found both in subgroups 1 and 2. While the number of subjects is small, these data might suggest a correlation between intermediate processing dysfunctions and white matter lesions. Spontaneous-EEG pattern. on the other hand. did not show any correlation to neuropsychological outcome. As outlined in the introduction. a number of RT experiments in CHI patients provided

evidence that a reduction of processing resources is a main factor in CHI-induced cognitive impairments 1401. In the present study, patients displayed a poor performance in the featureabsent condition of Experiment 1. pointing to impairments in serial. capacity-limited starch. The hit rate was strongly reduced in the patients compared to the controls, and the significant group x task interaction indicated stronger impairments in the serial task than in the parallel task. (The fact that RT of target detection LVLIS more delayed in the feature-present than in the feature-absent condition in the patients compared to the controls seemed to contradict this interpretation. However, reanalysis of the data showed that in the feature-absent condition. patients responded mainly to those targets that wcrc placed near the fixation point and therefore could be detected quickly and easily.) In addition. ER P analysis rcvcalcd differences between patients and controls in the intcrmcdiatc as well as in the later components: the N 1 amplitude in the patients was rcduccd. and the separation between positike and negative trials was more delayed in the feature-absent than in the feature-present condition. T~vo possible explanations might account for these findings. First. CHI may cause dysfunctions ofserial processes that are required for feature coiljunction in the feature-absent condiGon. By this interpretation, parallel as ~vcll as serial proccsscs arc impaired by (‘HI. Altcrnati\.ely. or in addition. CHI might act upon processes that mediate both parallel and serial proccsscs. This assumption implies that some of the processes that arc used during parallel search may also be used for identifying objects during serial starch: since man> ob.jects must be identified in each stimulus array in the serial starch task and only one must be identified in the parallel search task. the cffecty of impairments in thcsc processes would bc repeated for each item in the array during serial starch, producing ;I larger owrall elrect. The results of the L~.cx and HII I V\III) studv [20] and their replication in the prcscnt cxpcrimcnts in the control subjects suggest that ca;ly~fcaturc discrimination is in\olvcd both in parallel and serial search as indexed bq the contralatcral posterior N1 component that is present in both the feature-present and feature-absent conditions. Therefore. the lack of this component (and the reduction of the N I wuw) in the patient group in both condi!ions might suggest that CIHl causes impairmcn~s of proccsscs that mcdiatc parallcl as well as serial proccsscs. From the prescnl cspcriments it cannot be decided exactly to what cxtcnt (‘HI-induced deficits of parallel and serial search are based on dysfunctions of common or scparalc proccsscs. In any USC. the data provide stron g e~idence that abnormalities of perceptual proccsscs contribute to impairments of both capacity-free feature detection and capacitylimited feature col?junction. These impairments might constitute important factors that underiy the general cognitive slo\vness as ;I core symptom of attentional deficits after (‘HI.

4. BINDER, L. Persisting symptoms after mild head injury: a review of the postconcussive syndrome. J. c/irl. r\-p. Neuropsychol. 8, 323 346, 1986. 5. CAMPBELL, K.. HOULE, S., LOKKAI~. D., DFACON-ELLIOT, D. and PROULX, G. Event-related potentials as an index of cognitive functioning in head-injured outpatients. In Cerehrtrl Ps~choplz~siolog~: Srudies in Ewnt-Related Po~en~iuls, (EEG Suppl. 38). W. C. M&AL.LUM, R. ZAPPOLI and F. DENOTH (Editors). Elsevier. Amsterdam, 1986. 6. CUKKY. S. H. Event-related potentials as indicants of structural and functional damage in closed head injury. Proyr. Brain. Res. 54, 507~ 515, 1980. 7. DONCHIN, E., KARIS, D., BASHOKE,T. R.. Coots, M. G. H. and GKATTON. G. Cognitive psychophysiology and human information processing. In Ps~~hoph~sioloyy: System.\, Processes and ,4pp/icorions, M. G. H. COLES, E. DONCHIN and S. W. POKC.ES(Editors), pp. 244-267. Guilford Press, New York, 1886. X. DUVCAN. J. Visual search and selective attention. In Allenlion und Pwfbrmunc~r XI, M. I. P~SN~R and 0. S. M. MARIN (Editors), pp. 85-106. Erlbaum. Hillsdale, NJ. 1985. 9. GKONWALL, D. Advances in the assessment of attention and information processing after head injury. In Nrurohelrurioral Rrc~ocery fk)m Heud Injury, H. S. LEVIN. J. GRAFMAN and H. M. EISE~BER(; (Editors). pp. 355 371. Oxford University Press, New York. 1987. 10. HAKTCK, M. R. and AINE. C. J. Brain mechanisms of visual selective attention. In l’trrirties 01 Arrenrion, R. PAKASUKAMANand D. R. DAVIES (Editors), pp. 292 321. Academic Press, London, 1984. 11. H~IN~E. H. J., LUC.K, S. J., MAN~;IJN. G. R. and HILLYARD. S. A. Visual event-related potentials mdex focused attention within bilateral stimulus arrays: I. Evidence for early selection. El~crroc~n~rphal~~~~r.c/in. Neurophy.~io/. 75, 51 I 527, 1990. 12. HILLYAKI), S. A. and PICTON, T. W. Electrophysiology of cognition. In Hundhod of‘ Ph~sioloy~ Higher Functions o/ t/w Nrrrou.s System, Seclion I: TheNrrrousS~xrem, lid. I’, Hi&v Functions ofrhu Bruin, Parf 2. F. PLUM (Editor), pp. 519-584. American Physiological Society, 1987. 13. HII.LYAKU. S. A.. MANC;UN, G. R., LUCK, S. J. and HFIN%~, H. J. Electrophysialogy of visual attention. In Mtrc,hinrry o/ l/ze .Mid, E. ROY JOHN, T. HARMONY. L. S. PKICH~.~’and P. A VALDF.S-SOSA(Editors). pp. 186 205. Birkhauser, Boston. 1990. 14. JULESL. B. Towards an axiomatic theory of preattentive vision. In D~tmmic A,\pwt~ ~~f’Nr~~rric~u/ Function, G. EI)I:LMAN, M. COWAN and M. D. GALL (Editors), pp. 120 148. Wiley, New York. 1984. 15. KUTAS, M.. MCCAKTHY, G. and DONCHIN, E. Augmenting mental chronometry: The P300 as a measure of stimulus evaluation time. Scicww 197, 792 795, 1977. 16. LEZAK, M. Nuurops?cholo!lic,ul Assc~s.smr~n~.Oxford University Press, Oxford, 1983. 17. LF~IN, H. S.. AMPARO. E.. EISFNHEKC;,H. M.. WILLIAMS, D. H., HIGH. W. M.. McA~v~rr. C. B. and WFINFK. R. C. L. Magnetic resonance imaging and computerized tomography in relation to neurobehavioral sequelae of mild and moderate head injuries. J. Nrurosury. 66, 706 713. 1987. 18. Lu(.K. S. J. and HILLYARI). S. A. Event-related potentials to visual “pop-out” stimuli. Socklyfbr ,~c,uro,s~.ierl‘,r .4hsrrtrcrs 14, 1013. 1988. 19. L1~c.l;. S. J.. HFIN~.F, H. J., MAF;~;uN. G. R. and HILL~ARI). S. A. Visual event-related potentials index focused attention within bilateral stimulus arrays: II. Functional dissociation of PI and Nl components. Elrctroe,lcep/ztr/o~/~. dig. Nrlrroph!,siol. 75, 528 -542. 1990. 20. LUCK, S. J. and HILLYAKI). S. A. Electrophysiological evidence for parallel and serial processing during visual search. Percept.P.~,rc/toph~.s. 48, 603 ~617. 1990. 21 MARR. D. C’isron: .4 Compututio~wl Inwsrigurion into the Hw~m Repwsentutio,l uml Processirtgl of l’is~trl In/;uwurion. Freeman, San Francisco, 1982. 22. MCCALLUM. W. C. and CUMMINGS, B. The effects of brain lesions on the contingent negative variation in neurosurgical patients. N~~~~oY)I(.c~I~u/oJ/T. din. Nutrrophysiol. 35, 449 456. 1973. 23. MCCARTHY. G. and DOYCHIN. E. A metric for thought: A comparison of P300 latency and reaction time. scienc~r 21 I, 77 80, 1980. 24. PAI’ANI(‘OLAOU.A. C.. Ltviu, H. S.. Ei~~rzat~c;. H. M., MOOKL. B. D.. GOETHE, K. E. and HIGH, W. M. JR. Evoked potential correlates of posttraumatic amnesia after closed head injury. J. Neurosurq. 14,67&678. 1984. 25. Posur:~. M. I. Selective attention in head injury. In Nrurohrku~iortri Rrcowry,fk~m Hetrd Injury. H. S. LF~IN. J. GRAFMAN and H. M. EISFNBERCI(Editors). pp. 390- 397. Oxford University Press. Oxford, 1987. 26. R~zzo. P. A.. AMA~IL~, G., CAPOKAL., M., SPARAUO. M., ZANASI. M. and MORO(.UT~I. C. A CNV study in a group of patients with traumatic head injuries. E/e~rrool~rpht/~j~/r. din. Nruroph~siol. 45, 281 -285. 1978. 27. RL.FF. R. M. Automatic detection vs controlled search: A paper-and-pencil approach, Perwpt. Mar. Ski//s 62, 407 416, 1986. 28. Rrk~. R. M. and WYLII:. T. F. Tile Strrl Dircqo N~urop.\~~l~~~l~~~~~~ Tesf Btr/rery. ild,,linistrtrrio,t hfu~~ud. University of California at San Diego. Head Injury Center. San Diego, 1986. 29. R~:FI,. R. M.. LIGHT. R. H. and EVANS, R. The RuR Figural Fluency test: A normative study with adulls. &r/ Nulrrop.\!,c,ho/. 3, 37 5 I. 1987. 30. Rucic;. M. D.. COWA\. C. P., NAC;Y. M. E.. MILNI:R. A. D.. JA(‘OBSFN. I. and BROOKS, D. N. Event-related

‘Press. Oxforj.

i9X7