3Journal ofNeurology, Neurosurgery, and Psychiatry 1994;57:1236-1240

Picturing unilateral spatial neglect: viewer versusobject centred reference frames

Anjan Chatterjee

AbstractVision theorists postulate that knowledgeof objects in space is shaped by differentspatial coordinate systems. An objectmay be represented by its location inrelation to the viewer, or it may be repre-sented by its own intrinsic spatial proper-ties. When patients with left sided neglectfail to respond to stimuli on the left, it isnot clear whether "left" refers to a vieweror an object centred reference frame. Touncouple these two reference frames,eight patients with neglect were asked tocentre lines and objects in photographs.Viewer centred neglect would result inimages appearing on the right side ofphotographs and object centred neglectwould result in images appearing on theleft. Four patients demonstrated viewercentred neglect and three demonstratedobject centred neglect. One patient hadvariable performance, perhaps resultingfrom competing effects of both viewerand object centred neglect. Stimuli char-acteristics did not affect the spatial coor-dinate system in which neglect occurred.These results suggest that viewer centredand object centred reference frames arefunctionally dissociable, and that patientsmay have spatial neglect predominantlyin either coordinate system.

(_7 Neurol Neurosurg Psychiatry 1994;57: 1236-1240)

Department ofNeurology and theAlzheimer Center,University ofAlabamaat Birmingham,Birmingham,Alabama, USAA ChatterieeCorrespondence to:Dr A Chatterjee, UABDepartment of Neurology,454 Sparks Center, 17207th Avenue South,Birmingham, Alabama35294-0017, USA.Received 6 December 1993and in revised form 29 April1994.Accepted 25 May 1994

Patients with left sided neglect fail to orienttoward, respond to, or act on stimuli in theleft side of space.' When asked to bisect hori-zontal lines, these patients often place theirmark to the right of the objective midpoint ofthe line, demonstrating neglect of the left side.Embedded within this simple, time honouredtask is a potential confounding factor. Thetask does not discriminate between neglect ofthe left side of space into which the lineextends, and neglect of the left side of the lineitself. The magnitude of line bisection errorsare influenced by line length as well as thespatial location of identical lines2 3 suggestingthat neglect of the left side of lines and spacefrequently co-occur and may interact in com-plex ways. The purpose of this investigationwas to try to determine if one form of neglectpredominates.

Vision theorists postulate that the brainuses different spatial reference frames to codeentities in space.4-6 A viewer centred referenceframe locates an object from the vantage pointof the viewer-that is, the left or right of,above or below the viewer. This reference

frame may be further organised retino-topically, or with respect to the position of theviewer's head or trunk. An environmentcentred reference frame locates an object byits spatial environment. Thus a specific objectin a room may be located with respect to itsunique coordinates in the room and its spatialrelation to other objects. This environmentcentred reference frame is unaffected bychanges in the viewer's location.6 An objectcentred reference frame codes spatial aspectsof an object itself, such as the object's intrinsicleft and right, and top and bottom. This refer-ence frame remains stable even when theobject is moved about. These spatial referenceframes normally operate in concert, permit-ting us to manoeuvre and interact with objectsin our spatial environment.

Unilateral spatial neglect is commonly con-ceived of as a disorder of spatially directedattention.7-'0 With damage to the right hemi-sphere, the left hemisphere's vector of spatialattention, which is directed into right space,manifests itself as "neglect" of the left side.' Itis not clear, however, which left side (viewer,object, or environment) is actually being neg-lected. Environment centred reference frames,which potentially influence neglect,6 11-13 willnot be discussed as the investigation reportedhere focuses on disentangling viewer fromobject centred coordinates.

This study considers several questions aboutunilateral spatial neglect and spatial referenceframes. Do patients predominantly haveviewer or object centred neglect? Do bothviewer and object centred neglect occur? Ifthere is a double dissociation between viewerand object centred neglect, what is the form ofthis dissociation? Do some patients demon-strate viewer and others object centred neglect?Or does the nature of the stimuli encountereddetermine the spatial coordinate system inwhich neglect occurs, so that the same patientdemonstrates viewer centred neglect in onesituation and object centred neglect inanother? To explore these issues patients withleft sided neglect performed a task designed touncouple viewer and object centred referenceframes. This uncoupling was achieved byinstructing them to photograph objects so thatthese objects would appear exactly in the cen-tre of the frame of the picture. If a patientwere to neglect the left part of space in theview finder, then the objects would appear onthe right side of the photograph. If they wereto neglect the left side of the object, then theobjects would appear on the left side of thepicture (fig 1). Thus neglect in these twospatial coordinate systems predicts oppositepatterns of performance.

1236

on March 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.10.1236 on 1 O

ctober 1994. Dow

nloaded from

http://jnnp.bmj.com/

Picturing unilateral spatial neglect: viewer versus object centred reference frames

Figure 1 Prototypiceffects of viewer centred(left) and object centred(right) neglect onplacement of lines.

Subjects and methodsEight patients from the Spain RehabilitationHospital with right hemisphere strokes weretested. Seven patients had had ischaemicstrokes and one (patient 8) a haemorrhagicstroke. These patients showed evidence of leftsided neglect as defined by screening mea-sures described later. All the patients wereright handed. There were five men and threewomen in this group. Their average age was66 (range 53 to 76) years. All subjects had CTor MRI verification of their lesions. Theirlesions were mapped in a standard manner'4and then graphically reconstructed to showthe extent of the lesion along the cortical con-vexities. Patients were tested in my laboratoryafter consenting to participate in the study.They were seated in wheelchairs and all stimuliwere centred in front of them on their mid-sagittal plane.

SCREENING MEASURESPatients bisected three lines, 1 mm wide and16, 18, or 20 cm long. Neglect on line bisec-tion was defined as errors that were greaterthan an average of 0 5 cm to the right of theobjective midpoint. Patients also performedtwo cancellation tasks. In one there were 32"O"s distributed so that eight targets wererandomly located within each quadrant.There were no distractors in this task. In thesecond, there were 24 target "A"s distributedwithin 48 distractor letters, such that therewere six targets and 12 distractors within eachquadrant. Neglect in cancellations wasdefined as the omission of more targets on theleft sides of both arrays than on the right. Fordrawing tasks, patients copied a simple linedrawing of a house next to a fence and tree,and also drew a picture of a flower from mem-ory. Neglect on these tasks was defined as notincluding left sided elements in either draw-ing. Patients were included in the study if theydemonstrated neglect in at least one kind oftask-line bisection, cancellation, or drawing.Visual fields to confrontation and extent ofhemipareses were assessed by me. Table 1shows performances of all eight patients.

EXPERIMENT 1: LINE BISECTIONAll patients bisected lines 2 cm wide and 22,23, 24, or 25 cm long twice for a total of eightbisections. Lines were horizontally presentedin a random order and placed 30 cm in front ofthe patient centred on the mid-sagittal plane.Errors were recorded to the nearest mm devia-tion from the objective midpoint of the line.

EXPERIMENT 2: PHOTOGRAPHING LINESThis experiment was designed to uncoupleviewer from object centred reference frames in

Table 1 Results ofscreening measuresPatient LB C D VFD HP

1 23 07-05 + + +2 09 12-08 + + +3 09 32-23 + + +4 02 28-17 - - +5 43 11-06 + - +6 25 25-16 + + +7 45 04-02 + + +8 56 08-08 + - +

LB = Line bisection, errors reported in average mm deviation tothe right; C = cancellation tasks, number of targets for twotasks reported (maximum 32-24); D = drawing task; VFD =visual field defect; HP = hemiparesis; + denotes presence of adeficit; - denotes absence of a deficit.

attending to a line. Black lines of the samemeasurements as in experiment 1 were fixedon a wall against a white background in a hor-izontal orientation. The lines were placed ateye level about 45 cm in front of the patients.In their right hand they held a Polaroid 600SE Impulse auto focus camera. The viewfinder in this camera has no internal mark-ings, such as a central focus ring. Eachpatient's right elbow was supported by aboard attached to the armrests of their wheel-chair. Their forearm was held in a near verticalposition such that angular rotation of the fore-arm, pronation, or supination moved theimage to the left or right of the view finder.Movement at the wrist was not restricted.Seven patients used their right eye to view theline through the camera view finder. Onepatient (patient 2) insisted on using his lefteye, claiming that this was his "hunting eye".Eye movements were also not restricted.When the patient indicated that the line wasaccurately centred in the view finder, theinvestigator standing behind the subjectpressed the shutter release.

Each line was randomly presented twice fora total of eight photographs. Deviations fromthe objective midpoint in the photographswere determined by subtracting the distance(in mm) between the right border of thephotograph and the right end of the line fromthe distance between the left end of the borderto the left end of the line. A positive numberindicates that the line was placed towards theright of the frame of the photograph, repre-senting viewer centred neglect, and a negativenumber indicates that the line was placed tothe left of the frame of the photograph, repre-senting object centred neglect (fig 1).

EXPERIMENT 3: PHOTOGRAPHING OBJECTSThis experiment was conducted to determine ifthe nature of stimuli influences the referenceframe in which neglect occurs. Confrontingreal objects might predispose attention tobe deployed in an object centred referenceframe. Six objects-a calculator, stapler, pen,scissors, remote control, and tape dispenser-were used as stimuli. These were placed on atable in front of the patients such that theirlong axes were horizontally oriented, and werecentred on the patients' mid-sagittal plane.The objects were placed about 30 cm belowpatients' eye level and about 45 cm in front ofthe patients. The patients were instructed tofirst name the object before them and thenphotograph it as described in experiment 2.

1 237

on March 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.10.1236 on 1 O

ctober 1994. Dow

nloaded from

http://jnnp.bmj.com/

1238

Similar measurements of the location of theobject within the photograph were made. Onepatient, (patient 7) refused further testingafter experiment 2.

Table 2 Deviation (mm) for each experiment

Patient Exp 1 Exp 2 Exp 3 R-L p Value

1 53 +51 +34 14-0

Picturing unilateral spatial neglect: viewer versus object centred reference frames

DiscussionThis study confirms the existence of differentspatial reference frames that vision theoristspostulate by showing that neglect occurs inboth viewer and object centred referenceframes. These results extend findings of previ-ous studies that have examined the relationbetween these two reference frames andneglect. Farah et al 6 showed 10 patients withneglect for various line drawings (such as thatof a rabbit) containing letters. The patientswere asked to name the object and then readall the letters. By rotating the drawings, viewerand object centred reference frames were dis-sociated. The object centred reference framedid not have a discernable influence on whichletters were neglected, whereas the viewercentred reference frame continued to influ-ence the spatial distribution of neglect.Chatterjee et al 3 examined the effects ofhemispatial placement and attentional cueingon line bisection tasks. They found that thebehaviour of a patient with neglect wasaffected by both conditions. When comparedwith normal subjects, however, the patient'sperformance was qualitatively different withchanges in spatial location of the lines, andonly quantitatively different when attentionwas cued to the left or right side of the linesthemselves. These results were interpreted asshowing a qualitative defect in a viewer cen-tred reference frame and not an object centredreference frame. By contrast, Driver andHalligan'6 described a case in which neglectwas determined by object centred coordi-nates. When the patient was required tomatch two objects, a decrement in perfor-mance was seen when the informationrequired to make the discrimination fell onthe object's left side. This decrement occurredeven when the objects were rotated so that thecritical area on the left side of the object waslocated to the right of the viewer. Caramazzaand Hillis reported a case of a patient with alesion in the left hemisphere who neglectedthe right side of words in reading.'7 Thispatient continued to neglect terminal portionsof words (object centred) even when thesewords were presented in mirror reversed formso that the neglected portion of the word fellon the left side of the viewer.The study of Farah et a16 raises doubts

about whether object centred neglect occursat all. The other three studies report singlecases, limiting their generalisability. The stim-uli used in these studies were very different. Itis possible that the individual results, whetherimplicating viewer or object centred neglect,were a consequence of the kinds of stimuliand testing procedures used. None of thestudies showed a double dissociation ofviewer and object centred neglect by the sametesting procedures. The strongest evidencesuggestive of the occurrence of either vieweror object centred neglect comes fromdrawings by patients with neglect. In copyingtasks patients may either neglect objects onthe left side of an array or the left side ofobjects in the array.18 9 The within patientconsistency, however, of these drawing

performances is need of further investigation.This study supports the existence of a dou-

ble dissociation between viewer and objectcentred neglect. All the patients had left sidedneglect as traditionally defined. Whenrequired to place lines and objects in the centreof photographs their performances, however,diverged. Some patients placed objects on theleft side of the photograph, demonstratingneglect of the left side of objects. Othersplaced objects on the right side of the photo-graph, demonstrating neglect of the left sideof their view. These placements resulted pri-marily from patients' angular rotations of theforearm making directional hypokinesia anunlikely explanation for these findings. As lat-eral movements of the wrist were not specifi-cally constrained, this possibility cannot,however, be completely eliminated.The results of this study also distinguish

between two alternative ways in which a func-tional dissociation between viewer and objectcentred reference frames could occur.Attentional activation of these referenceframes could be driven by the nature of thesensory stimuli in a "bottom up" manner.The same patient might then demonstrateviewer centered neglect in situations thatwould normally activate viewer coordinatesand object centred neglect in situations thatwould normally activate object coordinates.Alternatively, attentional mechanisms them-selves might be organised along different ref-erence frames, and could be selectivelydamaged in a "top down" manner. Thensome patients would have viewer centredneglect and others object centred neglectindependent of stimuli encountered.Experiments 2 and 3 were designed to distin-guish between these alternatives. A line isarguably a poor object, as it only has the prop-erty of two dimensional linear extension,without the richness of critically defining fea-tures. The objects in experiment 3 were real,with their associated shapes, colours andfunctions, that patients named before takingtheir picture. If the spatial coordinate systemin which attention is deployed is determinedby the nature of the stimuli, one would pre-dict that photographing lines might induceviewer centred neglect and photographing realobjects might induce object centred neglect.In this study, the nature of the stimuli did notinfluence the spatial coordinate system inwhich neglect was found. Rather, patientsdemonstrated the same form of neglectwhether confronting lines or objects. Thesefindings suggest that attentional systemsthemselves are organised along spatial refer-ence frames and that these distinct attentionalsystems may be disparately damaged. Theseresults are in accord with Halligan andMarshall's view that the neglect syndromeencompasses quite different underlyingimpairment.20 Even patients with similar per-formances on standard line bisections furtherfractionate when asked to photograph thesesame stimuli (fig 2).Two patients at the extremes of perfor-

mance were interesting. Patient 4 did not have

1 239

on March 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.10.1236 on 1 O

ctober 1994. Dow

nloaded from

http://jnnp.bmj.com/

Chanerlee

neglect on line bisections, deviating only anaverage of 3 mm to the right of the true mid-point. Yet she demonstrated substantialviewer centred neglect. She may have had acompletely preserved object centred referenceframe allowing her to bisect accurately whenonly confronting the object, but whenattempting to place this object within a photo-graphic frame her abnormal viewer centredframe became manifest. She was the onlypatient with a lesion confined to the dorsolat-eral and medial frontal lobe. At the otherextreme, patient 8 made the largest errors online bisections, deviating an average of 95 mm-to the right of the true midpoint. She was theonly subject who was inconsistent in whereshe placed lines and objects in her pho-tographs. She may have had similar degrees ofboth viewer and object centred neglect. Thuswhen these different reference frames pro-duced biases in opposite directions her perfor-mance collapsed into inconsistency, and whenboth reference frames produced biases in thesame direction her performance deviated dra-matically to the right. She was the only patientwith a large putaminal haemorrhage extend-ing into the adjacent white matter. The inter-pretations of these two patients' behaviourremains tentative, and require confirmationfrom future studies. Additional patients withfrontal lobe and subcortical lesions will haveto be studied to determine if these brainregions have unique contributions to spatialcoordinate systems.A final comment about the brain-behaviour

relation in these patients needs to be made. Itis a tenet of vision research that visual infor-mation is processed in a modular fashion.2' Adorsal "where" stream processes visual infor-mation to spatially locate objects. A ventral"what" stream processes visual information toidentify objects. One might expect that thedorsal where stream is important in construct-ing a viewer centred reference frame and theventral what stream is important in construct-ing object centred reference frames. If atten-tion and higher level perceptual processing arefunctionally linked, then they might also haveclose anatomical corollaries. Then lesions pre-dominantly affecting the dorsal stream wouldproduce viewer centred neglect and lesionspredominantly affecting the ventral streamwould produce object centred neglect. Thelesions of the patients in this study did notsupport this notion that the anatomical sub-strate for attentional mediation of thesespatial coordinate systems directly corre-sponds to the dorsal and ventral anatomicaldivisions of perceptual processing. Thesepatients had relatively large lesions, however.Further studies with patients with strategicallyplaced circumscribed lesions may be needed

to determine if damage to the attentionalmediation of different reference frames resultsfrom damage to distinct cerebral loci.

In summary, this study confirms the notionthat visual information is organised in thebrain along viewer and object centred spatialreference frames. Brain damage that results inleft sided neglect may disparately impair spa-tially directed attention in either coordinatesystem. Some patients neglect the left side oftheir view, others the left side of objects theyencounter, and perhaps some neglect both.

I thank Lisa Santer, MD for helpful editorial comments andThomas Groomes, MD for patient referrals. This work wassupported by an NIH-NINDS Clinical InvestigatorDevelopment Award 1KO8NS0 170201.

1 Heilman KM, Watson RT, Valenstein E. Neglect andrelated disorders. In: Heilman KM, Valenstein E, eds.Clinical neuropsychology. 2nd ed. New York: OxfordUniversity Press, 1985:243-93.

2 Marshall JC, Halligan PW. Line bisection in a case ofvisual neglect: psychophysical studies with implicationsfor theory. Cognitive Neuropsychology 1990;7:107-30.

3 Chatterjee A, Mennemeier M, Heilman KM. The psycho-physical power law and unilateral spatial neglect. Brainand Cognition 1994;25:92-107.

4 Marr D. Vision. A computational investigation into thehuman representation and processing of visual information.New York: WH Freeman and Co, 1982:397.

5 Feldman JA. Four frames suffice: a provisional model ofvision and space. Behavioural and Brain Sciences 1985;8:265-89.

6 Farah MJ, Brun JL, Wong AB, Wallace MA, CarpenterPA. Frames of reference for allocating attention tospace: evidence from the neglect syndrome.Neuropsychologia 1990;28:335-47.

7 Brain WR. Visual disorientation with special reference tolesions of the right hemisphere. Brain 1941;64:224-72.

8 Kinsbourne M. A model for the mechanisms of unilateralneglect of space. Transactions of the American NeurologicalAssociation 1970;95: 143-7.

9 Heilman KM, Valenstein E. Frontal lobe neglect in man.Neurology 1972;22:660-4.

10 Mesulam M-M. A cortical network for directed attentionand unilateral neglect. Ann Neurol 1981;10:309-25.

11 Calvanio R, Petrone PN, Levine DN. Left visual spatialneglect is both environment-centered and body-cen-tered. Neurology 1987;37:1179-83.

12 Ladavas E. Is the hemispatial damage .produced by rightparietal lobe damage associated with retinal or gravita-tional coordiantes. Brain 1987;110:167-80.

13 Mennemeier M, Chatterjee A, Heilman K. Neglect rela-tive to gravity [abstract]. J Clin Exp Neuropsychol 1992;14:85.

14 Damasio H, Damasio AR. Lesion analysis in neuro-psychology. New York: Oxford University Press, 1989:227.

15 Binder J, Marshall R, Lazar R, Benjamin J, Mohr J.Distinct syndromes of hemineglect. Arch Neurol 1992;49:1187-94.

16 Driver J, Halligan PW. Can visual neglect operate inobject-centered coordinates? An affirmative single-casestudy. Cognitive Neuropsychology 199 1;8:475-496.

17 Caramazza A, Hillis AE. Levels of representation, co-ordiante frames, and unilateral neglect. CognitiveNeuropsychology 1990;7:391-445.

18 Gainotti G, Messerli P, Tissot R. Qualitative analysisof unilateral spatial neglect in relation to laterality ofcerebral lesions. J Neurol Neurosurg Psychiatry 1972;35:545-50.

19 MarshallJC, Halligan PW. Visuo-spatial neglect: a newcopying test to assess perceptual parsing.Jf Neurol 1993;240:37-40.

20 Halligan PW, MarshallJC. Left visuo-spatial neglect: ameaningless entity? Cortex 1992;28:525-35.

21 Ungerleider LG, Mishkin M. Two cortical visual systems.In: Ingle DJ, Goodale MA, Mansfield RJW, eds. Analysisof visual behavior. Cambridge,MA: MIT Press, 1982:549-86.

1 240

on March 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.10.1236 on 1 O

ctober 1994. Dow

nloaded from

http://jnnp.bmj.com/