icy s~dho ., odle. x',1
. No . I . pp 13 2`. IP%
.3?
Cp,rigti ( 1994 . Cheder Sci .ncc Lid
Prated in Great Britain . All righrn reared
W2s-39P ;94 $6
WI Dlla

Pergamon

FOCAL AND GLOBAL ATTENTION MODULATE THE
EXPRESSION OF VISUO-SPATIAL NEGLECT : A CASE STUDY
PFTFR W . HALLIGAN * an
R ivermea

JOHN C . MARSHALL,

Rehabilitation Centre . Oxfor OX I 4XD : an
Neuropsychology Unit . University Department of
Clinical Neurology, The Ra cliffe Infirmary, Woo stock Roa , Oxfor OX2 6HE, U .K .
e 7 Mar 1993 : accepte
(Rece

12 August 1993)

Abstract-We report a case of severe left visuo-spatial neglect consequent upon right-hemisphere
stroke . In Experiment l, we show that while the patient bisects horizontal lines significantly to the
right of true centre, he is extremely accurate in placing a mark at the centre of an equivalent outline
square. Experiment 2 confirms that the lateral an ra ial accuracy obtaine in the latter (square) task
cannot be pre icte from the lateral an ra ial isplacements foun in horizontal an vertical (ra ial)
line bisection : the experiment also shows that the in ivi ual horizontal lines of a square can be
bisecte significantly more accurately than the same lines presente in ivi ually . Experiment 3 shows
that the crucial figural component requisite for the accurate estimation of lateral extent is a large
vertical (ra ial) line in right space . The obtaine figural mo ulation of neglect is consistent with a
mo el in which a large, unitary stimulus in right space engages an enhances the lesione right
hemisphere's intrinsic capacity to eploy global attention .

INTRODUCTION
IN PREVIOUS stu

ies [5, 16], patients with left visuo-spatial neglect have been reporte to
show a very specific effect of figural mo ulation : Although they bisect horizontal lines
significantly to the right of true centre, the same patients are remarkably accurate in placing a
mark at the centre of squares with the same horizontal magnitu e . We now report another
patient who shows this effect an we explore sonic of the sufficient (an perhaps necessary)
stimulus con itions for the effect to emerge .
'The esign an interpretation of the experiments erive from two (conceptually) istinct
properties of normal hemispheric specialization . First, it is well-known that each cerebral
hemisphere is preferentially biase to orient an atten to contralateral space [8, 10] . These
biases are not absolute but rather manifest as "attentional gra ients" [9] . Secon , the
hemispheres iner in terms of the `spatial scale" at which they prefer to operate [12, 15] . The
left hemisphere respon s preferentially to local features of a stimulus array an the right
hemisphere to more global properties thereof . Again, it is unlikely that these biases are
absolute (within a "fixe ban "), but the relative preferences of the hemispheres for operating
at ifferent spatial scales seem well-establishe [14] .
It is the combination of these two in epen ent properties of the cerebral hemispheres that
pre icts that "global cueing" shoul engage an boost the ( amage ) right hemisphere's
capacity to eploy global attention an thereby overcome (at least partially) the (intact) left
hemisphere's focal rightwar bias [3, 6] . It is the overt manifestations of the latter bias that
have been tra itionally labelle "left visuo-spatial neglect" .
13

P . W . HAII10AN an

.

C . MARSHAL[

CASE REPORT
R .B . is a 61-year-ol , right-han e man who sustaine a right hemisphere stroke on 17
December 1992 . CT scan reveale an extensive right mi le cerebral artery infarct, which
involve primarily right temporal an paricral cortex without evi ence of any left
hemisphere or subcortical involvement (Fig . I ) . On a mission to Rivermea 3 months after
his stroke . B .B- ha a ense left hcmiplcgia, left sensory loss, an left homonymous
hemianopia with macular sparing . Visual acuity was 6,14 in each eye (glasses correcte ) . Full
occular movements were possible in all irections . Sacca es were hypometric right to left but
accurate left to right ; visual pursuit was somewhat jerky .
The outstan ing neuropsych ologicaI symptoms were emotionalism an severe left visuospatial neglect . Assesse on the Behavioural Inattention Test (BIT), B .B . score 45/146 ; the
curt-off for normality is 130/146 [4J . B .R . was always well-oriente in time an place : he was
aware of the physical effects of his stroke, but was not consistently aware of his left neglect .
The experiments that follow were con ucte over a 6-week perio , beginning 3 months
post-stroke . The magnitu e of B,B .'s teste neglect was not stable over the course of these
stu ies but coul vary quite substantially even within a ay . We accor ingly ensure the
vali ity of the results from our experimental manipulations by taking a new base-line
measurement of horizontal line bisection at the beginning of each experiment .

EXPERIMENT I
The purpose of this experiment was to compare the accuracy of estimations of the lateral
mi points of horizontal lines an squares .
Alerlu a
Horizontal lines of 180 in were rawn in black ink on sheets of horizontally-oriente white A4 paper (29N x 205
mm), one line per sheet . the wi th of each line was I mm . Outline squares, 180 x 150 mm, were likewise rawn on A4
paper . Each stimulus was centre on the page, both horizontally an vertically . Each stimulus sheet was place on
the esktop imme iately in front of the patient an was always centre laterally on the mi sagittal plane of the
patients hea an trunk . B .B . was requeste to bisect each line, an to placea small ot at the cent re of each square .
Responses were ma e with a line pen hel in the (preferre ) right han . All stimuli were isplaye in free vision,
without restraint on hea or eye movements . Although no time pressure was applie , each response was ma e
within 2 or 3 sec of stimulus presentationStimuli were blocke in the following or er of presentation : Lines In-5),
squares (n-5) . lines (n-5) an squares (n-5): that is . in the esign ARAB, where A is lines an B squares .

Results

the istance of each response from the objective mi point was measure to the nearest
millimetre, an expresse as minus (-) for leftwar vs plus (+) for rightwar isplacements .
For the squares, both horizontal (lateral) an vertical (ra ial) isplacements were calculate :
minus in icates placements too close to the hotly an plus too far front the bo y . The results
(mean isplacement an S .D .) for the four blocks are shown in Fig . 2 .
One-way ANOVA (with repeate measures on the single factor of horizontal isplacement) shows that the results from the four blocks are not rawn from the same istribution
(F=17 .4 : .f.3 .16 ; P<0 .0001) . Planne comparisons were then evaluate with the
(conservative) Schefle F-lest, with significance set at the 95% level . There is no significant
ifference between A, an A z or between B, an B e . A, an B, are significantly ifferent
(F=9 .04) as are A, an B . (F=8 .3) . Similarly, A, an B, are significantly ifferent (F=7 .4),
as arc A, an B, (F=9 .9) .
As reporte in previous cases of left neglect [5, 16], B .B, is substantially more accurate

FXI'NF'SSION OF VISUO-SPA I 1 AL NFGLE( - I

Fig . J . CT-scan of patient B .B . . showing infarction of the right mi
on the right of the image) .

le cerebral artery territory Ileft it

I XPRFSSIU\ IIF VISt,O-SPAIIAI . NHji .I ;( :r

I

AI

+26 2mm
(5 .4)

H - .2 .Bmm
(11 .3)
B

1

V = -2 .nmm
(1-8)

A

27 .4mm
(6 .8)

2

H = +5 .emm
(5 .8)
B

T
150mm

2
V - -2 .4mm
(2 .9)
E

Fig . 2 . Lateral lan

180mm ->

ra ial) isplacements IS .D .) for horizontal lines (an squares) in Experiment 1 .

with squares than with horizontal lines, where his transections are grossly outsi e normal
limits [I I] .
EXPERIMENT 2
This experiment replicates Experiment I an a s two further con itions : bisection of
vertical (ra ial) lines, an the bisection of all four lines that make up an outline square .
Metho
The stimuli were presente in four blocks of 10 trials each . Or er of presentation was (A) horizontal lines . (D)
vertical lines. (Cl ou inesquarres with each line to he bisecte , an (D) outline squares for placement ofa ot at the
centre . The relevant imension . or imensions . of the stimuli were again ISO man . In con ition C-the instructions
requeste that the patient begin with the top horizontal line an continue in a clock-wise irection . In all other
respects, the metho ology is i entical to that employe in the first experiment Experiment was run ones an thenn
repeate 6 weeks later .

Results

The results from each session, measure an score as in Experiment I are shown in Fig . 3 .
One-way ANOVA (repeate measures on the single factor of horizontal isplacements)
shows that the results from the four con itions in the first session are not rawn from the
same istribution (F-26 .2 : .f . 3, 36 : P<0 .0001 ) : the same conclusion hol s for the secon
session (F=35 .4 ; .f. 3, 36 :: P<0 .0001). Planne comparisons (Scheffe f-test) shows that
bisections of the top horizontal of a square are significantly more accurate than bisections of
a single horizontal line in both sessions (F=24 .4 an 28 .2, for the first an secon sessions,
respectively) . Bisections of the bottom horizontal of a square are only significantly better
than bisections of an in ivi ual horizontal line fit the first session (F= 13 .2) . In both sessions .

P . W . Hau-IGA\ ari

A

J . C . MNR51 ALL

C

B
2 .7mm
(6 .7)

+20 .0mm
(6 9)

D

-8 .8mm
(9 .3)
1

H- r4 .5rnm
(4 .9)
+8 .Omm
(5 .3)

-1 7mm
(6 .0)

V . -5 .1 mm
(2 .0)

Session (i)
-0 .7mm
(8 .2)
-1 .7mm
(8 .9)

+3 .5mm
(9 .6)
+24 .8mm
(51)

H = +0 .1 mrn
(8 9)
+5 .7mm
(3 .4)

.8 .2mm
(54)

V =. -6 .Omm
(4 7)

Session (ii)
+10 .0mm
(3 .9)
Fig 3 . Lateral an

ra ial

isplacements IS J) .) for lines an

square, in Experiment 2 .

placing a ot at the lateral centre of a square is significantly more accurate than in ivi ual
horizontal line bisection (F=7 .4 an 4 .5) . Bisection accuracy for the top an bottom
horizontals of a square is significantly ifferent only in the secon session (F=24 .6) . In
summary, the ata comparing single horizontal lines to squares provi e a goo replication of
the results of Experiment 1 . Bisecting the horizontals of a square also yiel s a significantly
more accurate performance than bisecting in ivi ual horizontal lines . There is a suggestion
that vertical bisections may iffer between single lines (con ition B) an the two verticals of a
square (con ition C) . Likewise, there seem to be ifferences between accuracy on con itions
C an D . Con ition D (placing a ot at the centre of a square) is the purest figural task an
accor ingly yiel s the best performance (although the reliability of this result requires further
stu y) .
PrehininaTV

iscussion

The results confirm previous reports that (some) patients with left visuo-spatial neglect can
estimate the lateral mi points of squares with consi erably greater accuracy than their
bisections of horizontal lines woul suggest . B .B .'s performance shows that this substantial
amelioration of left neglect when squares are presente can persist even when the task
eman s that the two horizontals of an outline square be bisecte .
The results are consistent with the notion that accurate perception of the lateral mi point
of a square is etermine by some "emergent property" of the figure [I] . Whatever figural
property is involve is clearly of sufficient "power" to over-ri e the gross left neglect seen on
the bisection of in ivi ual horizontal lines . In Experiment 3 . we accor ingly begin to explore
some of the minimal global cues that suffice to counteract the extreme rightwar bias foun
when in ivi ual horizontal lines are bisecte .

LxpRiSSIIN OF VIsL0-SPATIAL Nat 1 Ft i

19

EXPERIMENT 3
In this experiment, we again compare the accuracy of ju gements of the lateral mi points
of horizontal lines an outline squares, but a
10 con itions in which the figures are
subparts of squares .
Aloe 1,0(1

The stimuli consist of horizontal lines 1180 MIT If an squares (180 x 180 mml, plus every pair an every triple of
180 mm lines than can he regar e as subparts of (180 x 180 mm I squares . The experiment commence with the
bisection of a single horizontal lint (10 trials), followe by placing a ot at I he centre of a square (10 trials) . All pairs
an triples of lines that form subparts ofsquares to 10) were then presente (blocke ) in pseu oran om or er (It)
trials per block) . The instructions were to place a ot at the centre of the ("imaginary'') close figure that the lines
outline . The experiment conclu e with a repeat (10 trials) of the true square (i .e . the figure in which all four lines
were physically present) . In all other respects the stirnu Ii an metho s are i entical to thuse employe in the previous
experiments .

Results
The results are shown in Fig . 4 .
+31 .4mm
(16.2)
H - +38 .3mm
(9 .9)
V - -2 .0mn,
(2 .6)

H = +12 .5mm
(8,4)

H - +135mm
(48)

V = -8 .0mm
(6.1)

V = -4 .2mm
(4 .2)

H - +12 .6mm
(5 .8)

H -+42 5mm
(44)

H - +14 .Omm
(24)

(6 .6)

V = -7 .8mm
(44)

V = TO 7,11m
(2.8)

H = +41 9mm
(3 1)

H = +13 .9mm
(5 3)

H - +12 .1mm
(3 .8)

V= .13mm
(40)

V - -3.1 mm
(4 .8)

V = .0 .3mm
(3-9)

V = 2 ,0,,,,

2n

Ist set
H - 0 .3mm
(60)

H =

(2 7)

H = -0 .6mm
(3 3)

V = -4 .2mm
(3 .6)

V = -2.9mm
(36)

V = -5 .8mm
(4 .8)

Re 4 . Lateral an

ra ial

24 firm

set

isplacements (S .D .) for the stimulus configurations in Experiment 3 .

One-way ANOVA (on the factor of horizontal isplacements) shows that the results
across the 13 blocks are not rawn from the same istribution (F=45 .4 ; .f . 12, 117 :
P<0 .0001) . Scheffe F-tests confirm (at the 95% level) what is apparent by eye : All stimuli
that contain a right-si e vertical line yiel (lateral) isplacements that are significantly more

P . W . HkLLIC61N an

J C . MARSHALl

accurate thin those obtaine in any stimulus that o not contain a right vertical . The two
lines that make a bottom right corner yiel
isplacements that are significantly less accurate
than those foun in the secon block of squares (F= 1 .8) : the three lines that form a bottom
left an bottom right corner give isplacements that are significantly less accurate than those
foun in the secon block of squares (F=1 .3) . Displacement on the pair of horizontal lines
(top an bottom of a square) is significantly more accurate than on any figure that oes not
contain a right vertical . Accuracy on this configuration oes not iffer significantly from any
stimulus containing a right vertical other than the true squares (F-5 .2 an 5 .3) . No other
comparison across the set of results is significant .
In summary, the component of a square that is primarily responsible for the accuracy of
ju gements of the lateral mi point thereof is the right vertical . On a slightly less stringent
criterion for significance . a true square (all four lines) elicits more accurate responses than
any other stimulus with a right vertical (the improvement is circa 13 mm) .
It can be observe , however, that, in all three experiments, vertical (ra ial) isplacement is
consi erably smaller than (rightwar ) isplacement on (single) horizontal lines . It is possible
that R .B . coul have assume (or "seen") that in all figures of two or more lines, horizontal
extent was i entical to vertical (ra ial) extent . Accor ingly, he coul , in principle, have taken
the estimate mi point of any vertical line in right space, an then, from that point, move
leftwar by the half length of the vertical . Clearly, this "strategy" woul yiel lateral mi point
estimations that were as accurate as vertical (ra ial) mi point estimations . The hypothesis
pre icts a significant correlation between (unsigne ) lateral an ra ial isplacements . We
accor ingly calculate this parametric correlation across all pairs of lateral an ra ial
isplacements for all the figures in Experiment 3 that have a right vertical (n-80) . The
correlation obtaine was r-=0 .07 : the hypothesis is false .

GENERAL DISCUSSION
The expression of left visuo-spatial neglect in B .R, shows an exquisite sensitivity to figural
mo ulation . On the bisection of in ivi ual horizontal lines . B .B .'s (mean) rightwar
isplacements arc always outsi e normal limits (although there is substantial variability
from session to session in the magnitu e of this effect) . Nonetheless . when squares are
presente , R .B .'s ability to place a mark at the centre thereof is remarkably accurate . Even
when the task is to bisect the horizontals of a square figure . this improvement persists ; the
very presence of a square sulTices to eliminate (or consi erably re uce) the expression of
(otherwise severe) neglect .
The major source of the mo ulation effect is the presence of a vertical line as a figural
component on the right of the stimulus pattern (Experiment 3) . This suggests that an
exten e vertical (ra ial) couipunent on the right can engage an crilianee the ( amage )
right hemispheres intrinsic capacity to eploy global attention an hence see the figure "as a
whole" . The result is not epen ent upon an explicit transposition ol'the vertical mi point
into the horizontal imension, as witness the correlation of essentially zero obtaine between
lateral an ra ial accuracy in Experiment 3 . Rather, it is global attention that counteracts the
(intact) left hemisphere's (intrinsic) ten ency to orient rightwar s with the subsequent
"capture" of focal attention by that part of the stimulus array which is in the extreme right
position . It is this latter process that is most in evi ence when solitary horizontal lines are
bisecte : focal attention is capture by the local right termination of the line .
The etaile mechanisms involve must be eluci ate by further experimentation an

EXPRESS It SN OF VI5L'rl-St's'IIAt AF

Ill 'T

simulation [2 . 7, 13] . It is possible, for example, that the global amelioration of left neglect is
actually me iate by the (intact) left hemisphere . A global cue (in right space) may prevent
the left hemisphere from focusing- own to the small scale that it woul otherwise a opt to
resolve the right termination of a (thin) line . An interpretation of this nature woul be
consistent with the failure to fin any significant effect ue to the presence of a vertical line in
left space (Experiment 3) .
Both left neglect an global processing eficits are common among patients with right
brain amage . Nonetheless, our results clearly show that explicit global cueing can
counteract both impairments .1 he precise attentional an /or perceptual eterminants of the
current effects remain to be etermine , as o the respective contributions of the two
hemispheres .
Aelowrole pcments-'this work was supporte

by the Me ical Research Council an

the Stroke Association .

REFERENCES
I . BLrILI :R, K . Die Gestaltwohr nehman9erl . Spemann, Stuttgart . 1913 .
2 . CRAVEN, B . J . an WA ri, R . J .'I lie use offractal image statistics in the estimation of lateral spatial extent
. Spatial
Vision 4, 223 239 . 1989 .
3 . Dn .ls, D . C ., Ronutisox- L . C . an BALLIET. R . The break own an relmbilitation of viauospatinI asl'unetion
in brain-injure patients . Ins'rootionof Rehabilitation tic ieine 5, 132-138, 1985 .
4 HAI .LICAN, P. W . . CotKBURN, J . an
WILSON B . A . The behavioural assessment of visual neglect .
Nenr'ops'choloyicul RelmhilHmion 1, 5 32, 1901
5 HAt lit'AN . P. W . an
M,v vIAI .t, J . C . Figural mo ulation of visun-spatial neglect A case stu y.
Nemvpsyehnloyia 29, 619 628, 1991 .
6 H SLLIUAN . P. W. an MARSUnLI ., J . C . Homing in on neglect : A case stu y of visual search . Corte.x29,
29,167 174,
1993 .
HuMrearvs, G . W . an Rumor'e, M . 1 . Interactive attenlional systems an unilateral visual neglect . In
t-ililaerai A'eghri .I . H . Rour:R soN an J . C . MARS]tAtL IE itors) .pp . 139 167 . Lawrence Erlbaum, Lon on .
1993 .
8 . KINSUOUxvt, M . The cerchral basis of lateral asymmetries in atention . Am, P--vrchoI . 33, 193 201 . 1970 .
9. KINSBOUR'r, M . Oricnlational bias mo el of unilateral neglect : Evi ence from attentional gra ients within
hemispacc .In l4 . .fal, alNeglecr,1 . H .ROHIKistNan ,I .C .MARSHALI .(E itorsI .pp .63-86 .Frlhaum,Lon on .
1993 .
1(t
. LUCK . S J-, I IILLYAIto, S- A . . MANGUN . G . R. an GAZ.ZANIGA . M . S- In epen ent hemispheric altentional
systems me iate visual starch in split-bruin patients . Nature 342, 543. 545, 1939 .
I i MANNING, I_, HALLI(iAy, P . W, an MARSH4LL, J . C . In ivi ual variation in line bisection : A stu y of normal
subjects with application to the interpretation of visual neglect . .V'eu ropsvchololliu 28 . 647 55, 1990 .
12 . MARIAN, M . Hemispheric specialization for local an global processing . Neurops)cholnyia 17 . 33-40, 1979 .
13 . Mozns_ M . C- The Perception of Multiple Objects . MIT Press, Cambri ge, MA, 1991,
14 . RonLRrsoN L . C' .an DFLIS . D. C . 'Part-whole' processing in unilateral brain- amage patients : Dysfunction
of hi erarchical organization NeuroPercNalopis, 24, 363 370, 198615 . ROBERTSON, L. C . an LAMB
. M . R . Neuropsychological conirihutions to part-whole organization . Cotpnit .
Psrchol . 23 . 299 332, 1991 .
16 . 'frkea ri, R . an I .I:VANOIR . M . The inlluenee of stimulus properties on visual neglect . J. A'vurof . A'earos og an
P,t,i...it, 54, 882 887, 1991 .