TO HALVE AND TO HALVE NOT: AN ANALYSIS OF LINE BISECTION JUDGEMENTS IN NORMAL SUBJECTS A. D. MILNER, M. BRE~HMANN and L. PAGLIARINI Psychological Laboratory. Universit) of St. Andrew.
Fife, U.K
Abstract-Normal adults were tested in a series of three experiments to examine the influences of spatial location and cueing upon line bisection judgemcnts. Judgements in all three experiments were strongly influenced by cueing with a letter at one or other end of the line. The spatial location of the line (in left, central or right body space) also had a minor effect in Experiments 1and 2. where evidence was additionally found for a small constant error when lines were presented centrally. It is argued from the results of Experiments 2 and 3. where no explicit bisection response was required. that perceptual,‘attentional factors, rather than an orienting bias. play the major role in mediating the cueing elTcct. It is concluded that thcrc is a substantial attentional effect upon judgements of extent. whereby paying less attention decreases perceived relative line length. However the constant error and the effect ofspatial location, whilst mainly perceptual in nature. may also be partly determined by prcmotor orienting biases caused by differential hemispheric activation.
INTRODUCTION A FREQUENT FINDING in patients
with posterior right hemisphere lesions is that they make rightward errors when asked to bisect a horizontal line centrally (e.g. Ref. [22]). There are broadly two reasons why a patient might make such errors. One school of thought [l l] has it that patients make a pre-motor error, because of a leftward “directional hypokinesia”. This is thought to weaken the leftward vector in any spatially directed action, thus causing a net rightward error. The other school of thought places the error on the per~ptual side and suggests that patients underscale the leftward extent of the line (e.g. Ref. [15]), perhaps as a result of paying inadequate “automatic” attention to it [21]. Attempts have been made to decide between these two broad alternatives by examining the effects of manipulating (a) spatial location and (b) cueing, on the bisection behaviour of neglect patients. The results indicate that the first of these variables strongly affects bisection performance, with rightward error increasing when lines are placed in the left half of body space [I 1, 17,221, and decreasing when in the right half. However, this result does not decide between the two hypotheses, in that either attentional salience w pre-motor facilitation might decline progressively as a function of leftward location in space [6]. The second manipulation is to direct attention to the ends of the line to be bisected; letter cues are typically provided for the patient to report prior to bisection. The results are conflicting on whether cueing hiltrtertrlly causes any amelioration of the rightward error [ 11,211; however unilateral cueing on the left is consistently found to reduce the error [lo, 17, 19, 211. This result would appear to support the attentional theory, whilst not precluding the additional
infuence of a pre-motor factor, which would indeed account for the fact that cueing does not uniformly abolish the rightward error. In recent studies in this laboratory [lb] we have acquired further evidence for ;I perceptual/attentional factor in causing rightward bisection behaviour, by use of 21 “landmark” task (cf. Ref. [I 81). We have observed in three patients showing rightward bisection that a centrally pre-transected line is judged by them to be marked to the /e/i of UYH?VP. This indicates a leftward underestimation of the line’s length. Indeed. since they were asked to poir~t to the end of the line nearer to the landmark, the patients were pointing in ii direction (leftwards) oppcaifr to that predicted from any presumed directional hypokinesia. Other recent experiments 12, 6. 10, 131 provide similar support for the conclusion that ;I perceptual.attentional effect operates in causing rightward bisection responses in most if not all neglect patients. though there is also evidence for a response-based component which may e~cn predominate in some individuals [2, 6. 141. If a disturbance in the perception of line length is indeed the major cause of bisection errors in neglect patients. could such a perceptual disorder be directly attributed to a 11 asymmetrically-biased distribution of I isual attention’? If so, then one would predict that in normal subjects too the distribution of spatial attention should have difrerential elects 011 bisection judgements. There is now evidence that subjects do make bisection errors towards whichever end of a line is cxplicitlq cued [7, 17, 191. Thus they may indeed perceptually overestimate lint length on the side to which attention has been drawn. relative to the otha side. The present experiments \vcre designed to examine further the nature of this cueing effect on the bisection judgements of normal subjects, and also to re-examine the cffcct of spatial location. In general. the bisection errors of normal subjects are smaller in magnitude and if anything opposite in direction. relative to those of right-hemisphere patients [4]. This tendency to err leftwards in normal subjects is generally attributed to a right-hemisphere dominance in determining bisection responses, but this could be for either of two reasons [3]. The dominance could be conceptualized, once more, either in perceptual:attcntional terms [Sy] or in motor;orienting terms (cf. Ref. [I I]). In other words, the normal subject’s tendency to bisect left of centre could either reflect a relatively magnified percept of the leftward part of the line (perhaps due to an attentional bias toward the left), or it could reflect a response bins in the form ofa predominantly leftward orienting tendency (perhaps due to an activated right hemisphere). In normal subjects. as in patients, the bisection bias can be inlluenced not only by cueing asymmetrically. but also by \:arying the spatial location of the line [S. 17, 221. As with the bisection disorder in neglect. however. again neither manipulation provides a decisive test of the motorloricnting theory. In contrast. the landmark task. which requires no explicit act of bisection [7. 16. 201. should provide direct evidence on the role of such factors. 111 this task, the subject is given a pre-transected line_ and is required to indicate, either manually or verbally. the end that is nearer to the transection mark. If the lint is in reality centrally bisected, but the left half appears longer to the subject (the perceptual hypothesis). then a ricqhrward response should be made. If instead there is no perceptual asymmetry but only a pre-motor bias towards the left. then the subject should. if responding verbally, indicate tither end randomly. Indeed, if responding manually, any postulated orienting bias should incline the subject to indicate the I+ end. Thus. the pre-motor bias and the perceptual bias accounts of performance in the standard bisection task predict different and even opposite response preferences in the manual landmark task. Analogous logic can be applied to an
l.INE BISCCTION JUDGEMENTS
IN UOKMAL
SUBJtCTS
517
analysis of the effects of cueing or of spatial location: if the independent variable causes a bisection error in a given direction, then the manual landmark task will always place the two types of error factor in mutual opposition.
EXPERIMENT
1
The first experiment was performed in order to ensure that our particular stimuli, and our cueing and spatial manipulations, would replicate the effects previously described by others. We therefore examined performance in the traditional line bisection task as a function of the spatial position of the stimulus (to the left, right or centre of the subject’s body midline), and as a function of attentional cueing to either or both ends of the line.
5frhjrct.s. Twelve subjects (seven female. fi\c male) ranging in age from 23 to 36 years (mean = 27). participated as unpaid volunteers m the experiment. All were strongly right-handed as assessed both by self-report and by admmistration of the 12.item A~F~~TT Handedness Inventory [I]. Muterids. Twenty-four black Letrtwt lines of 20 cm length and 1.5 mm width were placed horizontally and centrally on twosheetsofA4paper. I2linesoneachsheet. Ofthese24lines,subsetsofsix hadeitheraletterat theleft end. a letter at the right end. a letter at both ends. or no letter at either end. The letters were separated from the end of a line by a I mm space. Cued and uncued lines were ordered pseudorandomly on the sheets. and different orderings were used in each spatial condition for a given subject. The letters used as cues were chosen as visually similar to each other (0. Q. C or G ). so that they would require fixation for identification. Bilaterally cued lines always had different letters at the two ends, and in all conditions the letters occurred equally often at left and right. f’,occ~/~u. Subjects were seated at a table opposite the experimenter. who ensured that the subject’s body positjon remained constant throughout the experiment (though without drawing attention to the body midline as such). Subjects were instructed to name the letter(s), if any. and then to centrally bisect the line as accurately as possible using the right hand, proceeding line by line through each sheet. After each transection that line was covered with a card in order to pre\ent comparison ofthe present response uith previous bisections. Head and eye movements were not restricted in any way. The set of 24 lines was presented once in each spatial location (left, right or central to the subject’s body midline). The order of locations was counterbalanced within and between subjects according to a Latin square. In the central condition the viewing distance was approximately 45 cm from the line that was to he bisected. In the left and right hemispace presentations the sheets were located such that the centre ofeach line lay at a distance of 30 cm from the sagittal body midline. Errors in line bisection for each subject and for each condition were measured to the nearest millimetre and aberaged across the six instances. Errors to the right of the midpoint were given a positive value and those to the left a negative value. Sratisticcrl LI~IUI~WY. The mean error scores and their standard deviations were analysed. They were subjected to three-wsay analyses of variance with sex as a between-subjects factor, and apace (left, centre. right) and cueing (no ictter, two letters, letter left, letter right I aswithin-subjects factors. The significance of main effects and interactions involving repeated measures were assessed using a Geisscr Greenhouse adjustment to the degrees of freedom where appropriate. Finally. significant main effects and Interactions were exammcd in detail through the Newman Keuls testing procedure, using the 5 “0 level ofsignificance throughout. In addition, one-sample r-tests were used to test for constant error (departures from a mean of zero) in particular test conditions.
Rrsults Analysis of the mean error scores showed significant main effects for both spatial location IF(1.2, 12.0)=12.17. P
P
,
,
432-
No letter
1 o-1 -
_
Two
M
Letter left
letters
7
Letter right
-2 -3 -4 -
Analysis of the variability (SDS) of the bisection responses (F-C I for both main ell’ects and for their interaction).
showed
110
significant
etfccts
The results show clearly that hcmispace as well as acing atfectcd performance in the bisection task. Leftward bisection errors in left space and (non-sigiiiticant) rightward errors in right space were found. This may bc seen as a relative undcrcstimation of the leftwnrd extent of lines presented in right hemispace and of the rightward extent of stimuli in left hemispace. The result agrees with most previous work in other laboratories [S. 221 and in our own [7]. but is directly opposite to the result reported by NIC‘HI-I.I.I ~‘r trl. [I 71. The reasons for this disagreement are unclear. although it is notable that NK’HKI.LI (7~ trl. positioned their laterally-placed lines at only a 70 mm eccentricity, so that part of any line longer than 140 mm would hale appeared on the “wrong” side of the sagittai midline. Their data show that it was in fact only with the longest lines used (200 or 240 mm) that thcit paradoxical effect appeared. NI(‘HEI.I.I 6’1 tr/.‘s subjects were also older than ours (mean 64 years). although it is unclear why that \vould CLILISC the observed reversal: ncvcrtheless it is
LINE
BISE(‘TION
ILDGEMENTS
IN
YOKMAL
SUB,E(‘TS
519
interesting that other investigators using older subjects have failed to find the usual effect of spatial location [S]. In any event, it is clearly not possible on the present data to endorse NICHELLI et a/.‘~ proposal that a tendency to attend centrally causes a perceptual overestimation of the inward parts of lines placed laterally. Independently of spatial location, unilateral left or right cues caused bisection to err towards the side of cueing. This replicates previous results 17, 171, and is clearly consistent with the idea that attention directed toward one end of a line causes a relative perceptual overestimation of that part of the line. Bisection of centrally presented lines erred significantly leftwards, in agreement with the findings of BRADSHAW et ul. [4]. Others have found no clear population bias among their subjects (e.g. Ref. [13]). As indicated in the Introduction, it could either be argued that this bias, when it is seen, is due to an orienting/motor asymmetry or to a perceptual/attentional asymmetry: either of these might plausibly follow from a right hemisphere dominance for spatial processing. In other words, a response-based tendency due to a greater right hemisphere activation by the visuospatial processing demands of the task. rather than a perceptual asymmetry, could cause the leftward bias. In the same vein, it could be further argued that both the space and cueing effects might be attributable to a tendency for the subjects to respond in the direction contralateral to the more activated hemisphere [ 12, 201. Thus the use of lateral presentation conditions might, through activating the contralateral hemisphere, cause an increased lateral orienting response that results in “overshoot” bisection responses. Similarly. the effect of cueing, at least with a centrally-located line. could be to activate the contralateral hemisphere. with the result of causing an increased motor-orienting tendency in the direction of the cue. In other words. all of the three influences on bisection found in Experiment 1 could be the result of the operation of either a motor.‘orienting bias or a perceptual;attentional bias. The problem with the standard bisection task is that it requires a motor response and therefore inevitably confounds perceptual and pre-motor factors. Accordingly in the following experiment bisection judgements were studied using a “landmark” task, to enable an examination of perceptual effects in isolation.
EXPERIMENT
2
In this experiment. the subjects were not asked to bisect a line. but instead were presented with lines that were already transected. The subjects had to make a forced-choice judgement as to whether each line was transected nearer to its right end or its left end. This version of the landmark test transformed the subjects’ task into a purely perceptual one, in that no directional motor response was required.
S~rRjwr\. A neu sample of 12 subject5 (six female. six male) participated as unpaid volunteers in the expenmcnt. All wcrc strongly right-handed as asscsscd both by self-report and by administration of a six-item quehtionnalre. They were aged hetueen 22 and 35 years. .Iltrt~r~rr/s. Sixty-four black lines of 70 cm length and I .5 mm wdth were drawn horizontally and centrally on sheets ofA paper (used sldcways). one line per sheet. Each lint was transected with a lint 10 mm in length and I mm in width. set at right-angles IO the long line. Of the 64 lines, 32 mere asymmetrically pre-transected, with 16 being hiaected 5 mm to the right of true centre and the remaining lines 5 mm IO the left of centre. These 32 asymmetrical stimuli uere otherwise unmarked. and served as lillcrs intended to encourage the impression that all 64 lines might he asymmctricallq bisected. The remaining 32 lines uere transected at the midpoint. Of these centrally transected lines 25’!:, had a letter at the left end, 25 “,
C‘ucd and uncucd lint\ were prcscntcd p~eudorandoml>. in dllfcrcnt acquenccs for cnch space condltlon. Apain the letters used were visually similar to each other (0. Q. c. G). The vieuing dlstancc was the umc ‘15in Experiment I. and again left and ripht hemispace prcwltations Ia) at a distance of 30 cm from the subject’.\ sagittnl midline. Pror f~/rrrv. Again subjects were seated at a table opposite the experimenter. uho ensured that their body pwtion remained constant throughout the experiment. They lcerc falsclq mformcd that none ofthc tranwztions wcrc placed at the exact ccntrc ofa line. They ucrc asked to name the letter(s) if there \bereany. and then to tell the evpcrimcnter Mhich end ofthc lint they thought the transection “.‘a\closer to. Subjects were forced to make ;L left right choice c\un lflt was ncccsaary to guess: no other response wah permitted. The block of64 trials was prcbented three times. once in each spatial location (left. right orccntral with respect to the subject’!, body midlinc)and the order ofthe three block\ W;I\ counterbalanced hetwccn subject\ In ;I Latin-square fashion. Again head and eye movement< wcrc in no \\a! rc\trlctcd. .Srcrti.tfkcr/ I~W/J,WY Judpcments ofthc atimull prc-tranxctcd 5 mm from the midpoint were perfect and theref(,re not analyscd. I-or the centrally pre-trnnaccted lines. the ,~~r,r~hc,,~,f~iqh/~c.rrrt/(.I~~~(.~~ (out ofeight maximum) for each suhJcct in each condition wa!, subjcctcd to a three-waq analysis of variancewith ICY ii\ a hctwccn-suhJects fxtol- and \pati;d location and cueing a’r within-suhjcct~ factors. (Although the di5trihution of choIcea mas apprortlmatcl! rectangular. ANOVA is sulliclentl> robust to cope with such data.) Other statistical procedurc~ were a’r hcforc
There was a significant main effect of space [F( 1.9. 19.3)=4.32. P-cO.0251. Rightward judgements (indicating that the right half of a line appeared shorter than the left) were significantly more frequent in left or central space than in right space (see Fig. 2). Only in left and central space did these choices significantly differ from 50% chance pcrformancc (r =Z.ZO. f-'
0
I
I
1
I
L
C
R
Left Space
Centre Space
b
No letter
-
Two letters
b
Letter left
U
Letter right
Right Space
b-is. 3. I~\pcrimcnt 2 Verbal) landmarh Judfement\ ;I\ it function
There was also a highly significant main efTect of cueing [F (2.2,77.0)=21.54, P
LINE BISECTJON
JL’DGEMEh’TS
IN NORMAL
SUBJECTS
521
the line as being closer to the transection mark, whereas a unilateral right cue had the opposite effect: subjects saw the left end of the line as being closer (see Fig. 2). Unilateral cueing on either side caused reliably asymmetrical judgements (t = 5.35, P~O.001 for left cues and t = 3.19, P < 0.00 1 for right cues). Responses under the no-cue and bilateral-cue conditions showed slightly more right than left responses (significantly different from chance performance in the case of the no-cue condition, with t =2.79. P~0.05). There was also a significant space x cue interaction [F (3.8, 38.3)= 3.17, P~O.051. This seems to have been due to an anomalous pattern ofjudgements of bilaterally cued lines: they were judged rather like left-cued lines in left hemispace but more like right-cued lines in central and right space. There were again no significant effects involving sex.
.4s in Experiment 1, a strong effect of cueing was found. In this experiment, the half of a symmetrically transected line opposite a letter cue was regularly chosen as appearing shorter, indicating a relative perceptual overestimation of length on the cued side. This result supports the view that the effect of cueing in Experiment 1 was also due to a similar relative overscaling of the cued side. Also as in Experiment I, spatial location affected the results. Presentation of the line stimuli in left and central space led to a significant preponderance of “right” choices (i.e. subjects indicated that the right half of a line appeared shorter than the left). Right spatial presentation led to a non-significant tendency in favour of“left”choices. These data, in a task that involved no directional motor response, indicate that the leftward biases observed in active bisection in left and central space (Experiment 1) must be attributable in part to the leftward extent of stimuli appearing relatively enlarged. In this experiment, again as in Experiment 1, a complementary effect of rightward overestimation occurred, though more weakly, in right space. Taken together, then, Experiments 1 and 2 indicate that the effects of spatial location and cueing on line bisection cannot be wholly attributed to asymmetrical motor-orienting tendencies. Even when no directional response is made, biases remain. These effects of cueing and hemispace are superimposed upon an overall constant error in both experiments. This introduces a tendency both for leftward bisection (Experiment 1) and for rightward choices in the landmark test (Experiment 2), both of which reach significance in central space presentation. Demonstrating the existence of any of these perceptual biases, including those brought about by spatial location or cueing, however, does not imply that motor orienting biases do not crlso contribute to the effects observed in active bisection (Experiment 1). Therefore a third experiment was carried out in which the two factors-perceptual and orienting-were specifically set in mutual opposition.
EXPERIMENT
3
In this experiment we again used the landmark task, but this time the instruction was to make a lateralized motor response (pointing) to the side that the transection mark was thought to be nearer. It was reasoned that when faced with a difficult (impossible‘?) psychophysical judgement, a subject’s manual response may be swayed by any orienting biases that might be operating. Thus if such biases are strong enough to cause error in the
standard bisection task, then they might cause the observer to veer toward pointing in that same direction in this manual landmark task. Three influences were apparent in the bisection behaviour described in Experiment I : (a) an overall leftward bias, (b) a bias towards the cued side, and (c)a sideward bias with laterally placed lines. As argued earlier, bias (a), the constant error whereby subjects tend to bisect a line to the left of its true centre, might be due to a leftward pre-motor bias. in addition to a perceptual overestimation of the left part of the line. If the predominant bias were leftwardmotor, then in the pointing version of the landmark task. subjects would. if anything. be expected to point generally lsfi rather than right, when faced with a centrally transected line. The contrary effect would occur with a predominantly perceptual bias: if subjects perceived the left half of a line as being longer, riylltward pointing would be predicted. If the two effects are both present, then they would presumably subtract from one another. For each of the effects under examination, the use of the pointing version of the landmark task sets perceptual and pre-motor factors in opposition: if a pre-motor bias is operating. it should work against the perceptual bias observed in Experiment 2. It should be emphasized that the task is objectively impcssible, and is perceived as difficult by subjects. Thus it seems unlikely that a firm perceptual decision is made on each trial that determines the subject’s response in a manner immune to putative output biases.
Suhjwr\. Twclvc further subjects (heben fcmcdc, ii\e male) ranging in age from 22 to 37 !cars (men~~=X). participated as unpaid volunteers in the experiment. All wcrc strongly right-handed ah asserscd both hy self-repor? and by administration of the 12-item AZ~I.TT Handcdncsa Inventory Ll]. :Ilrrrc~iu/\. Eighty-four black Urtr,wt lines of 20 cm length and I .5 mm width were used. each line placed horizontally and centrally on ;L separate sheet of A4 paper. Sixty ofthesc lines uere asymmetrically prctramected. SIX lines each being marked at I. 7. 3. 4 or 5 mm to the right or the left of the true centre. Thcsc extra asymmctrlcal \ttmuli wcrc added to incrcabe task difficulty (subjects easily dlstinguishcd the 5 mm deviations of Experiment 2 from central transcctlon\). and also with the intention to estimate the psychophqslcal “neutral” point at which a non-central transection aould appear suhjecti\cly central The remaining 24 lines were transcctcd in the centre and cued as in Experiment 2. with ais Inner in each of the lout c‘ucconditions. The set of X4 1ine.r wab then divided into tv,o subsets of 42 lines. each subset containing an equal number ol’as~mmctrv_zall~ and centrally transected. cued and unwed lines. Each suhact ~;IS presented once in each spatial locatwn (left. right or central with respect to the subject’s body mldline). the order of spatial presentation\ thus belngcounterhalllnccd within as well BS between subjects. The viewing distance and spatial locatlom ofthc lint\ was the \ame 3s In Experiments I and 2. P~~~c~dwc~. Subjects mcrc instructed as in Experiment 2. except that after naming any letters they \\crcI-equlrcd 1t1 put~r directly to the end of the line that appeared closer to the trnnscction. As in Experiment 2. “middle” re\pon.a \blercnot permltted. and head and cy mo\ementa wcrc unwstrictcd. Srt/ri,>ric,tr/ ctr~~/~~.w.Subjects’ pcrformancc on the stimuli prctransectcd 3 5 mm from the true mldpolnt I\;LII pcrfcct for I I ~ubyxts (one subject judged all leftward transections as trnn&ectcd tu the right when line.\ wcrc praentcd in right \pacc). SubJectivcl> \ubyxt\ found it as hard to judge transections that doiatcd I or 2 mm from either side ofthe midpoint as lines that were actuall> centralI! hisccted. However the distrihutlon ofthe el-rors on the I and 2 mm daintinp line\ uas equally distributed het\\een left and right rcsponheh. f.or tho hne\ prctransocted 111the ccntrc. the t~~rwh~rra/~I~//~~wwY/ rc,~prww (m;tuimum 51x per cell) \\crehubjsctcd to ;I three-Ma) annlqsi\ of \arlancc ;I\ 111Fupcrimcnt 2.
Unlike Experiments 1 and 2. no main effect of spatial location was found [F (1.2. 12.3 ) < I]. Cueing, however, remained highly significant [F (2.4, 24. I ) = 43.08. P< 0.001 )]. The twocue and no-cue means did not differ significantly. but all other pairs of cueing conditions did differ from each other at the 0.05 level (see Fig. 3). The direction of the unilateral cueing 2. and their magnitude was if anything greater effects was the same as in Experiment cue conditions differed highly reliably than in Experiment 2. Both unilateral
from chance performance (t=5.01, PO.lO], nor did any effects involving sex.
2 B E .cs,
II________ _-----
3--
___--
----_____
50 %
No letter
0
I L Left Space
Fig. 3. Experiment 3. Manual mean number of forced-choice
I C Centre Space
-
Two letters
b
Letter left
+
Letter right
I R Right Space
landmark judgments as a function of spatial location and cueing: the manual indications (out of six) that an objectively central transection IS placed right of centre.
Discussion The hemispatial asymmetry present in Experiments I and 2 could not be found in this experiment. If this result is due to the change of response modality (directional pointing rather than a verbal response), it would suggest that a pre-motor bias is instrumental in the genesis of the spatial location effect. That is, the “centrifugal” bisection behaviour in which subjects bisect laterally displaced lines too far laterally (Experiment 1) may be caused partly by a perceptual distortion (Experiment 2) but equally also by an output (overshoot) bias, independent of the actual perception (Experiment 3). In contrast, the effect of cueing was strongly replicated in Experiment 3; the subjects tended to point to the side opposite the letter cue as appearing shorter. Cueing thus had a rather robust effect, consistently strong over all three experiments. It appears that drawing attention to one side by means of unilateral cueing produces a relative overevaluation of that side, and that the nature of the required response is irrelevant to the size of this cueing effect. Finally, the constant error that was present in bisection (Experiment 1) and which appeared as a perceptual effect in Experiment 2. was not seen in Experiment 3. This could be due to a failure to replicate a weak phenomenon; alternatively, however, the negative result
514
could be interpreted therefore contribute
A. D. Mtt WK. M. BKXHMAW
and L. PAGLIARINI
as due to the leftward operation of a pre-motor bias which might to the constant error often reported in active bisection [4. 5. 71.
GENERAL
DISCUSSION
Experiment 2 showed that the leftward bias in line-bisection tasks in normal subjects (e.g. in Experiment I ) is partly the result of a perceptual bias. Subjects really do see a line bisected centrally as being bisected slightly to the right. and so presumably will see a line bisected slightly to the left as being centrally bisected. (Direct evidence for this latter difference has been found in unpublished work [7].) In other words, we subjectively overscale the left half of a mid-transected line, relative to its right half. This finding clearly lends support to perceptual or attentional theories of the bisection bias. However it does not ipso /ircto provide evidence against an orienting bias that might affect bisection responses (cf. Ref. [I 11). That theory. considered alone, would have predicted no bias in Experiment 2 and (if anything) a leftward bias in the pointing task used in Experiment 3. However both factors (perceptual and premotor) operating together would produce the pattern of data we have observed across the two experiments. Thus both of the factors may be partial determinants of the constant error seen in active bisection. That is, our data would be consistent with a dual role of the right hemisphere in both enhancing the perceptual salience of spatial stimuli in the left hemispace, and also in activating leftward orienting response tendencies. Experiment 1 confirmed most previous reports [S, 7, 221 that the leftward error in bisection tends if anything to increase in the left half of egocentric space, but to reverse direction in the right half, Experiment 2 extended this finding by demonstrating a parallel spatial effect on landmark judgements, in which subjects perceive midpoint-transected lines placed on either side as transected nearer to the spatial midline. (N.R. Superficially similar results have also been reported for judgements of tachistoscopically-presented stimuli [7, 201; however those effects were demonstrated to be hemiretinally based 1201, while the present data are obtained in free-gaze conditions, and are therefore probably hemispatially based.) However, Experiment 3 indicates that this perceptual bias can be nullified by a requirement to make a motor response to the end of the line subjectively nearer the l~~tldmark. Our provisional interpretation is that in this second landmark experiment, two opposing effects may have cancelled each other out. That is, a perceptu~~lly-determined preference toward the midline for laterally placed lines (Experiment 2) may have been counteracted by a motor-orienting bias tending to cause a lateral overshoot (Experiment 3). If so. then that same orienting factor probably also contributes to the hemispatial “overshoot” effect in active bisection evident in Experiment I. Nevertheless it must be admitted that it could be argued that the hemispatial effect wasjust not strong enough to be replicated in Experiment 3. Certainly it was statistically weaker than the cueing effect in both Experiments 1 and 2 (cf. Ref. [S]). In contrast, cueing procedures* provided clear and robust results throughout. Use of a single letter consistently biased landmark judgements toward the other end of the line, i.e. caused a relative overscaling of the cued end of the line. This effect was not reduced by the use of a directional motor response (Experiment 3 vs Experiment 2). This supports the idea that the influence of cueing upon * It should be noted that the letter cueb are tra.wrw~l to operate by drawping attention 10 a part ofthe atimttlus array. It IS ofcourse possible that they also alter the percep~uai point olbalance in a figure; the importance this could hc tested by the use of non-visual cueing.
of
active line bisection (found in Experiment I and by others) operates at the perceptual level and little if at all at the level of any putative orienting response bias. (The latter, without additional assumptions, could in any case only have explained the cueing effects seen in central space.) It seems likely that this perceptual bias can be attributed to a differential attentional salience of the two ends of a unilaterally cued line. It is commonly assumed that a right-hemisphere mechanism is at work in causing the constant error in normal bisection behaviour. If so, the above considerations make it plausible to argue that this may itselfcome about partly through an attentional bias. In turn. it may be that patients with spatial neglect suffer such a gross distortion of their attentional field that a pre-existing slight bias for size overestimation in left hemispace is transformed into a much larger bias for underestimation in that hemispace. There is neuropsychological evidence for such a perceptual mis-scaling: GAIKOTTI and TIAKI [9] reported that patients with leftward neglect judged the rightward of two similar patterns presented together as appearing comparatively larger. We have replicated and further examined this phenomenon [163. There is thus mounting evidence that the perception of spatial extent is subject to attentional influences, and that unilateral posterior lesions, in causing an abnormal attentional gradient, may ipsofacfo cause abnormal errors of size perception. It is primafacie surprising that one’s perception of spatial extent might be altered by attentional biases: we tend to think of size perception as a rather low-level process, and even as pre-attentional. But in fact size perception is computationally rather complex, and is probably the end-product of several stages and types of input analysis. It is therefore not implausible that a system for spatial perception might have evolved which is routinely and intrinsically distorted by attentional factors. It may indeed be important for survival that biologically unimportant items be subject to an illusory underestimation of their size relative to important items. At the same time, however, there would have been selective pressure for animals to have a veridical metric to permit accurate size computations for the visual control ofprehension. Consequently it seems unlikely that the perceived size ofattended items would be subject to illusory enlnryernent. but rather that unattended ones might be subject to subjective diminution. A~lino~c/edC/er,le,lts~The
authors
are grateful
for financial
support
from the Wellcome
Trust
and the Medical
Research Council.
REFERENCES 1. ANNEX. M. The binomial distribution 2. 3. 4. 5. 6. 7. 8.
of right. mixed and left handedness.
Q. J. taup. Ps~~hol. 19, 327 333. 1967. BISIA~H. E.. GEMINIANI, G., BLXTI. A. and Rr;scok~, M. L. Perceptual and premotorfactors of unilateral neglect. ,Veuro/oy~ 40, 1278 128I, 1990. BOWERS. D. and HEILMAN. K. M. Pseudoneglect: Effects of hemispace on a tactile line bisection ~asl\. ,‘Veurops~c,/lologiil 18, 49 1 49x. 1980. BRADSHAW.J. L., NETTLETON. N. C.. NATHAN, G. and WILSOV, L. Bisecting rods and lines: effects of horizontal and vertical posture on left-side underestimation by normal subjects. .Yrtcr~~~.s~c~ho/~~~ia 23, 421 425. 19~s. BKADSHAW. J. L.. NATHAY, G.. NETTLETON.N. C.. WILSOY, L. and PIEKSON.J. Why is there a left side underestimation in rod bisection‘? Neurop.s~~llo/o(/in 25, 735-738, 1987. COSLETT, H. B.. BOWEKS. D.. FITZPATKICK. E., HAWS. B. and HIXLMAN. K. M. Directional hypokinesia and hemispatial inattention in neglect. Bmin 113, 475486, 1990. DUDGEOIZ, K. Hemispheric attentional control: the effect of hemispace and cueing on normal dextral and sinistral line bisection performance. Unpublished BSc. dissertation. University of St. Andrew, 1988. FUKATSI!, R., FLJII, T., KIMUKA, I., SASO, S. and K~GLKC, K. Effects ofhand and spatial conditions on visual line bisection. T&&r J. erp. .!4url. 161, 329-333. 1990.
526
A D. MII Kf K. M. BKFCHCIAUXnnd L. PA(;LI.\KI~I
G~I\OI II. G and TIA(X.I, C. The relationships bet&een disorders of visual perception and unil;lteral sp;itial neglect. ,~europ.\!“,ho/ocliu 9, 45 I 4%. 197 I. IO HALLICIAU, P. W. and MAKSHALI . J C. Perceptual cueing ;md perceptuo-motor compatibility in visuo-spati:tl neglect: a stngle case study. (i~qn. ,Yrur~p.sycho/. 6, 423 435. 1989. HI ILMAI, K. M. and Vnr.r.wrr.rx. E. Mechaniama underlying hemispatial neglect. -Inn. .l’ewr,(. 5, 166 170. II lY7Y. I7 K~usnor R~I.. M. Mechanisms of unilateral neglect. In .l’e~rrr,plt!,.\ir,/~~~/;(,~~/ trntl .l’clr,r,p\!,~/lo/o!/i[ trl .d.\pw~,\o/ Spctric~l Nrcqlr~l, M. JI ~XVEKOI) (EdItor). pp. 69 86. Elsevicr. Amsterdam. 19X7. L.. HALL.IC;A~X. P. W. and MAKSHALL. J. C. Individual variation in line bisection: A study of normal 13 MANNfUG. 9
subjects with application to the interprctatlon of visual neglect. .\‘~,~~r~)p,\!,c~l~~,lo(liu 28, 647 655, 1990. 14. MAKSHALL. J. C. and HALLIG,ZK. P. W. When right goes left: An investigation oflinc hisection in H case ofvisual neglect. Cwtru 25, 503 515. 19XY. 15. MIL~K. A. D. Animal models for the syndrome of spatial neglect. In !~r,u~opll!,\i~,/~J~i~~~/ rtrd ,Yru~opa!.~,hr,Io~qkul .-l.spw/~ r!/ Sputitrl ,‘L’cyhr. M. JE~\%~NLKOI> (Editor). pp. 259 28X. Elserier, Amsterdam. 1987. 16. MILVI.K, A. D.. BKKH~IA~~. M.. ROIKK~S. R. C. and FOKSTEK. S. V. Leftward underestimation of linear hi/e in pxticnts with viauospatial neglect. Eur. J. ,Yrurr~wi., suppl.5, in press, 1992 17. NICH~.I.LI. P.. RINALIII. M. and CI HLLLI. R. Selective spatial attention and length representation in normal subjects and in patients wth unilateral spatial neglect. Bwin C‘oqnrr. 9, 57 70. 19X9. IX. Por-rl.. W. Diswciation of spatial discrimination deficit\ following frontal and parietal lesion\ in monkqa. .I c’on,p. Phr.\i~/. /?s~~ho/. 82, 277 239. 1973. 19. RFI.T~K-LOKENZ, F. A. and Posr;f K. M. I. Components ofneglect from right-hemisphere damapc: an analysis 01 line bisection. ,Ycu~op.s~~kolf,clltr 28, 327 333. IYYO. 20. RI I T~:K-LOKI:NZ. P. k.: KI&~KUI-. M. and Mosc.o\ II~C’H.M. Hemispheric control ofspatial attention. Bw~rr C‘oqnit. 12. 24Ok266. 1990. /~r,/~xq~c~ 21, 21. Rllwc)cx. M. J. and HLMfwKbYs. Ci. W. The effect of cueing on unilatcrnl neglect. ,“\;rw~>p,\~ 5x9 599. 1983. 72. S(.HI.NKI.NHI:KG.T.. BKAI)I.OKI). D. c‘ and AJAX. E T. Line bisection and unilateral vfsual neplcct ln paticnta with neurological Impairments. ,Vwr~~lo(q~ 30, SOY 517. 1980.