visuospatial neglect in normal subjects: altered spatial representations induced by a perceptual...

7
Neuropsycholo`ia\ Vol[ 25\ No[ 4\ pp[ 358Ð364\ 0887 Þ 0887 Elsevier Science Ltd[ All rights reserved \ Pergamon Printed in Great Britain 9917Ð2821:87 ,08[99¦9[99 PII] S9917Ð2821"86#99020Ð9 Note Visuospatial neglect in normal subjects] altered spatial representations induced by a perceptual illusion JAMES FLEMING and MARLENE BEHRMANN$ Department of Neuroscience\ University of Pittsburgh\ Pittsburgh\ Pennsylvania\ 04159\ U[S[A[^ $ Department of Psychology\ Carnegie Mellon University\ Pittsburgh\ Pennsylvania\ U[S[A[ "Received 17 May 0886^ accepted 14 Au`ust 0886# Abstract*Previous studies have suggested that the representation of space is distorted in patients with hemispatial neglect[ To examine the nature of the altered spatial representation\ we developed an analog of neglect in normal subjects using bisections of the Judd visual illusion "e[g[\ ³ **Ð ³# whose direction "facing left:right:both# and _n angle "03>:34>:65># were orthogonally varied[ Subjects made bisection errors that deviated signi_cantly in a direction opposite to that in which the _ns pointed\ and these deviations decreased as _n angle increased[ We replicated these _ndings in a second task in which subjects were given the midpoint and placed the two _ns at the ends of an imaginary shaft "e[g[\ ³ = ³\ note] dot true center#[ These _ndings suggest that the geometric properties of the display have a strong in~uence on the representation of space\ and thus may be used to induce a distorted perception of space analogous to that observed in patients with visuospatial neglect[ Þ 0887 Elsevier Science Ltd[ All rights reserved[ Key Words] neglect\ perceptual illusion\ spatial representation[ Introduction The search for the relation between space of the {physi! cal world|\ i[e[ the world which exists apart from that mentally instantiated by organisms\ and {psychological space|\ i[e[ space represented by neurons or ensembles of neurons\ has persisted for many years ð0L[ Though the link between neural processing and spatial representation remains poorly understood\ important insights into the mechanisms underlying spatial representation have been obtained from patients with a visuospatial de_cit acquired following brain damage[ Patients with left hemi! spatial neglect "typically after lesions to the right parietal region# may fail to respond to or report information on the contralesional side of space "see Bisiach and Vallar ð1L for an overview#[ In an attempt to understand the mechanisms responsible for spatial coding\ recent studies have been concerned with establishing the nature of the spatial map derived by these patients and the cor! Address for correspondence^ tel[] "301#157!3026^ fax] "301#513!8087^ e!mail] j~emingÝandrew[cmu[edu 358 respondence between these perturbed internal spatial rep! resentations and veridical physical space[ One potentially useful method for studying the spatial anisometry in neglect is to induce {neglect| in normal subjects and to use this analog to analyze the resulting spatial topography[ Several recent studies have shed light on the ~exibility of spatial and attentional processing "see\ for example\ ð2Ð4L# by unleashing or exaggerating the attentional biases of normal subjects or by manipulating their spatial representations[ In line with these studies\ we induce neglect!like performance in normal subjects by using visual illusions to alter their perception of spatial extent\ and then we examine their subsequent behavior[ A study by Mattingley and colleagues ð5L\ though not explicitly directed at inducing neglect!like performance\ had subjects bisect the Judd visual illusion\ a variant of the Mu llerÐLyer illusion[ They demonstrated that the perception of spatial extents can be manipulated so as to induce neglect!like bisection errors "ð5L] see also ð6L^ see Fig[ 0#^ speci_cally\ on both the unilateral!_n and bilat! eral!_n Judd illusions\ normal subjects and neglect pat! ients made bisections that were displaced away from the

Upload: james-fleming

Post on 15-Sep-2016

216 views

Category:

Documents


0 download

TRANSCRIPT

Neuropsycholo`ia\ Vol[ 25\ No[ 4\ pp[ 358Ð364\ 0887Þ 0887 Elsevier Science Ltd[ All rights reserved\ Pergamon Printed in Great Britain

9917Ð2821:87 ,08[99¦9[99PII] S9917Ð2821"86#99020Ð9

Note

Visuospatial neglect in normal subjects] alteredspatial representations induced by a perceptual

illusion

JAMES FLEMING� and MARLENE BEHRMANN$

� Department of Neuroscience\ University of Pittsburgh\ Pittsburgh\ Pennsylvania\ 04159\ U[S[A[^ $ Department of Psychology\Carnegie Mellon University\ Pittsburgh\ Pennsylvania\ U[S[A[

"Received 17 May 0886^ accepted 14 Au`ust 0886#

Abstract*Previous studies have suggested that the representation of space is distorted in patients with hemispatial neglect[ Toexamine the nature of the altered spatial representation\ we developed an analog of neglect in normal subjects using bisections ofthe Judd visual illusion "e[g[\ ³**г# whose direction "facing left:right:both# and _n angle "03>:34>:65># were orthogonally varied[Subjects made bisection errors that deviated signi_cantly in a direction opposite to that in which the _ns pointed\ and these deviationsdecreased as _n angle increased[ We replicated these _ndings in a second task in which subjects were given the midpoint and placedthe two _ns at the ends of an imaginary shaft "e[g[\ ³ = ³\ note] dot � true center#[ These _ndings suggest that the geometricproperties of the display have a strong in~uence on the representation of space\ and thus may be used to induce a distorted perceptionof space analogous to that observed in patients with visuospatial neglect[ Þ 0887 Elsevier Science Ltd[ All rights reserved[

Key Words] neglect\ perceptual illusion\ spatial representation[

Introduction

The search for the relation between space of the {physi!cal world|\ i[e[ the world which exists apart from thatmentally instantiated by organisms\ and {psychologicalspace|\ i[e[ space represented by neurons or ensembles ofneurons\ has persisted for many years ð0Ł[ Though thelink between neural processing and spatial representationremains poorly understood\ important insights into themechanisms underlying spatial representation have beenobtained from patients with a visuospatial de_citacquired following brain damage[ Patients with left hemi!spatial neglect "typically after lesions to the right parietalregion# may fail to respond to or report information onthe contralesional side of space "see Bisiach and Vallarð1Ł for an overview#[ In an attempt to understand themechanisms responsible for spatial coding\ recent studieshave been concerned with establishing the nature of thespatial map derived by these patients and the cor!

� Address for correspondence^ tel[] "301#157!3026^ fax]"301#513!8087^ e!mail] j~emingÝandrew[cmu[edu

358

respondence between these perturbed internal spatial rep!resentations and veridical physical space[

One potentially useful method for studying the spatialanisometry in neglect is to induce {neglect| in normalsubjects and to use this analog to analyze the resultingspatial topography[ Several recent studies have shed lighton the ~exibility of spatial and attentional processing"see\ for example\ ð2Ð4Ł# by unleashing or exaggerating theattentional biases of normal subjects or by manipulatingtheir spatial representations[ In line with these studies\we induce neglect!like performance in normal subjects byusing visual illusions to alter their perception of spatialextent\ and then we examine their subsequent behavior[

A study by Mattingley and colleagues ð5Ł\ though notexplicitly directed at inducing neglect!like performance\had subjects bisect the Judd visual illusion\ a variant ofthe Mu�llerÐLyer illusion[ They demonstrated that theperception of spatial extents can be manipulated so as toinduce neglect!like bisection errors "ð5Ł] see also ð6Ł^ seeFig[ 0#^ speci_cally\ on both the unilateral!_n and bilat!eral!_n Judd illusions\ normal subjects and neglect pat!ients made bisections that were displaced away from the

J[ Fleming and M[ Behrmann:Neglect by perceptual illusion369

Fig[ 0[ The results from Mattingley et al[ ð08Ł using Judd _gures having _n length � 2[4 cm\ shaft length � 04 cm\ and _n angles of39>[ Notice the marked bisection errors generated by normal subjects on both Leftward and Rightward _gures[ The shafts of the

Judd _gures and the control line are aligned[

objective midline\ and shifted towards the outward!projecting _n and away from the inward!projecting _n[Relative to a no!_ns control line\ normal subjects mademean errors of approximately 7[7) and −4[1) on Right!ward "_ns going to the right# and Leftward Judd _gures\respectively[

Mattingley and colleagues proposed that the bisectionerrors arise from a perceptual compromise ð5\ 7\ 8Ł betweencon~icting cues derived from the horizontal extent of theshaft and from the horizontal extent of the entire _gure[They argued that\ for the inward!projecting _n on theleft side of a Rightward Judd _gure as in Fig[ 0\ theleft endpoint of the shaft is "mis#perceived as beginningslightly to the right of its true position[ Hence\ bisectionmarks are displaced to the right of the actual midpoint[Similarly\ an outward!projecting _n on the right side ofthe Rightward _gure causes the shaft to be "mis#perceivedso that it extends further to the right\ thereby leading torightward bisection displacements[ The same perceptualcompromises hold for Leftward _gures[

While a perceptual compromise interpretation mayprovide a general description of the illusory e}ects\ per!haps a more _ne!grained explanation is that the _nsinduce a distorted representation of the spatial extent ofthe shaft^ inward _ns distort the shaft\ pushing the mid!line to the opposite side whereas outward _ns pull themidline towards themselves[ Indeed\ several studies usingvariants of the Mu�llerÐLyer illusion have demonstratedsuch distortion e}ects "see Greene and Nelson ð09Ł for abrief review#[ Importantly\ a perceptual distortion ofspace resulting in a deviation from the midline in a linebisection task is considered to be one of the de_ningcharacteristics of hemispatial neglect ð00Ł[ Given thatsuch a distortion is also considered to give rise to bisectionerrors in normal subjects on visual illusions\ in this paper\we use the bisection of the Judd _gure as an analog ofneglect and explore the altered visuospatial rep!resentations induced in normal subjects[

Our study has three main parts]

0[ We initially replicate the basic _nding of Mattingleyet al[ ð5Ł to con_rm that the Judd _gure gives rise tothe reported robust displacements in line bisection andthat these displacements vary with the direction of the_gure[ Whereas shaft length is known to play somerole in determining bisection errors "albeit small inMattingley et al[ ð5Ł and large in Greene and Nelsonð09Ł#\ numerous studies have consistently dem!onstrated that the _ns play a substantial role in dis!torting perceived length and distance "see Greene andNelson ð09Ł for a brief overview#[ More speci_cally\the magnitude of the _n angle\ formed by the _n andshaft\ in~uences the distortion[ Although\ the exactrelation between _n angle and nature of the distortionfunction\ whether monotonic "see for example\ ð01\02Ł# or curvilinear "such as ð03\ 04Ł#\ is still contro!versial[ Given the consistent e}ect of _n angle\ weexamine the in~uence of the magnitude of the _nangles on bisection deviation by measuring subjects|bisection performance using displays with small "03>#\intermediate "34># and large "65># _n angles[

1[ Because we do not know what particular aspects ofthe illusion are critical for inducing the errors\ weanalyze systematically the in~uence of various geo!metric properties of the display on the bisection error[More speci_cally\ we evaluate the individual con!tribution of the three components of the _n anglegeometry\ the W "length of the _n#\ X "portion of theshaft that underlies the _n#\ and Y "vertical dimensionof the _n angle# components "see Fig[ 1#\ by holdingone of the single components "W\ X\ or Y# constantat a pre!speci_ed measurement while allowing theother two components to vary with _n angle[ If thebisection error varies when a given component is heldconstant\ then that _n component cannot be totallyresponsible for producing the errors[

J[ Fleming and M[ Behrmann:Neglect by perceptual illusion 360

Fig[ 1[ An example of a bilateral!_n Judd _gure[ The geometric components of a _n angle are shown[ "W# is the length of the _n\"X# is the length of the portion of the shaft that underlies the _n\ and "Y# is the height of the top of the _n to the shaft[

2[ We verify our results using an alternative techniquedeveloped by Bisiach and colleagues in order to dem!onstrate disordered spatial representations in patientswith visuospatial neglect ð05\ 06Ł[ In this procedure\subjects are given only the midpoint of a virtual shaftand are required to place the _ns at the correspondingend!points[ We predict that even in the absence of theshaft\ the end!point _ns would induce a deviationof the subjective midpoint biased by the angle anddirection of the _ns[

Our claim\ then\ is that irrespective of the technique"line bisection\ end!point _n placement#\ simply varyingthe angle and direction of the _ns will give rise to pre!dictable changes in the severity of the distorted perceptionof the shaft[ The rightward distortions with Rightward_gures may mimic left!sided neglect\ and leftward dis!tortions with Leftward _gures may mimic the less com!mon right!sided neglect ð07Ł[ Thus\ line bisection on Juddillusion displays may potentially provide insight into thenature of the spatial representational anisometry inneglect[

Method

Subjects

Three groups of ten\ right!handed subjects were recruited\one for each of the W!\ X!\ and Y!display versions of theexperiment[ The subjects were undergraduate and graduate stu!dents\ as well as friends of varying educational backgrounds[Subjects had no signi_cant medical\ neurological\ or psychiatrichistory and had normal or corrected!to!normal vision[ The "W#group consisted of 4 males and 4 females with mean age of 13[1yrs "S[D[ � 2[57#\ the "X# group consisted of 2 females and 6males with a mean age of 19[8 yrs "S[D[ � 1[40#\ and the "Y#group consisted of 4 females and 4 males with a mean age of11[8 yrs "S[D[ � 2[64#[

Materials and procedures

Geometric dimensions known to be robust in eliciting anillusion ð04Ł were used as a guide in selecting the geometric

components of the Judd _gures[ Because the distortion e}ectsare robust at a length of 04 cm "see Fig[ 1\ ð5Ł#\ all displays hadshafts that were 04 cm long and 9[914 cm wide[ Fins were also9[914 cm wide\ and _n angles were identical above and belowthe shaft[ The displays were presented in black ink on 7 0:1×03in[ white paper[ The shaft was positioned parallel to the hori!zontal edge of the paper and displaced vertically 0[4 cm and 1[3cm to the right of a center position on the page[ This was doneto prevent subjects from using the corners of the page to guidethe transection towards a shaft with its midpoint at the centerof the page[

Fin angle and direction were both manipulated within!subject whereas the geometric properties of the _n angle "W\ X\and Y# were manipulated between!subject "see Fig[ 2#[ The _nangle varied from 03>\ 34> and 65>$ and the _ns opened outeither Ri`htwards "³**г#\ Leftwards "×**Ð×#\ or Both"׳**Ð׳#[ This last display is a control condition in which thecon_gural properties should elicit both rightward and leftwardperceptual distortions that potentially cancel out one anotherand result in accurate bisections[ This produces a total of nineconditions[

The dimensions for the between!subject factor of geometriccomponents "W\ X\ and Y# were as follows "see Appendix 0 forfurther details#]

"a# W!displays] the "W# component or _n length was held con!stant at 6[4 cm in length\ i[e[ 49) as long as the shaft length"04 cm#\ while the X and Y components varied with _nangle "see Fig[ 2\ Panel A#^

"b# X!displays] the "X# component or horizontal portion of theshaft covered by the _n was held constant at 2[0 cm whilethe W and Y components varied with _n angle[ This Xdimension was chosen so that all stimuli would _t on the 70:1×03 in[ paper "see Fig[ 2\ Panel B#^ and

"c# Y!displays] the "Y# component or height was held constantwhile W and X components varied with _n angle[ The Ydimension "Y � 0[74 cm# was equal to that at the 03> _nangle conditions of the W!display\ and was chosen to avoidcrossing of the _ns in the control conditions[ Thus\ the03> _n angle conditions were identical to the 03> _n angle

$ The 03> control conditions in the W! and Y!display versionshave a narrow space between the inward tips of the _ns[ Toprevent subjects from using this space to guide bisection of theshaft\ either to the top left\ inward!projecting _n or the bottomright\ inward!_n was 6[54 cm long instead of 6[4 cm[ Theseconditions were presented randomly and caused the narrowspace to be displaced such that it did not overlie the true centerof the shaft[

J[ Fleming and M[ Behrmann:Neglect by perceptual illusion361

Fig[ 2[ Test and control Judd _gures^ "A# W!displays\ "B# X!displays and "C# Y!displays are illustrated[ The _n angle mag!nitudes are noted to the left of the stimuli\ and the directionthat the _ns open is noted above the stimuli[ Note that whilethe _n dimensions "W\ X\ Y# vary between experiments\ thesame magnitude of _n angles and shaft length are used in each

experiment[

conditions of the W!displays\ but the 34> and 65> displaysdi}ered "see Fig[ 2\ Panel C#[

Subjects were seated at a table and a sheet of paper containinga Judd display was presented aligned with the sagittal mid!plane of the body and positioned parallel to the edge of thetable[ The distance between the display and the body was variedso that subjects were comfortable while performing the task[Subjects were instructed to evaluate the horizontal extent of theshaft of the _gure\ to _nd the midpoint of the shaft\ and to placea transection mark through the perceived midpoint[ Carefulobservation ensured that subjects performed the task correctlyand that they did not use their hands or other instruments todetermine the midpoint of the shaft[ Ten trials were presentedin each of the nine conditions in randomized order[ Displayswere presented for an unlimited duration until transection[

The displacement of the transection mark from the truecenter of the shaft was measured to within 9[94 cm for eachdisplay\ divided by 04 cm "shaft length# and multiplied by 099to yield a percent bisection error[ As is the convention\ errorsto the right of the midpoint were denoted with a positive valueand those to the left with a negative value[ The main analysis

involved a repeated!measures analysis of variance "ANOVA#with direction "Both\ Rightward\ Leftward# and _n angle "03>\34>\ 65># as within!subject factors\ geometric component display"W\ X\ Y# as a between!subjects factor and percent bisectionerror as the dependent measure[ To understand the relativecontributions of direction and component display and theirinteractions with angle\ we also performed post!hoc testing ateach level of the factor of interest[ Pairwise comparisons weredone using Tukey HSD tests and a correction procedure tomaintain family!wise at 9[94[

Results

The main analysis designed to examine the e}ects ofdirection and angle\ as well as the contribution of eachgeometric component of the _n angle\ revealed a sig!ni_cant three!way interaction between _n angle\ direc!tion\ and component display ðF"7\097#�4[46\P³ 9[990Ł[ The mean bisection errors as a function of_n angle and direction are shown in Fig[ 3 separately forthe three component displays[ As in Mattingley et al[ ð5Ł\we observed a signi_cant di}erence in the direction of theerror depending on the direction of the _ns\ðF"1\3#�139[69\ P³ 9[90Ł^ relative to the control con!ditions averaged over all angles and component displays"mean ) error "MPE#�9[42#\ there were greater left!ward errors with Leftward _gures "MPE�−2[18# andgreater rightward errors with Rightward _gures"MPE�2[73#[%

A comparison of our 34> conditions with the 39> con!ditions of Mattingley et al[ showed no di}erences betweenthe bisection errors on Leftward stimuli "Mattingley etal[|s MPE�−4[07^ our MPE�−3[64#[& There was\however\ a signi_cant di}erence between out Rightwardcondition and that of Mattingley et al[ "Mattingley etal[|s MPE�7[68^ our MPE�3[08#[ This di}erence maypossibly be attributed to the di}erence in age betweenthe control groups in these studies "Mattingley et al[age�50[0\ S[D[�05[1\ our subjects age�13[1\S[D[�2[57#[ A comparison of errors obtained by us witha group of elderly control subjects "not presented here#to those of Mattingley et al[ revealed equivalent perform!ance across groups for both theRightward ðF"0\01#�1[18\P× 9[94Ł and Leftward ðF"0\01#�9[88\ P× 9[94Łconditions[

To examine the extent of the bisection error as a func!tion of both _n angle and direction\ we performed threerepeated measures ANOVAs at each level of the between!subject factor of the component displays "W\ X\ Y#[ Allthree ANOVAs revealed signi_cant e}ects of direction

% An analysis using the absolute values of bisection errorsrevealed that the errors on Leftward stimuli were not stat!istically di}erent in magnitude from the errors on the cor!responding Rightward stimuli ðF"0\16# � 9[710\ P×9[94Ł[

& Planned comparisons involved uncorrected separate vari!ance t!tests[ The errors from the 39> conditions involved inthese comparisons are shown in Fig[ 2\ p[ 310 of Mattingley etal[ ð5Ł[

J[ Fleming and M[ Behrmann:Neglect by perceptual illusion 362

Fig[ 3[ The mean ) bisection errors and standard error bars as a function of _n angle for test and control _gures are shown[ Positivepoints on the Y!axis indicate rightward bisection errors\ negative points indicate leftward errors[ Sample displays are included for

reference[ "A# W!display experiment\ "B# X!display\ "C# Y!display[

ðW] F"1\07#�79[76\ P³ 9[990^ X] F"1\07#�36[32\P³ 9[990^ Y] F"1\07#�067[89\ P³ 9[990Ł as well asdirection by angle ðW] F"3\25#�15[29\ P³ 9[990^ X]F"3\25#�08[31\ P³ 9[990^ Y] F"3\25#�57[42\P³ 9[990Ł[ As is evident from Fig[ 3\ for both the W! andthe Y!display experiments\ the errors with all Rightwardconditions di}ered signi_cantly from each other anddecreased monotonically as the _n angle increased\ andsimilarly all Leftward conditions di}ered from each otherwith the same pattern[ For the X!display experiment\only the Rightward "03> vs 34># and Leftward "03> vs34># comparisons were not signi_cantly di}erent fromeach other\ although the trend is in the correct directionfor both[

In order to determine the relative contribution of eachcomponent display and _n angle to the bisection error foreach of the three component displays\ separate ANOVAswere performed at each level of direction[ We considerthe control condition before moving on to the analysis ofthe Leftward and Rightward trials[ In the control "Both#condition\ the _rst important post!hoc _nding is thaterrors did not vary with _n angle ðF"1\16#�9[04\P× 9[94Ł or with component display ðF"1\16#�0[38\P× 9[94Ł\ indicating the suitability of the control con!ditions[ As expected\ the interaction of angle by com!ponent display was not signi_cant ðF"1\3#�9[79\P× 9[94Ł[ Furthermore\ for all three componentdisplays\ most of the errors\ with the exception of the 03>W!displays "MPE�9[39#\ 34> X!display "MPE�0[02#\and 03> Y!displays "MPE�9[49#\ did not di}er fromzero\ indicating the stability of the control condition[

Interestingly\ there were no di}erences in the extent ofthe error across the three component displays for trialsin the Rightward condition ðF"1\74#�0[39\ P× 9[94Ł\indicating that the magnitude of the _n angle\ irrespectiveof the dimensions of its geometric components\ is respon!sible for the errors on Rightward _gures[ However\ therewas a main e}ect of angle ðF"1\74#�39[75\ P³ 9[990]

Rightward\ averaged over component display] 03>MPE�5[13^ 34> MPE�2[64^ 65> MPE�0[44Ł\ con!_rming the previous _ndings of bisection errors varyingwith _n angle ð01\ 02Ł[

In contrast\ errors for trials in the Leftward conditiondid vary with component display\ as re~ected in theANOVA with the within!subject factor of angle andbetween!subject factor of component displayðF"1\16#�5[07\ P³ 9[90^ component display errorsaveraged over angle] W MPE�−3[76^ XMPE�−0[71^ Y MPE�−2[06Ł[ As expected from theabove\ there was a main e}ect of angle "Leftward 03>MPE�−4[48^ 34> MPE�−2[22^ 65> MPE�−9[83#\as well as an interaction of angle by component displayðF"1\3#�7[35\ P³ 9[94Ł for Leftward stimuli[ This inter!action is likely to be the underlying cause of the three!wayinteraction of direction by angle by component displaymentioned above[ Therefore\ we set out to determinewhich speci_c pairwise di}erences account for this angleby component display interaction[ Importantly\ pairwisecomparisons revealed that for the Leftward direction theonly signi_cant di}erence was between the W! and X!displays at 03> "W MPE�−6[30^ X MPE�−2[91#[

Because this single di}erence between component dis!play versions appears to have no obvious theoretical sig!ni_cance\ we suggest that the major _nding is that themagnitude and pattern of the bisection errors do notdi}er among the three component versions of the experi!ment[ In fact\ as seen from the displays in Fig[ 2\ thedimensions giving rise to a given _n angle are quitedi}erent across the three component displays\ and yetnearly the same bisection errors were generated by that_n angle regardless of whether W!\ X!\ or Y!componentwas held constant[ Consistent with this\ there was nomain e}ect of component display "W!\ X!\ or Y!display#ðF"1\16#�1[97\ P× 9[94Ł as re~ected in the main three!way ANOVA[

To verify that the results from the Judd bisection task

J[ Fleming and M[ Behrmann:Neglect by perceptual illusion363

are due to a distorted representation of space\ we utilizeda second procedure\ the end!point placement task\developed by Bisiach et al[ to demonstrate that neglectpatients have a distorted representation of space ð05\ 06Ł[Another group of 09 healthy right!handed subjects whomet the previously de_ned inclusion criteria participatedin this task[ Subjects _rst bisected 04 cm long lines with!out _ns[ They were then presented with trials containinga single dot on a page and were required to positionleft! and right!sided _ns having the same dimensions\ _nangles\ and directions as those of the W!display experi!ment at the end!points of a virtual shaft of 04 cm inlength[ An ANOVA comparing the line bisection pro!cedure involving the W!displays and the end!point _nplacement procedure\ and again with _n angle and direc!tion as within!subject factors\ revealed no di}erencebetween procedures ðF"0\07#�0[92\ P× 9[94Ł[ Fur!thermore\ there were no signi_cant two! or three!wayinteractions involving this factor of procedure[ There washowever\ as expected\ a main e}ect of directionðF"1\25#�095[93\ P³ 9[990Ł as well as an interaction ofdirection by angle ðF"3\61#�31[17\ P³ 9[990Ł[

Discussion and Conclusion

The goal of this study was to develop and study ananalog of visuospatial neglect in normal subjects using avisual illusion that distorts horizontal spatial extent[ The_rst important _nding was that\ as in Mattingley et al[ð5Ł\ the Judd _gures successfully induced a displacementof the subjective midline relative to the control conditionand the direction of the error varied with the direction ofthe _ns[ In the control conditions\ bisection was approxi!mately equivalent to the objective midpoint of the shaft^the presence of inward and outward _ns on each endof the shaft likely cancelled out inward! and outward!directed distortion e}ects[ In contrast\ when only a singleinward _n appears on one end of the shaft and an out!ward _n on the other\ bisection errors are systematicallybiased away from the inward _n and towards the outward_n[ These bisection errors may thus re~ect a distortedperception of the horizontal extent of the shaft^ theinward!projecting _ns may induce an inward!directeddistortion of the space encompassing the shaft\ and theoutward!projecting _ns may induce an outward!directeddistortion[

Interestingly\ the extent of this bias is contingent uponthe angle of the _n[ As the _n angle increased\ the bisec!tion error declined monotonically[ These _ndings areconsistent with many psychophysical studies usingMu�llerÐLyer variants indicating that the degree of per!ceived distortion decreases monotonically with increasing_n angle "for example\ ð01\ 02Ł#[ The analysis of the geo!metric components of the _n angle suggests that the mag!nitude of the _n angle itself rather than any singlegeometric component is responsible for the distorted rep!resentation of the subjective midpoint[ The central result\

then\ is that the nature and extent of the bisection errorvaries signi_cantly as a function of _n angle magnitudeand direction\ irrespective of the particular geometricdimensions comprising the _n angle[ Furthermore\ thesemodulations of bisection error by the direction and angleof the _ns were veri_ed with the end!point _n placementtask\ thereby demonstrating the robustness and e.cacyof the distortion e}ects[

The results cannot be fully accounted for by the prin!ciple of perceptual compromise as suggested by Day ð7\8Ł and Mattingley et al[ ð5Ł[ Mattingley et al[ argued thatwith an outward!projecting _n\ the extent of the shaft isperceived as a compromise between its true magnitudeand that of the entire _gure\ thereby leading to bisectiondeviations toward the outward!projecting _n[ In the caseof an inward!projecting _n\ the end!point of the shaftnearest the _n is perceived as being shifted away fromthe vertex of the _n angle\ and the subjective midpoint isthus also shifted because the shaft has shifted[ This thenleads to bisection errors[ However\ in the X!displayexperiment a perceptual compromise between the lengthof the shaft and that of the overall _gure cannot explainthe observed modulation of bisection error[ For bothinward! and outward!projecting _ns in this experiment\the "X# component of the _n angle was held constant[Thus\ the overall length of each _gure and the length ofthe shaft did not change across conditions as the _n anglevaried from 03>\ 34>\ and 65> "see Fig[ 2\ Panel B#[ If aperceptual compromise is responsible for bisection error\then this compromise should have been of the same mag!nitude for all test conditions\ yielding identical resultswith all _n angles[ This was obviously not the case[

This paper has shown that several tasks involving ananalysis of the horizontal space of the Judd _gure maybe used to distort the representation of space in normalsubjects[ While these distorted representations yieldeddeviations of subjective midpoint that are relatively small"up to 6)#\ these errors are robust across subjects andare systematically a}ected by the geometric properties ofthe display[ Moreover\ these distortions may serve as agood analog of neglect as they mirror the same behavioralpatterns seen with neglect patients on similar tasks[ Eva!luating how the representation of space is distorted innormal subjects bisecting the Judd illusion may thereforeyield a greater understanding of the disordered mediumof space in neglect[ Moreover\ once this analog is furtherexplored and understood\ it may be used further to shedlight on the relation between the space of the {physicalworld| and {psychological space|[

Acknowled`ements*The research reported in this paper wassupported by a graduate fellowship from the Center for theNeural Basis of Cognition and by grant MH43135!90 from theNational Institute of Mental Health to MB[ We thank JimNelson for his help with data collection and Jason Mattingleyfor providing us with his normal subject bisection error data[We also thank Edoardo Bisiach\ Jason Mattingley and ananonymous reviewer for their helpful comments[

J[ Fleming and M[ Behrmann:Neglect by perceptual illusion 364

References

0[ O|Keefe\ J[\ Kant and the sea!horse] An essay in theneurophilosophy of space[ In Spatial Representation]Problems in Philosophy and Psycholo`y\ ed[ N[ Eilan\R[ McCarthy\ R[ Brewer\ Blackwell\ Oxford\ UK\and Cambridge\ USA[\ 0882\ pp[ 32Ð53[

1[ Bisiach\ E[\ and Vallar\ G[\ Hemineglect in humans[In Handbook of Neuropyscholo`y\ ed[ F[ Boller\ J[Grafman\ G[ Rizzolatti\ and H[ Goodglass\ Vol[ 0[Elsevier\ Amsterdam\ 0877\ pp[ 084Ð111[

2[ Behrmann\ M[\ Moscovitch\ M[\ and Mozer\ M[ C[\Directing attention to words and nonwords innormal subjects and in a computational model]implications for neglect dyslexia[ Co`nitive Neuro!pyscholo`y\ 0880\ 7"2:3#\ 102Ð137[

3[ Drain\ M[ and Reuter!Lorenz\ P[\ Vertical orientingcontrol] Evidence for attentional bias and neglect inthe intact brain[ Journal of Experimental Psycholo`y]General\ 0885\ 014"1#\ 0Ð19[

4[ Siero}\ E[ and Posner\ M[ I[\ Cueing spatial attentionduring processing of words and letter strings in nor!mals[ Co`nitive Neuropsycholo`y\ 0877\ 4"3#\ 340Ð361[

5[ Mattingley\ J[ B[\ Bradshaw\ J[ L[ and Bradshaw\ J[A[\ The e}ects of unilateral visuospatial neglect onperception of Mu�llerÐLyer illusory _gures[ Percep!tion\ 0884\ 13\ 304Ð322[

6[ Ro\ T[ and Rafal\ R[ D[ Perception of geometricillusions in hemispatial neglect[ Neuropsycholo`ia\0885\ 23"09#\ 862Ð867[

7[ Day\ R[ H[\ Natural and arti_cial cues\ perceptualcompromise and the basis of veridical and illusoryperception[ In Human Information Processin`] Mea!sures and Mechanisms\ ed[ D[ Vickers and P[ L[Smith\ Elsevier\ Amsterdam\ 0878\ pp[ 096Ð018[

8[ Day\ R[ H[\ The foundations of veridical and illusoryperception[ In Co`nitive Approaches to Human Per!ception\ ed[ S[ Ballestros\ Lawrence Erlbaum Associ!ates\ London\ 0882\ pp[ 132Ð157[

09[ Greene\ E[ and Nelson\ B[\ Evaluating Mu�llerÐLyere}ects using single _n!set con_gurations[ Perceptionand Psychophysics\ 0886\ 48"1#\ 182Ð201[

00[ Halligan\ P[ W[ and Marshall\ J[ C[\ Toward a prin!cipled explanation of unilateral neglect[ Co`nitiveNeuropsycholo`y\ 0883\ 00"1#\ 056Ð195[

01[ Dewar\ R[ E[\ Stimulus determinants of the mag!nitude of the Mu�llerÐLyer illusion[ Perceptual andMotor Skills\ 0856\ 13\ 697Ð609[

02[ Pressey\ A[ and Martin\ N[ S[\ The e}ects of varying_ns in Mu�llerÐLyer and Holding illusions[ Psycho!lo`ical Research\ 0889\ 41\ 35Ð42[

03[ Nakagawa\ D[\ Mu�llerÐLyer illusion and retinalinduction[ Psycholo`ia\ 0847\ 0\ 056Ð063[

04[ Restle\ F[ and Decker\ J[\ Size of the Mu�llerÐLyerillusions as a function of its dimensions] Theory anddata[ Perception and Psychophysics\ 0867\ 10"5#\ 378Ð492[

05[ Bisiach\ E[\ Rusconi\ M[ L[\ Peretti\ V[ A[\ andVallar\ G[\ Challenging current accounts of uni!lateral neglect[ Neuropsycholo`ia\ 0883\ 21\ 0320Ð0323[

06[ Bisiach\ E[\ Pizzamiglio\ L[\ Nico\ D[\ and Antonucci\G[\ Beyond unilateral neglect[ Brain\ 0885\ 008\ 740Ð746[

07[ Mennemeier\ M[\ Vezey\ E[\ Chatterjee\ A[\ Rapcsak\Z[\ and Heilman\ K[\ Contributions of the left andright cerebral hemispheres to line bisection[ Neu!ropsycholo`ia\ 0886\ 24"4#\ 692Ð604[

Appendix

Dimensions of the W\ X and Y components "in cm# of the _nangles

Experiment Fin angle "># W X Y

W!display 03 6[4 6[24 0[7434 6[4 4[24 4[2765 6[4 0[84 6[2

X!display 03 2[1 2[0 9[7434 3[4 2[0 2[165 02[9 2[0 01[5

Y!display 03 6[4 6[24 0[7434 1[51 0[74 0[7465 1[96 9[49 0[74