CLASSIFICATION OF THIS REPORT DOCUMENTATIO AD ...ASC 05 08 Space perception; Visual association €ortex; Visual attention 19. ABSTRACT (Continue on reverse If necessary and identify

UNCLASSIFIEDSECUITY CLASSIFICATION OF THIS PAGE

REPORT DOCUMENTATIO AD-A232 450 No08

Is. REPORT SECURITY CLASSIFICATION -

Unclassified2a. SECURITY CLASSIFICATION AUTHORITY 3. OISTRIBUTIONIAVAILABILITY OF REPORT

2b, DECLASSIFICATION / DOWNGRADING SCHEDULE

4. PERFORMING ORGANIZATION REPORT NUMBER(S) S. MO0NITORING ORGANIZATION REPO T-NUMBUSAFSAM-JA-88-43

6. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a. NAME OF MONITORING 0 fA~@ ,T -25,USAF School of Aerospace (if applicable) A L _F TMedicine USAFSAM/VNB MAR i51991

6c. ADDRESS (City, State, and ZIP Code) 7b. ADDRESS (City, State, and P " f)

Human Systems Division (AFSC) , )Brooks AFB, TX 78235-5301

8a. NAME OF FUNDING/SPONSORING 8b. OFFICE SYMBOL 9. PROCUREMENT INSTRUME-rIDENTIFICATION NUMBER

ORGANIZATION USAF School of (If applicable)

Aerospace Medicine USAFSAM/VNBi8c. ADDRESS (City, State, and ZIP Code) 10. SOURCE OF FUNDING NUMBERS

SPROGRAM PRO JECT TASK WORK< UNIT"

Human Systems Division (AFSC) ELEMENT NO. NO. NO. A CCESSION NO.Brooks AFB, TX 78235-5301 62202F 7930 20 01

11. TITLE-(ncude--SecurtyClassificatih). ...

Functional Specialization in the Lower and Upper Visual Fields in Humans: Its Ecological

Origins and Neurophysiological Implications

12. PERSONAL AUTHOR(S)

Fred H. Previc

13. TYPE OF REPORT 13b. TIME COVERED 14. DATE OF REPORT (Year, Month, Day) is. PAGE COUNTInterim I FROM 9/84 TO 9/90 September 90 76

16. SUPPLEMENTARY NOTATION

17. COSATI CODES 18. SUBJECT TERMS (Continue on reverse if necessary-and identify by block number)FIELD GROUP SUB.GROUP Evolution; Magno/parvocellular; Near/far; Primate

ASC 05 08 Space perception; Visual association €ortex; Visual attention

19. ABSTRACT (Continue on reverse If necessary and identify by block number)

Functional specialization in the lower and upper visual fields in humans is analyzed inrelation to the origins of the primate visual system. Processing differences between thevertical hemifields are related to the distinction between near (peripersonal) and far(extrapersonal) space, which are biased toward the lower and upper visual fields,respectively. Nonlinear/global processing is required in the lower visual field in orderto percelie the optically degraded and diplopic images in near vision, whereas objects infar vision are searched for and recognized primarily using linear/local perceptualmechanisms. The functional differe-nces befw.een near and far visual space are correlatedwith their disproportionate representations in the dorsal and ventral divisions of visualassociation cortex, respectively, and in the magnocellular and parvocellular pathways thatproject to them. Advances in far visual capabilities and forelimb manipulatory skills mayhave led to a significant enhancement of these functional specializations.

20. DISTRIBUTION /AVAILABILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATIONX' UNCLASSIFIED/UNLIMITED 0 SAME AS RPT. 0 DTIC USERS Unclassified

22a. NAME OF RESPONSIBL.Effi['VIDUAL 22b. TELEPHONE (include Area Code) 22c. OFFICE SYMBOLFred H. Prevtc, P (512) 536-3521 VNB -

DD Form 1473, JUN 86 Previous editions are obsolete. SECURITY CLASSIFICATION-OF THIS PAGEUNULAS'LA)

-. Accesion For

-,. FUHAVIORAL AND BRAIN SCIENCES (1990) 13, 519-575 NTIS CRA&I P---Printed in the United States of America DTIC TAB 0

Unannounced []JustificationDistibution .......................

By ....

Distribution I

Availability Codes

Avail and/or

Dist Special

Functional specialization in the RIlower and upper visual fields inhumans: Its ecological origins and 4neurophysiological implications

Fred H. PrevicCrew Technology Division, USAF School of Aerospace Medicine, BrooksAFB, TX 78235-5301Electronic mall: previc%[email protected]

Abstract: Functional specialization in the lower and upper visual fields in humans is analyzed in relation to the origins of theprimate visual system. Processing differences between the vertical hemifields are related to the distinction between near(peripersonal) and far (extrapersonal) space, which are biased toward the lower and upoer visual fields, respectively. Non-linearlglobal processing is required in the lower visual field in order to perqeive the optically degraded and diplopic images in nearvision, whereas objects in far vision are searched for and recognized primarily using linearllocal perceptual mechanisms. Thefunctional differences between near and far visual space are correlated with their disproportionate representations in the dorsal andventral divisions of visual association cortex, respectively, and in the magnocellular and parvocellular pathways that project to them.Advances in far visual capabilities and forelimb manipulatory skills may have led to a significant enhancement of these functionalspecializations.

Keywords: evolution, magno/parvocellular, near/far vision, primate, space perception, visual association cortex, visual attention,visual fields

1. Introduction 1.1. The distinction between near and far vision in the

One of the little-known and poorly understood features of primate

the human visual system is how it processes information I first discuss how important changes in the visual en-in its lower and upper visual fields (hereafter referred to vironment of primates dramatically increased the segre-as the LVF and- UVF). In this theoretical review, I gation of near and far visual space. There are fourhypothesize that increased functional specialization in advances of particular importance to far vision: (a) thethe LVFand UVF in primates was promoted by advances tremendous increase in the optical resolution of thein near (peripersonal) visuomotor manipulatory skills and primate eye (Polyak 1957), (b) the greater reliance onfar (extrapersonal) -visual capabilities, respectively. Pro- colored fruits as a food source, made possible by thecessing in the LVF is believed to be more non- evolution of spectrally selective cone pigments (Polyaklinear/global because of its involvement-in reaching and 1957; Snodderly 1979); (c) the use of the face as another manipulations performed in peripersonal space, important instrument of emotional expression and otherwhereas processing in the UVF is primarily linearllocal social communication (Allnan 1977), and (d) the emer-and linked to visual search and recognition mechanisms gence of a voluntary saccadic system independent af headdirected toward extrapersonal space. Finally, the en- movements. Conversely, two major developments great-hanced segregation of near versus far visual space may ly expanded the visuomotor skills used in peripersonalexplain many unique neurophysiological aspects of the space. First, the increased body size and the assumptionprimate visual system, especially regarding ttie spe- of a sitting or partially erect posture resulted in ancialization of its dorsal and ventral cortical divisions, elevation of the eyes relative to the rest of the body and

© 1990 Cambridge University Press 0140-52SX190 $5.00+.00 9 1 2 04 20851.9

Previc: Visual field specialization

facilitated the use of the hands and arms for primarily 'tmanipulative behaviors rather than postural support (Os-man Hill 1972). Second, changes in the shape of the handled to sophisticated reaching behavior in the higher ---.,.,,,....--

primates (Bishop 1962).The quartet of far visual developments presumably (Z

operated synergistically, in that: (a) the dramatic im- ",-1provement in optical quality made the emphasis on far ,'I ,vision possible; (b) the frugivorous diet and enhanced roleof facial expression provided the functional impetus for _scanning, recognition, and memory functions directed .toward distal space; and (c) the emergence of an indepen- -dent saccadic eye movement system provided an efficientmechanism for exploring extrapersonal space. The im-provement in optical quality paralleled the emergence ofa well-defined fovea in diurnal primates, although the ,ienhanced exploration of the extrapersonal visual environ-ment would extend considerably beyond the boundary ofthe fovea. The contribution of color vision is evidenced in(a) the red-green and yellow-blue spectral opponency of '

the primate geniculostriate pathways (useful for perceiv-ing the longer-wavelength colors in fruits against a blue-green forest background), (b) powerful color-specific at-tentional capabilities (instrumental in searching for andlocating fruits), and (c) mechanisms for achieving color (5\constancy (valuable in detecting fruits under differentspectral illuminations). The importance of increased fa- I.cial expression - facilitated by the evolution of a mobile I (- ._r,. \upper lip (Allman 1977)- is reflected in the extensive and t%sophisticated facial processing performed by neurons in ..the anterior temporal lobe of rhesus monkey., (Perrett et ~~al. 1984). Finally, the freeing of saccadic eye movements .from head movements - not found in nonprimate mam-mals such as the cat (Guitton et al. 1984) - resulted in an , -'4enormous expansion of the output of the saccadic system .(culminating in more than 100,000 saccades per day, ,.accord'ng to Schiller 1986). Although the increased em-phasis on far vision may have led to the advances in VV4,,saccadic eye movement control and, in turn, to the i 1 , 1.1Imarked expansion of prefrontal brain areas engaged insaccadic scanning (Goldman-Rakic 1987), it is also pos-sible that the reverse scenario occurred.

Meanwhile, the expanded use of the arms and hands in (b)retrieving and ingesting fruits and other food objectspromoted specialized mechanisms for operating in, and Figure 1. An illustration of the relationship between the lowerswitching to, the near visual environment. One such visual field (LVF) and primate reaching behavior. The latralmechanism is the "near reflex," involving a triad of ocular view (a) shows that both reaching for objects and transportingresponses (accommodation, convergence, and pupillary them toward the mouth are normally accomplished in the LVF.constriction) designed to focus on nearby objects. This The rear view (b) depicts the substantial diplopia in the proximal

reflex, along with the related capability of pursuit track- LVF as the hand approaches the object.ing, is a phylogenetically recent phenomenon largelyconfined to primates (Jampel 1959). Other specializedperceptual capabilities would prove more valuable in the above aspects of near vision apply to the LVF as well.reaching and related activities. One of these is a con- Consequently, the LVF is functionally linked to (a) oculartralateral spatial attentional systei that alluvb Litt lidl iuvuni , pwbuit and -crgcncc) associated withto be accurately guided from the visual periphery to the tracking an object as it is brought to the mouth, which isfixated object, even though it cannot be directly viewed, located inferior to die eyes, (b) an attentional systemA second capability, termed "global perception," also capable of monitoring the motion of the contralateralassists in monitoring the reaching hand despite the distor- hand from the lower visual quadrants, and (c) globaltions and reduced contrast caused by its rapid motion as perceptual capabilities that can overcome the visual deg-well as the diplopia and defocus resulting from the more radations produced by the rapid, diplopic motion of thedistal fixation (Figure 1). upper limbs in peripersonal space. Other ecological pres-

It is important to note that visible peripersonal space is sures generated by the greater motion flow beneath thealmost exclusively contained in the LVF in primates, so horizon during forward locomotion and the heightened

520 BEHAVIORAL AND BRAIN SCIENCES (1990) 13:3


shadowing found close to the ground could also contrib- ry. First, the proposed functional specializations are farute to the LVF's greater relative sensitivity to motion and from absolute and are subject to many counterexamples.luminance (see Gibson 1966). The influence of these Indeed, with the possible exception of color recognition,factors is reflected in the greater forward vection (self- neither human clinical data (Martin 1988) nor animalmotion) elicited by optical flow in the LVF as compared to neurophysiological evidence (DeYoe & Van Essen 1988)the UVF (Young & Oman 1974), and by the importance of support a strict allocation of individual perceptual func-vertical luminance gradients in judging the direction of tions to one system or the other. Not only is this con-earth's gravity (Barbour & Coss 1988). These features sistent with the extensive anatomical connections be-cannot account, however, for those LVF advantages that tween the two systems (DeYoe & Van Essen 1988), but itare unique to primates, as they are also present in the suggests that the dorsal and ventral pathways differ morevisual environments of many other mammals. Nor do in their processing strategies in different regions of visualthey relate to the major functional specializations of those space than in the particular types of information theycortical visual areas in which the LVF is predominantly process. Second, a host of dorsal-ventral differences to berepresented, as will become evident in later sections. discussed in later sections have not been addressed by

In contrast to the link between the LVF and periper- previous dichotomies. These include differences in bin-sonal space, the relationship between far vision and the ocular disparity tuning, visual field representation, visu-UVF is not nearly as exclusive, as both vertical hemifields omotor outputs, and visual attention.represent the extrapersonal portion of visual space. In- The following examples illustrate the first of the abovedeed, it will be shown that virtually no UVF advantages objections: (a) counter to the central/peripheral distinc-exist in sensory processing per se. Nevertheless, a func- tio,- most neuronal fields in parietal cortex overlap thetional link between far vision and the UVF clearly exists, foveal region and/or are influenced by foveal fixation andbased on the relationship between height in the visual pursuit (Andersen 1987; Sakata et al. 1985), whereasfield and perceived distance (see Sedgwick 1986). During inferotemporal fields average 25 degrees in diameterbinocular viewing, objects in the UVF generally appear (Dehimone el ai. 1985); moreover, the retinal distribu-more distant than those in the LVF, a phenomenon that tions ef tho twignocellular and parvocellular pathwayshas been related to the greater strength of uncrossed- (which dominal' the dorsal and ventral systems, respec-versus crossed-disparity mechanisms in the UVF and tively) exhibit coksiderably more overlap than previouslyLVF, respectively (Breitmeyer et al. 1977). A further link believed (I vingstone & Hubel 1988b); (b) counter to thebetween vergence distance and height in the field has spatial/object dichotomy, the perception of certain typesrecently been demonstrated, with divergence accom- of objects (e.g., fragmented or foreshortened ones) ispanying elevation of the head or eyes into the UVF and impaired by parietal lesions (Warrington & Taylor 1973;convergence accompanying their descent into the LVF Vaina 1989), whereas memory for the topographical rela-(Heuer et al. 1988). It is not surprising, therefore, that tionships among objects in extrapersonal space can bevisual search and saccadic scanning in far visual space are disrupted by temporal lesions in humans (Goldstein et al.more efficiently performed in the UVF. 1989); and (c) the assignment of depth and motion pro-

Processing differences between the LVF and UVF will cessing to the dorsal system conflicts with evidence thatbe reviewed and interpreted in the framework of the certain motion and depth percepts (e.g., short-rangenear-far dichotomy. Although the LVF-near and UVF-far motion; local stereopsis) clearly remain unimpaired fol-links are far from absolute, the differences between the lowing dorsal system damage in humans (Rothstein &vertical hemifields will serve as basic "markers" for less Sacks 1972; Zihl et al. 1983).extensively studied peripersonal and extrapersonal dif- A more general challenge to the functional parcella-ferences. The distinction between near and far vision will tions contained in previous dorsal-ventral schemes is theremain the overriding theme of this paper, however, and teleological one. Why, for instance, should the processingwill be directly addressed whenever relevant evidence of the features of an object be divorced from the process-exists. ing of its relation to other objects? Or, why, when

focusing on an object in front of us, should we attend to itsshape with one part of our brain and its motion and depth1.2. The relationship between the dorsal and ventral with another? Such divisions are cintradicted by the

systems and near and far vision unity of our phenomenological experience, by the impor-The enhanced split of near and far vision ultimately led to tance of motion and depth in shape processing (DeYoe &important transformations in the primate visual system. Van Essen 1988), and by the fact that objects and placesPerhaps the most important of these involved the in- can be perceived only as a particular spatial configurationcreased function,,1 segregation of the dorsal (occipito- of individual features or elements. Moreover, a grossparietal) and ventral (occipito-temporal) visual cortical division of the brain according to the particular informa-pathways. Many diffememit dichlutumiieb have beeni plu- tion piocessed ignores the fact that primates are betterposed to characterize the visual specializations of these able to attend to broad regions of space than to particularpathways, including spatiallperipheral versus object/ stimulus attributes within limited spatial regions (Nakay-central vision (Ungerleider & Mishkin 1982), motion ama & Silverman 1986). Given that the shaping of theversus color-form processing (Maunsell & Newsome higher visual pathways depends on those visual experi-1987; Van Essen & Maunsell 1983), and global (move- ences that are actively attended (Singer 1985), functionalment/depth) processing versus object identification specialization in the primate brain should above all corre-(Livingstone & Hubel 1988a). spond to the three-dimensional structure of visual space

Although all of these distinctions have merit, there are and the powerful attentional mechanisms associated withseveral reasons why none of them is altogether satisfacto- it. I

BEHAVIORAL AND BRAIN SCIENCES (1990) 13 3 521


The hypothesis that specialization in the dorsal and ary of peripersonal and extrapersonal space is easier toventral cortical pathways is linked to the different percep- define physically (i.e., at the edge of arm's reach) thantual requirements of near and far visual space offers functionally. It is obvious that visual scanning and localseveral advantages over previous schemes. For example, perceptual analyses typically directed toward extraper-it can explain most of the dorsal-ventral differences (e.g., sonal space can also be performed on objects withineye movements and visual field representational biases) reach, just as the body-centered spatial coordinate sys-neglected by previous theories (see sections 3.2 and 3.3). tem ordinarily used in peripersonal visuomotor activitiesThe near-far distinction can also account for the fact that (see sections 2.5 and 3.2) can project beyond the reach offunctional differences between the dorsal and ventral the animal. A secondary objection is that dorsal-ventralsystems are only relative, not absolute; after all, motion, specialization may be created by neurodevelopmentaldepth, and form perception must be carried out in both shaping forces (e.g., the proximity of vestibular corticalnear and far space, although not necessarily in identical areas to the dorsal pathways) that relate to, but do notways in the two sectors. Perhaps most important, the explicitly distinguish between, near and far visual spacenear-far dichotomy has a clear ecological basis that is (see section 4.1). Neither of these criticisms underminesreflected neurologically in the separate neuronal pools the basic tenet, however, that functional specialization intuned to near versus far disparities (Poggio & Poggio the higher visual pathways conforms largely to the funda-1984) and in the clinical illusions (e.g., teleopsia and mental division of the primate visual world into near andmacropsia) that the entire world is either closer or farther far visual space.away (Critchley 1953; Penfield & Rasmussen 1950).Moreover, selective impairments of the crossed- or un-crossed-disparity systems (Mustillo 1985; Richards & 1.3. Local versus global visual perceptionLieberman 1985) or peripersonal as opposed to extraper- Before reviewing the functional differences betweensonal visual functions in various developmental disorders LVF and UVF processing, I will briefly discuss thesuggest how near and far visual perception are neu- distinction between linear/local and nonlinear/globalrodevelopmentally shaped into the dorsal and ventral perception. This is arguably the most important percep-pathways (see section 4.1). tual dichotomy referred to in this paper, and serves as a

Despite its relative advantages, the near-far dichotomy cornerstone of many other perceptual and neurophysio-is by no means incompatible with previous dorsal-ventral logical theories.schemes. It can, for example, account for the specializa- "Linear" and "nonlinear" will be defined in accordancetien of the dorsal system for most motion and depth with their neurophysiological usage (Enroth-Cugell &operations, since they are more frequently performed in Robson 1966). For example, a "linear" neuron respondsperipersonal space (where motion is most rapid and to the precise spatial profile of a luminance gradient in itsconvergence and disparity information most useful). Con- receptive field and will, in the proper phase, display aversely, the specialization of the ventral system for color "null" response. The majority of such neurons also re-and object recognition can be attributed to the greater spond in a fairly linear fashion to increments in theimportance of these processes in distal space (i.e., color is contrast of the image, and do not show response satura-of little importance in monitoring the position of the limbs tion at high contrasts (Shapley & Perry 1986). By com-during reaching and objects rarely enter peripersonal parison, nonlinear cells respond to luminance gradientsspace unless already recognized). In addition, the cen- in many regions of their receptive field and summate intral/peripheral differences can be incorporated into the such a way that no spatial phase produces a null response.near-far dichotomy, in that most far visual processes are They also exhibit temporal nonlinearity (transient re-confined to the central 30 degrees because of the poor sponsiveness) and saturate at contrast levels well belowspatial resolution of the peripheral retina (see section those of linear neurons, despite their greater luminance2.5), whereas peripheral visual inputs must be attended sensitivity. Accordingly, a linear perceptual systemand processed during reaching and other peripersonal would transmit precise spatiotemporal phase informationactivities (see section 3.2.1). and thereby mediate "local" perceptual processes,

The near-far distinction has recently been emphasized whereas a nonlinear perceptual mechanism would bein the neurophysiological literature (e.g., Rizzolatti et al. more adept at processing transient, low-contrast informa-1985), although the affinity between the occipito-parietal tion in a spatially distributed ("global") fashion.pathways and near vision has long been recognized (see The difference between local and global processing isMounceastle 1976). No direct theoretical link has yet illustrated in Figure 2. For example, the small E's inbeen made between the occipito-temporal pathways and Figure 2a, the corners of the equilateral triangle in 2b,far vision, however. In Rizzolatti et al.'s (1985) scheme, and the individual motions in Figure 2c are perceivedfor instance, far vi.ual functions reside in area 8 of using contour-dependent local processing. in contrast,prefrontal cortex and the superior colliculus. Yet neither the large S in Figure 2a, the illusory triangle in Figure 2b,of these structures contains neurons capable of per- and the group motion in Figure 2c require global percep-forming the extensive computations required of complex tual processing - i.e., correspondence-matching amongform recognition in extrapersonal space (see Bruce 1988; elements rather than contour-extraction.Goldberg & Robinson 1978), so they may be more prop- The difference between local and global processing iserly considered part of the far visual system's output also illustrated by the perceptual properties of the chro-pathways. matic system, which is generally considered a spatiotem-

It must be conceded that the near-far dichotomy also porally linear system. Many aspects of global perceptionfalls short as a complete explanatory scheme. Perhaps its either weaken or collapse entirely with stimuli composedmost significant weakness lies in the fact that the bound- only of equiluminant colorcontrast (Livingstone & Hubel



EEEE perceived even over large spatial excursions, whereasshort-range motion perception depends more on an anal-E 7ysis of local positional displacements. A truly linear mo-tion system cannot perform a "speed-invariant" analysis -i.e., it cannot code the velocity of a moving grating

E independent of its spatial frequency (Maunsell & New-some 1987) - nor can it properly signal overall imagemotion when the linear sum of individual motions isambiguous, as exemplified by the aperture problem(Movshon et al. 1985). These and other distinctions be-(a) (b) tween local and global motion analysis will be moreextensively treated in connection with the dorsal system'srole in reaching behavior (see section 3.2.1).

Of course, some global percepts may be achieved viahigher-order transformations of linear outputs, especiallywhen features are integrated into recognizable faces andobjects. Nevertheless, most global perceptual processes- and especially those not highly dependent on contour: :0 i and contrast boundaries - will be shown to utilize the

. 0 0 . 0 * nonlinear pathways of the visual system, which ultimatelyS* 0 4 e * 0 project into the highest levels of the dorsal system.

0 0 0 0 2. The functional specializations of the LVF andUVF

Having briefly presented the conceptual background ofthis theory, I now review LVF/UVF (near/far) perceptualasymmetries in the foliowing major areas: reaction-time? T - * , - (RT) performance, eye movements, visual thresholds,motion perception, and visual attention. Data from

0- human visual evoked potential (VEP) studies will also beexamined in conjunction with the behavioral evidence.

Local Global Subsequent sections will then review and integrate ani-mal neurophysiological and human neuropsychologicalfindings from the standpoint of these functional spe-

(C) cializations. Much of the evidence discussed in this sec-tion was cited in a recent review by Skrandies (1987), who

Figure 2. Illustrations of the difference between "local" and concluded that perception in the LVF is generally superi-"global" perception: (a) local (small Es) vs. global (large S) or to that in the UVF. I, too, will highlight LVF advan-letters; (b) local (disks and lines) vs. global (illusory triangle) tages in many basic visual functions, emphasizing informs! and (c) local (individual element) vs. global (group) addition the relatively greater role of UVF processing inmotion percepts for alternating frames of dots. many aspects of far vision.

2.1. Reaction-time performance1988a), such as: (a) global form perception (Gregory One of the best-studied and most reliable functional1977), including the perception of the illusory triangle differences between the LVF and UVF is in reactionshown in Figure 2a; (b) global motion analysis (Cavanagh times. It was already recognized more than a century agoet al. 1985), including the group motion shown in Figure (Hall & Von Kreis 1879, cited in Woodworth 1938) that2c; and (c) global depth perception (Lu & Fender 1972), the latency of RTs to most stimuli is shorter in the LVF. Inas in random-dot stereograms. Although Livingstone and a widely cited study, Payne (1967) showed that the meanHubel (1988a) also claim that local motion and depth latency advantage of the LVF is approximately 8-10 msecperception are impaired at equiluminance, their argu- at the vertical meridian but increases to more -than 20ment is contradicted by numerous perceptual studies msec in the nasal hemiretinae. Recent studies (Gawrys-(including the above-mentioned ones) and by clinical zewski et al. 1987; Rizzolatti et al. 1987) confirm the LVFdissociations between local and global motion and depth latency advantage, at least under valid or neutral atten-perception (De Hamsher 1978; Rothstein & Sacks 1972; tional cueing. One RT study (Cocito et al. 1977), how-Zihl et al. 1983). ever, reported that the LVF RTadvantage may be limited

Motion perception also illustrates the differences be- to gratings in the low spatial frequency range.tween linear and nonlinear processing. For example, a The basis for the RT latency advantage is not entirelydistinction is often made between short-range (local) and clear, but two possibilities suggest themselves. First, itlong-range (global) motion perception (see Anstis 1978). may reflect a basic sensitivity difference between theAs illustrated in Figure 2c, long-range motion can be LVF and UVF (Skrandies 1987), in accordance with the

BEHAVIORAL AND BRAIN SCIENCES (1990) 13-3 523


greater receptor density in the tipper hemiretina, which (b) vergence movements (to fixate in the same depthprocesses LVF input (Osterberg 1935; Perry et al. 1984; plane as the approaching object), and (c) pursuit move-Van Buren 1963). The asymmetrical receptor density, ments (to adjust to the lateral slips that occur as the objectillustrated in Figure 3, parallels that found in primary is brought toward the body). Not surprisingly, the pursuitvisual cortex kTootell et al. 1988b; Van Essen et al. 1984) and vergence systems are !nearly additive (Miller et al.and may underlie the LVF advantage in luminance and 1980), exhibit similar tentooral frequency responsescontrast threshold sensitivities (see section 2.3). A second (Hine & Thorn 1987), and, along with stereomotionpossibility is that the RT differences are produced during detection, are biased toward the LVF (see section 2.4).sensorimotor integration stages, since saccadic eye move- Furthermore, pursuit, vergence, and stereomotion defi-ment latencies to similar targets generally exhibit an cits are commonly observed together after damage to theopposite asymmetry (Heywood & Churcher 1980). In- dorsal system (Girotti et al. 1982; Zihli et al. 1983). Thatdeed, the manual RT advantage in the LVF is highly the pursuit system is primarily an instrument of nearpredictable from the previously described perceptual vision is also supported by evidence that the most eflfec-relationship between the LVF and peripersonal space (in tive stimuli for the ocular following system are substantialwhich the arm and hands exclusively operate) and paral- retinal slips (more likely in peripersonal space because oflels the shorter manual RTs to crossed- versus uncrossed- motion parallax) and large targets (see Miles & Kawanodisparity targets (Gawryszewski et al. 1987). 1987).

The LVF superiority in pursuit initiation is paralleled2.2. Eye movements by the greater slow-phase gain of horizontal optokinetic

nystagmus (OKN) in the LVF (Murasugi & HowardVertical asymmetries in eye movements are highly de- 1989). Even though pursuit and OKN drive the eyes inpendent on the type of movement executed. For exam- opposite directions while following an object against apie, Tychsen and Lisberger (1986a) reported a striking moving background, their neurophysiological substratesasymmetry in eye movement accelerations to pursuit are closely entwined in that both kinds of movements aretargets in the two hemifields, with greater accelerations impaired by vestibulo-cerebellar damage (Magnusse" etin the LVF for both upward and downward target motion. al. 1986; Zee et al. 1981) and damage to the dorsal vis,'iTheir results contrast, however, with those for saccadic system (see section 3.2.1). The vestibular system may beeye movements to static targets. As reviewed by involved because of its important role in signalling theHeywood and Churcher (1980), the majority of saccade head movements that typically accompany these ocularstudies have shown a UVF advantage (particularly be- movements. The Iowpass spatial tuning and high tem-yond 10 degrees eccentricity), with no study showing a poral resolution of OKN (Schor & Narayan 1981) indicatesignificant opposite trend. that it is probably mediated by the magnocellular path-

The most plausible explanation of the LVF specializa- ways, which project into the higher stages of the dorsaltion for processing and pursuing moving targets is that visual system and are biased toward the LVF (see sectiontracking of objects as they are brought into peripersonal 3.1).space for ingestion or manipulation usually involves an The closer link between the saccadic system and theinitial descent into the LVF. This is especially true of food UVF may arise from the importance of saccades in objectbrought to the mouth, which is below the eyes. In fact, to scanning and visual search in extrapersonal space. Astrack approaching objects calls for a combination of (a) noted earlier, the two major neural components of thestereomotion detection (to discern the object's course), voluntary saccade system - the superior colliculus and

frontal eye fields - have been assigned to the far visualsystem (Rizzolatti et al. 1985). The control of saccadic eyemovements is clearly dissociable from pursuit control, as

1000 evidenced by the differential effects of vestibular lesionson the two movements (Magnusson et al. 1986) and thenumerous clinical reports of cortically damaged patientswith pursuit deficits but normal scanning of scenes and

l oo objects (Girotti et al. 1982; Pierrot-Deseilligny et al.1986) and vice versa (Luria et al. 1963). That vestibularlesions fail to disrupt saccadic eye movements is con-

_J sistent with the fact that most saccades are placed within_ _ the foveal attentional field (approximately 15 degrees

10 osurrounding the fixation point), whereas most head32 movements only accompany saccades that extend beyond5 413211 this point (Bahill et al. 1975).

5432 1 LOWER UPPER In summary, the reverse vertical asymmetry of differ-_ _ _ _ __ ent types of eye movements contrasts markedly with the

0300 600 go reliable LVF advantage for manual RTs. This difference00CENTRICITY makes sense from an ecological standpoint in that the

ECCENTRICITY oculomotor system subserves at least two major functions

Figure 3. Ganglion cell distribution in the human retina along - locking onto and following a target moving in periper-the vertical meridian. Data are from Van Buren (1963), as sonal space versus saccadic scanning in extrapersonalredrawn by Skrandies (1987), with permission fioi Springer- space - whereas arm and hand movements always occurVerlag and Dr. W. Skrandics. within the confines of peripersonal space.



2.3. Visual thresholds server (Cogan 1979). Of course, the tilt of the verticalhoropter is less relevant during movement indoors, and a

The next body of evidence to be reviewed pertains to person's fixation may occasionally be directed to somethreshold data in the areas of luminance sensitivity, intermediate distance along the ground in front. Based oncontrast sensitivity, temporal resolution, stereoacuity, subjective experience, however, the images most fre-and color perception. quently and reliably encountered at large crossed dis-

The distribution of luminance thresholds across the parities in the LVF are those contained in elevatedretina largely parallels the receptor density distribution. peripersonal space (e.g., the arms and hands).Hence, most normal subjects can detect dim targets At least one psychophysical study (Manning et al. 1987)much further into the LVF periphery, especially in the has reported a crossed-disparity advantage in both thenasal hemifield (see Riopelle & Bevan 1953; Sloan 1947). LVF and the UVF, in accordance with the results of anThis, of course, is why the typical perimetry exam shows earlier VEP study (Fenelon et al. 1986). In addressing thesuch a pronounced skewing toward the lower nasal field, discrepancy with previous research, Manning et al. notedSloan's data, however, clearly suggest that the vertical the failure of earlier studies to align the fixation point andasymmetry in luminance thresholds is much greater for stimulus frame in the same depth plane. An alternativelarger targets, which may relate to the contrast sensitivity explanation is that the vertical differences may interactand acuity findings described below. with an overall perceptual advantage for crossed-dispari-

Data from several contrast threshold studies have es- ty random-dot stereograms (Grebowska 1983; Harwerthtablished that there is a greater LVF sensitivity in the & Boltz 1979; Lasley et al. 1984; Mustillo 1985), derivedlow-to-moderate s. 'tial frequency range (Lundh et al. from the basic link between global stereopsis and near1983; Murray et al. 1983; Rijsdijk et al. 1980; Skrandies vision discussed in section 2.7.1985a). On the other hand, 'he LVY' superiority at medi- Finally, thresholds for colored stimuli appear to beum-to-high spatial frequencies appears to be substantially asymmetric in the nasal portions of the two hemifields,reduced, if it is present at all (Lundh et al. 1983; Rijsdijk with sensitivity to red, green, blue, and yellow lights allet al. 1980). This is consistent with the absence of vertical slightly greater in the LVF (Carlow et al. 1976; Hurvichasymmetry in hi"s (Cocito et al. 1977) and luminance i.JOl. These asymmetries probably relate more to thethresholds (Sloan 1947) when high spatial frequencies or distribution of luminance thresholds than to color per se,small targets are used and may account for the muddled since it is necessary to detect a target before its hue can bepicture regarding visual acuity differences. Although identified. Indeed, recent evidence indicates that equi-some studies have reported better acuity in the LVF luminant red-green color sensitivity does not differ be-(Finke & Kosslyn 1980; Low 1943; Millodot & Lamont twebn the LVF and UVF (Anderson et al. 1989). One1974), others have shown UVF superiorities (Julesz et al. researcher (Pennal 1977) claims to have found better1976; Weymouth et al. 1928). A considerable overlap of color matching in the lower left visual quadrant in nor-the UVF and LVF visual acuity distributions undoubted- mals, but the methodology and data interpretation in hisly exists, based on Low's (1943) data from 100 subjects. study can be challenged. 3

A number of studies have investigated differences in In summary, luminance and contrast thresholds fortemporal resolution across the retina (Hylkema 1942, low-spatial and high-temporal frequency stimulation mayPhillips 1933; Skrandies 1985b; Tyler 1987; Yasuma et al. be lower in the LVF. Vertical asymmetries in visual1986). All used the critical flicker fusion technique and all acuity, stereoacuity, color discrimination, and thresholdbut Yasuma et al. (1986) showed a greater flicker resolu- sensitivity outside the above ranges appear to be muchtion in the LVF. 2 For example, 13 of 20 subjects in less reliable, however.Hylkema's study had a higher fusion limit in the LVF,while only two subjects showed the opposite trend. Using 2.4. Motion perceptionthe somewhat more difficult double-flash resolution tech-nique, Skrandies (1985b) also found superior LVF tem- Vertical asymmetries in motion perception also dependporal resolution but, once again, Yasuma et al. (1986) on the type of processing required. Two perimetric stud-reported no differences between the hemifields. It ies have investigated frontal-plane motion thresholdsshould be noted that the LVF superiority observed by using small targets moving at low velocities (McColginMurray et al. (1983) in both pattern and motion detection 1960; Regan & Beverley 1983). In the latter study, mostat all spatial frequencies tested may actually ha.'e re- subjects were reported to have exhibited vertical symme-flected a LVF superiority in transient processing, as a try for both in-phase and antiphase motion, although therelatively high (15-Hz) flicker rate was used in that study. authors did not display their group data quantitatively.

Like visual acuity, stereoacuity appears to be roughly McColgin's (1960) findings present a somewhat moreisotropic across the vertical hemifields (Richards & Began comphcated picture in that movement thresholds exh-b-1973). It has also been shown, however, that the ability to ited overall vertical symmetry, despite a slight interac-detect random-dot stereograms - in which global corre- tion involving horizontal and vertical motion detectionspondences determine the percept - is faster in the LVF (lower vertical thresholds in the LVF; lower horizontalfor convergent (near) disparities but faster in the UVF for ones in the UVF). In more recent studies using lowdivergent (far) ones (Breitmeyer et al. 1975; Fox 1982, grating velocities, no UVF/LVF differences have beenJulesz et al. 1976). This finding is especially intriguing reported in either detecting or perceiving the direction orsince a truly vertical line should be seen at a crossed velocity of drifting/counterphasing gratings (Anderson etdisparity in the UVF and at an uncrossed disparity in the al. 1989; Smith & Hammond 1986).LVF during a distant fixation, because of the substantial Because all of the above studies arguably investigatedtilt of the vertical horopter toward the base of the ob- "short-range" motion perception, it may be concluded

BEHAVIORAL AND BRAIN SCIENCES (1990) 13:3 525


that little vertical asymmetry exists for this type of per- visuospatial attentional mechanisms investigated in re-cept. In contrast, several studies suggest that global cent studies that have required manual RT responsesmotion perception - especially that involving "motion- (Gawryszewski et al. 1987; Hughes & Zimba 1987;in-depth" - is better performed in the LVF. For exam- Rizzolatti et al. 1987). Gawryszewski et al.'s results, inpie, Regan et al. (1986) demonstrated that motion-in- particular, suggest a three-dimensional cubic frameworkdepth perception occurs over a larger region of the LVF for visuospatial attention, with fundamental divisionsthan the UVF. This was true for all five subjects tested ,ccurring along the lateral (left-right), vertical (up-down),and confirmed a previous finding for a single subject and depth (near-far) axes. Preliminary evidence indicates(Richards & Regan 1973).4 Regan and colleagues argue that this attentional structure may in most subjects bethat stereomotion perception differs from the perception biased toward the proximal LVF, based on the greaterof both static depth and frontal plane motion. (Indeed, "cost" of attending to near/LVF space when far/UVFstereomotion detection can be completely absent in UVF stimuli are presented than vice versa (Gawryszewski et al.regions in which static disparities are readily detected.) 1987; D. L. Robinson, personal communication). TheSince visually guided reaching in peripersonal space is basic structure of the visuospatial attentional system,arguably the most frequently encountered movement-in- along with its alleged proximal LVF bias, points to adepth, it is not surprising that stereomotion is better relationship between it and the visuomotor coordinationperceived in the LVF. Given that the image of the required in peripersonal space. Obviously, an ability toreaching hand is optically degraded and diplopic during attend specifically to the contralateral, proximal LVFfixation on the reached-for object, its motion-in-depth would allow the trajectory of the reaching hand to bemust be detected via more global mechanisms than could more accurately monitored.be used to detect static disparities. Thus, stereomotion The proximal LVF bias of the visuospatial attentioncan be perceived over large disparity ranges (Cynader & system is reflected in the vertical asymmetry associatedRegan 1982), whereas local stereopsis requires small with the "neglect" syndrome. This attentional disorder isdisparities and point-to-point correspondences. It further typically produced by right parietal lobe damage and isappears that the closely related vergence system is like- generally more pronounced in the contralateral LVFwise stimulated by global perceptual mechanisms (Jones (Bender & Furlowe 1945; Butter et al. 1989; Morris et al.& Kerr 1972; Julesz 1978) and suffers from "blind spots" 1986; Nathan 1946; Rapcsak et al. 1988; Rubens 1985).that are closely aligned with regions of poor stereomotion Examples of the LVF bias, which is particularly evidentperception (Regan et al. 1986). during the immediate recovery period, are illustrated in

No research has specifically addressed whether other Figure 4. These LVF deficits are attributable to atten-"long-range" motion percepts are vertically asymmetric,although they would be predicted to be on the basis ofclinical evidence that links the long-range process tostereomotion detection (e.g., Zihl et al. 1983). The abilityto extract an object's shape from the pure motion informa-tion generated by its three-dimensional rotation may beless readily achieved, however, at uncrossed disparitiesand by individuals who lack a crossed-disparity mecha-nism (Richards & Lieberman 1985). The link betweennear vision and this long-range percept (termed "struc- (a) (b)ture-from-motion") makes intuitive sense in that rotation-in-depth frequently occurs during many visuomotor ma-nipulations in peripersonal space (e.g., the rotation of a (d)food object as it is brought to the mouth), and may, at leastin the case of object foreshortening, be rarely encoun- "tered except in peripersonal space. The detection of A

structure-from-motion accordingly depends on the integ- -rity of middle temporal cortex (Andersen 1988), an impor-tant dorsal brain structure involved in near vision andbiased toward the LVF (see section 3.2.1).5

It can tentatively be concluded, therefore, that where- (C) (e)as short-range motion is perceived equally well in boththe UVF and LVF, stereomotion detection and other Figure 4. Examples of the LVF bias in drawings of patientsglobal motion percepts may be biased toward near vision suffering from the "neglect" syndrome. Note how the neglect ofand/or the LVF. the contralateral visual field is more pronounced in the lower

quadrant, with the LVF neglect even crossing the verticalmeridian into the ipsilateral quadrant in some cases. Figure 4a is

2.5. Visual attention reproduced from Mountcastle (1976, Figure 3), with permissionIn this section, vertical asymmetries in visual attention from MIT Press and Dr. V. B. Mountcastle; Figure 4b is fromIn texamind withverc tosymme utrie in g visual atnin Weinstein (1980, Figure 2), with permission from Cambridgeare examined with reference to two countervailing sys- University Press and Dr. E. A. Weinstein; Figures 4c and 4d aretems: a body-centered one for monitoring visuomotor from Critchley (1953, Figures 104 & 108), with permission fromactivities in peripersonal space and a retinotopic one for MacMillan Publishing Co., Figure 4e is from Benton, Levin andvisual search and scanning in extrapersonal space. Van Allen (1974, Figure 1), with permission from Pergamon

The former system is probably synonymous with the Press and Dr. A. L. Benton.



tional factors because (a) they can occur in the absence of bourne 1972). Because oculomotor biases are believed tovisual field loss per se, (b) they frequently reappear reflect heightened activation of those brain regions thatduring simultaneous left-right visual field testing, and (c) ordinarily direct eye movements to the same region ofthey can be alleviated by unilateral vestibular activation space, the cortical areas most responsible for UVF sac-(see section 4.1). Indeed, all 18 patients in Rubens's cades - that is, the more ventral regions of visual cortex(1915) study exhibited LVF neglect (in 13 of them the and the frontal eye fields (Bender 1980: Wagman 1964) -neglect was limited to the LVF), even though actual visual are presumably more active during the s. --h for objectsdefects were reported in only eight of them. Evidence of in extrapersonal space.an additional proximal bias is based on the observation The recognition of visual forms may also utilize atten-that parietal (area 7b) "neglect" in monkeys is biased tional mechanisms that are biased toward the UVF. Onetoward peripersonal space, whereas damage to the arcu- of the first researchers to note this was Piaget (1969), whoate region of prefrontal cortex (containing the frontal eye performed several experiments related to the classicfields) leads to a greater neglect of extrapersonal space illusion in which a vertical line intersecting a horizontal(Rizzolatti et al. 1985). This coincides with evidence in one appears shorter in the UVF (1) than in the LVF (T).humans that parietal neglect - along with many other Piaget argued that subjects' "centrations" (attentionalvisual lateralization phenomena - appears to be framed foci) are shifted toward the UVF in this task and aremore in terms of body-centered rather than retinotopic paralleled by a bias in their ocular fixations. Althoughcoordinates (Bradshaw, Nettleton et al. 1987; Gazzaniga alternative explanations for this illusion exist, Piaget's& Ladavas 1987; Kooistra & Heilman 1989). interpretation is supported by recent evidence from line

In contrast to the proximal bias of the body-centered bisection tasks in which most subjects bisect a verticalvisuospatial coordinate system, the extrapersonal atten- line above the midpoint (Scarisbrick et al. 1987). Sincetional system is associated more with the search for and patients suffering from the "neglect" syndrome bisectrecognition of objects in the extrapersonal visual environ- lines in the direction of the unneglected field or quadrantment. This type of attention can serve as the "glue" (see Morris et al. 1986), the bisection findings in normalswhereby color-and form cues are properly integrated into arguably reflect an "attentional shift" toward the UVF.the feature conjunctions that define an object (Treisman The relationship between form recognition and the& Schmidt 1982), but it also has an important prior stage UVF is somewhat attenuated by the tendency to fixateknown as "feature selection," which occurs approx- near the effective center of a form (Kaufman & Richardsimately 150-250 msec after the stimulus (Previc & Harter 1969), at least if its diameter subtends less than 101982). Feature-selection is generally performed in paral- degrees. This central tendency would be expected, oflel across the central visual field and greatly increases the course, given that critical components of objects andefficiency of visual search, since only those objects that scenes may be located in any sector of the fixated image.have a reasonable probability of being the ones actually UVF facial features (e.g., the eyes and bridge of the nose)searched for (i.e., those sharing one or more features with appear to be more critical, however, in facial recognitionit) serve as targets for subsequent saccadic eye move- (see Cloning & Quatember 1966; Hines et al. 1987),ments (Williams 1966). Thus, it may be assumed that, like which cannot be attributed merely to physical saliencethe saccadic system, visual search is tied to a retinotopic because the largest single facial feature is the mouth,(as opposed to body-centered) spatial coordinate system. located in the LVF. Also, Schwartz and Kirsner (1982)Although the exact diameter of the feature search field demonstrated a significant UVF RT advantage of approx-depends on the nature of the target and background imately 20 msec when both name-matching and physical-information, feature attention generally falls off rapidly matching of letter pairs was required, but no attempt tobeyond 15 degrees from fixation (see Haber & Hershen- replicate this study has apparently been made. In otherson 1973, Fig. 9.7). This distance also represents the letter and shape recognition studies, no consistent ver-maximum radius of most naturally occurring saccades, as tical hemifield differences have been reported (Engelwell as those unaccompanied by head movements (Bahill 1971; Ikeda & Takeuchi 1975).et al. 1975). Since shape detection (Engel 1971) and In conclusion, various evidence points to a UVF-linkedsaccadic accuracy (Jeannerod & Biguer 1987) both de- attentional system in humans that aids in visual searchcrease beyond this point, the size of the extrapersonal and object recognition in extrapersonal space. This sys-attentional field is probably ultimately limited by the tem presumably opposes a peripersonal visual attentionalpoor spatial resolution of the retina beyond the central 30 system that is directed toward the proximal LVF so as todegrees. [See Tsotsos: "Analyzing Vision at the Complex- prevent serious attentional and fixational imbalancesity Level" BBS 13(3) 1990.] from occurring. In fact, the LVF neglect that follows

The proposed relationship between visual search and parietal damage is mirrored by a UVF neglect created byextrapersonal space is further supported by the former's damage to structures that apparently mediate attention tobias toward the UVF. For example, visual search usually extrapersonal space (see section 4.2).commences in the UVF (especially the upper left quad-rant) and proceeds from left to right (Chedru et al. 1973;Jeannerod et al. 1968), which may account for why UVF 2.6. Visual evoked potentialstargets are more frequently identified in briefly present- The final group of studies to be reviewed in this sectioned displays (Chaiken et al. 1962). Furthermore, the are those that have recorded VEPs from the scalp ofduration of search in the UVF is typically greater than in humans in response to UVF and LVF stimulation. Thethe LVF (Chedru etal. 1973), which parallels the finding transient VEP to pattern-reversal stimulation is com-that, while performing a search of a visual display in posed of three primary components, the two earliest ofmemory, subjects typically elevate their eyes (Kins- which (Ni and P1) are maximally recorded over the

BEHAVIORAl. AND BRAIN SCIENCES (1990) 13'3 527


posterior scalp at O, and are apparently generated in VEPs (particularly P1) and RTs involves the similar mo-primary visual cortex (Previc 1988). It has been reported notonic latency increase as a function of spatial frequencyby many researchers that, at least for gratings, NI is (Lupp et a]. 1976; Parker & Salzen 1977; Vassilev &generated by UVF stimulation whereas P1 is generated Strashimirov 1979). Such lowpass tuning resembles theby LVF stimulation (Kriss & Halliday 1980; Michael & spatial contrast sensitivity function at high temporal fre-Halliday 1971; Previc 1988). One widely accepted expla- quencies (Kelly 1977), as well as the spatial tuning of thenation for the opposite polarities of N1 and P1 relates to magnocellular system. Thus, both P1 and manual RTthe inverted orientations of their UVF and LVF dipole latencies may be dominated by transient mechanisms ingenerators, located on opposite sides of the calcarine the visual system (Lupp et al. 1976; Parker & Salzenfissure (Michael & Halliday 1971). What is significant 1977).about the relationship between NI and P1 and the UVF In conclusion, VEP evidence suggests that LVF pro-and LVF, respectively, is the strikingly different func- cessing is specialized for a nonlinear analysis of rapidlytional characteristics of these components. Based on the moving (transient) visual inputs in the low spatial fre-results of many studies - including those of Plant et al. quency and low contrast ranges, whereas UVF processing(1983), Previc (1988), Ristanovic and Hajdukovic (1981), is more restricted to a linear analysis of higher spatialand Strucl et al. (198,-) - a recent study (Previc 1988) frequencies and contrasts.concluded that N1 and P1 probably reflect the outputs ofthe parvocellular and magnocellular pathways of the 2.7. Conclusionsvisual system, respectively. This conclusion was based onthe fact that N I's response is spatially linear and limited to Based on the preceding review, the specializations of themedium-to-high spatial frequencies and contrasts, UVF and LVF in man are summarized in Table 1. Thesewhereas P1 is sensitive to motion transients and predomi- specializations should be considered only relative, asnates at low spatial frequencies, high temporal frequen- even tlh, most extreme differences between the twocies, and low contrasts. Because P1 is also prominent at vertical hemifields cannot overshadow the extensive pro-high contrasts and spatial frequencies (see Figure 5), cessing that they have in common. Nevertheless, thehowever, it could alternatively manifest both magno- very existence of these vertical anisotropies providescellular and parvocellular processing. Translated into important clues as to the origin and function of variousvisual fields, this would mean that both the magnocellular visual perceptual mechanisms in man and other primates.and parvocellular pathways process LVF inputs, whereas Perhaps the most pronounced asymmetries involve theonly the latter processes UVF inputs (see section 3.1). LVF superiorities in the low spatial and high temporal

The shorter latency of NI is difficult to explain if it truly frequency ranges. Certain types of nonlinear (global)represents a UVF version of P1, given the faster RTs in processing - especially those related to transient motionthe LVF. The onset of P1, however, may be reflected in perception - accordingly appear to be performed betteran early positive potential frequently masked by NI. In in the LVF. Crossed (near) disparities are detected morefact, VEP studies have generally revealed that compara- readily in the LVF, consistent with the fact that the LVFble components are recorded at shorter latencies when appears closer to us. There further exist LVF advantageselicited by LVF stimulation (see Skrandies 1987), and in the execution of manual RTs and pursuit, vergence,that the LVF latency advantage (10-20 msec) approxi- and optokinetic eye movements, as well as a LVF bias inmates that for manual RTs. Another parallel between the peripersonal attentional system impaired by parietal

lobe damage. Conversely, the UVF is more closely tied tofar vision, and uncrossed (far) disparities may be betterproce- ed in it. Mereover, the latency of saccadic eye

(a) (b movements is shorter when they are directed toward thecldeg N,O - UVF, in accordance with the link between the UVF and

an extrapersonal attentional mechanism that facilitateso - object search and recognition.

U,.The above summary clearly indicates that whereas6o many of the LVF specializations lie within the realm of

sensory processing (e.g., low spatial/high temporal fre-40 quency analysis), the UVF specializations are more of an

•,.c attentionallperceptual nature. Not surprisingly, clinical2$ patients with altitudinal hemianopia suffer nv.,ch more

P . , severe functional impairment when LVF vision is dis-I I I rupted (Berkley & Bussey 1950), at least if the hemi-

12 4 8 12 anopia does not originate from cortical damage. TheSPATIAL FREQUENCY (c/de) above difference is predictable given the almost exclusive

confinement of peripersonal space to the LVF as opposedFigure 5. The spatial tuning of the N1 and P1 components of to the much more vertically isotropic expanse of farthe visual evoked potential (VEP) across four spatial frequen- vision. Despite the LVF advantage in certain areas,cies. The "raw" VEPs illustrating the N, and P1 components are hisoevemsiteth loca pc tae ces ( eashown in (a), while the same data are plotted in (b) as relative N, however, most local perceptual processes (e.g., visualand P1 amplitudes, referenced to the largest component ampli- acuity, stereoacuity, color vision, short-range motiontude (usually that of P ) for a given subject across all experimen- detection) are performed equally well above and belowtal conditions. Reproduced from Previc (1988, Figure 1), with the horizontal meridian. This observation conflicts withpermission from Pergamon Press. Skrandies's (1987) view that the greater receptor density



Table 1. Functional specializations of the LVF and UVF

Function LVF UVF

Depth perception crossed-disparities uncrossed-disparities(appears closer) (appears farther away)

Motor output oculomotor (pursuit, saccadic eyevergence, OKN); RTs movements

Attention peripersonal extrapersonal(body-centered) (visual search)

Spatial vision more sensitive inlow frequency range

Temporal vision more sensitive inhigh frequency range

Perception more global (e.g., more local (e.g.,stereomotion) object perception)

in the upper hemiretina is associated with a general LVF in peripersonal space. By contrast, such mechanismsprocessing superiority. Rather, I argue that the LVF would hardly seem necessary in far visual space, for twoprocessing edge (and the corresponding upper hemi- principal reasons: First, the greater distance of far objectsretinal representational bias) is limited to the transient, ensures that they are typically smaller and slower mov-low spatial frequency, and low contrast information typ- ing, which together renders them more amenable to localically encountered in peripersonal space. perceptual analysis; and second, we generally attend to

The above statement clearly accounts for the many far visual space only when we are fixating in the sameLVF specializations that can be linked to processing in depth plane, so relevant visual information in extraper-peripersonal space but do not require local disparity and sonal space generally occurs at or near zero disparity (seecontour processing, including pursuit and vergence section 3.3.1). By contrast, biologically important visualmovements, stereomotion perception, and crossed-dis- processing in near space must be monitored even when itparity detection (Bridgeman 1989; Cynader & Regan is located at a considerable crossed disparity relative to1982; Jones & Kerr 1972; Julesz 1978; Steinbach 1976). It the fixation point.may be speculated, therefore, that the primary function The link between near vision and global perception hasof global form and motion perception in the primate repeatedly been illustrated by the act of reaching duringvisual system is to facilitate visuomotor coordination in fixation on a more distant object. But the global process-peripersonal space. With few exceptions, global percepts ing superiority in peripersonal space is unlikely to be anare achieved better at crossed disparities, including exclusive consequence of reaching and the rapid imagestructure-from-motion (Richards & Lieberman 1985), motion that it entails, for several reasons. For one,random-dot stereograms (Grabowska 1983; Harwerth & although visual guidance may be critical in the develop-Boltz 1979; Lasley et al. 1984; Manning et al. 1987; ment of reaching (McDonnell 1975), only subtle reachingMustillo 1985), and various illusions and masking phe- decrements are produced when peripheral vision is oc-nomena (Fox 1982; Fox & Patterson 1981). Moreover, the cluded in adults (Paillard 1982; Perenin & Vighetto 1983).perception of illusory forms such as the triangle in Figure Second, illusory contour perception evidently occurs in2b almost always requires that the form occlude the young infants (Ghim & Eimas 1988) as well as cats (Bravobackground. In addition, global depth, form, and motion et al. 1988), both of whom rely on near vision yet engagepercepts may all be mediated by the low spatial frequen- in reaching patterns that are vastly inferior to those of thecy, "transient" channels in the visual system (Bonnet aduit human. Third, virtually all adult humans engage in1987; Ginsburg 1986; Julesz 1978; Nakayama 1983; similar types of reaching, yet only those with goodRamachandran & Cavanagh 1987; Rogers & Graham crossed-disparity systems appear to be highly proficient1982; Shulman et al. 1986; Tynan & Sekuler 1975), which at extracting structure-from-motion and related percepts.is reflected in the well-documented resistance of these Finally, the differences between the cerebral hemi-percepts to optical blurring. Finally, the inhibition of spheres in global versus local perception (see section 4.1)local perception by the global system (Navon 1977) close- cannot easily be explained by their differential involve-ly parallels the inhibition exerted by the crossed-disparity ment in reaching, since the vast majority of humans reachand transient systems over the uncrossed and sustained with the right hand (controlled by the, left hemisphere),ones, respectively (Breitmeyer 1980; Richards 1972). whereas the right hemisphere is apparently more adept at

To date, no satisfactory explanation has been provided performing global perceptual computation5. Thus, globalfor what Fox and Patterson (1981) describe as the "front" perception may ultimately be linked to near vision and(near) effect in visual perception, but the important the LVF because of the nonlinear perceptual analysecability of a global/nonlinear system to operate under required at crossed disparities in peripersonal space,degraded optical conditions would be of greatest benefit regardless of the specific visuomotor activity.



3. Neural correlates of near and far vision

This section focuses on the differences between the twomajor streams of processing in the geniculostriate systemof primates as they are manifested at both the subcortical(magnocellular/parvocellular) and the cortical (dorsal vs. V Iventral) level. The functions of these divisions will be ,examined with special reference to: (a) the biased repre-sentations of the magnocellular and parvocellular systemsin the two cortical divisions; (b) LVF-UVF anisotropies inthe visual field maps at various stages within these neural Tstreams; and (c) the.different functional requirements in V P Tiperipersonal and extrapersonal space and their role in ITshaping the unique characteristics of the two visualsystems.

3.1. Magnocellular and parvocellular pathways V3 A PO

The geniculostriate portion of the primate visual systemexhibits a considerable segregation of its two major divi- -,7sions, extending all the way from the retina to the highest 2 D~pcortical centers. At early stages, these two divisions arereferred to as magnocellular and parvocellular, hereafter IV4termed magno and parvo. Compared to other mammals, V1other visual pathways (such as the W-cell and the accesso-ry optic) are substantially reduced in importance, whilethe functional segregation of the parvo and magno path-ways is much more pronounced (Guillery 1979). For VP AITexample, cells with very different functional properties(X-cells and Y-cells) are highly intermixed in the lateralgeniculate nucleus (LGN) of the cat, whereas cells in thefour dorsal (parvo) and two ventral (magno) layers of the Figure 6. Two views of the visual areas of the macaquemonkey geniculate are anatomically and functionally monkey (from Maunsell & Newsome 1987, Figure 2). Thedistinct, abbreviations are the same as those used in the text. Re-

It has been established that the segregation of the produced, with permission, from the Annual Review of Neuro-magno and parvo pathways extends into primary visual science, vol. 10, 1987, by Annual Reviews Inc. and Dr. J. H. R.cortex (area 17, or area VI) and the higher cortical visual Maunsell.areas beyond. The major projection of the parvo pathwaysis to layers 4A and 4CP of area 17, whereas the majormagno projection is to layer 4Ca, and, in turn, 4B (Blasdel& Fitzpatrick 1984; Blasdel & Lund 1983). Using variousstaining techniques, it has been further established that 7 S I I_ D Ithe magno system largely projects dorsally to areas V2,V3, V4, MT (middle temporal cortex), MST (middlesuperior temporal cortex), and, ultimately, area 7a (pos- V4terior parietal cortex), whereas the parvo system is di-rected more ventrally toward V2, VP (ventral posteriorcortex), V4, and, ultimately, IT (inferotemporal cortex)(see DeYoe & Van Essen 1988; Maunsell 1987; Maunsell& Newsome 1987). For reference purposes, the majorvisual cortical regions and their connections are mappedand diagrammed in Figures 6 and 7.

Much of what is known about the functional specializa-tions of the magno and parvo systems is derived fromphysiological recordings in the LGN. The fundamental 2distinctions between the two systems - based on a con-sensus of many reports and reviews (Blakemore & Vital-Durand 1986; Derrington & Lennie 1984; Derrington etal. 1984; Dreher et al. 1976; Kaplan & Shapley 1982;Marrocco et al. 1982; Schiller 1986; Schiller & Malpeli Figure 7. A diagram of the hierarchy and connections of1978; Shapley & Perry 1986) - are summarized in Table macaque cortical areas (from Maunsell & Newsome 1987, Fig-2. Cells in the magno layers tend to have (a) larger ure 4). Reproduced, with permission, from the Annual Reviewreceptive fields, (b) good contrast and luminance sen- of Neuroscience, vol. 10, 1987, by Annual Reviews Inc. and Dr.sitivities, (c) lowpass spatial tuning, (d) greater non- J. H. R. Maunsell.

530 BEHAVIORAL A.D BRAIN SCIENCES (1990) 13:3


Table 2. Differences between magno and parvo cells It has been more difficult to delineate the perceptualconsequences of a loss of the magno system, but the belief

Function Magno Parvo that these outputs are neutralized at equiluminance ledLivingstone and Hubel (1988a) to perform a series of

Spatial vision lowpass tuning; bandpass tuning; perceptual experiments using equiluminant color-con-

more nonlinear more linear; high trast stimuli. On the basis of their findings, Livingstone

resolution and Hubel inferred that the magno system is responsiblefor the perception of depth and movement, as well as a

Temporal vision transient; high sustained, lowpass whole host of global cues including perspective, textureflicker resolution tuning gradients, and motion parallax. The ability of Livingstone

Contrast response high sensitivity; low sensitivity; and Hubel's studies to isolate the role of the magnosaturation at high no saturation at system in visual perception may be challenged on threecontrasts high contrasts major grounds, however. First, the exclusive ability of

the parvo system to operate at equiluminance is stillColor vision broadband opponent (e.g. controversial (Derrington et al. 1984; Schiller & Colby

red-green) 1983; Schiller et al. 1989), so it cannot be definitely

concluded that all magno inputs were silenced in thoseexperiments. Second, other factors besides a loss of

linearity (i.e., relatively fewer "null" responses), (e) magno input may contribute to the difficulties in perceiv-greater responsiveness to transient or high temporal ing images at equiluminance; these include the poorfrequency stimulation, and (0 a preference for broadband overall contrast sensitivity for red-green gratings, es-as opposed to colored stimuli. By contrast, parvo neurons pecially at mid-to-high spatial frequencies (Mullen 1985),are more likely (a) to have smaller receptive fields, (b) to and the "unnaturalness" of most equiluminant stimuli.exhibit a greater degree of spatial linearity, (c) to respond Indeed, we never encounter real-world scene layouts atbest at intermediate-to-high spatial frequencies and high equiluminance, which is significant in view of the percep-contrasts, and (d) to exhibit a greater preference for tual learning required to utilize many monocular depthchromatic (opponent) stimulation. It should be empha- cues (Deregowski 1989), Moreover, the degradation ofsized that the above differences are far from absolute, as visual perception at isoluminance evidently also extendsmany cells with "mixed" properties (e.g., lowpass but to facial perception (Perrett et al. 1984), which is gener-linear spatial responsiveness) exist near the boundary of ally believed to be performed by the ventral system andthe parvo and magno layers and in the magno layers its parvo inputs. Third, the failure at equiluminance tothemselves. Some magno units certainly respond to pure perceive local stereopsis and short-range motion - alongcolor-contrast and show better spatial acuity and linearity with the Ponzo, corridor, and other "spatial organiza-than many of their parvo counterparts. Nevertheless, tional" illusions - may be questioned on the basis of bothmagno units are clearly distinguished by their greater perceptual (Cavanagh et al. 1985; Gregory 1977; Lu &ability to rer ond to "transient" stimulation presented at Fender 1972) and neurophysiological evidence (Schillerlow spatial frequencies and contrasts, rendering them et al. 1989).6 In fact, neurons in parvo projection area VPclearly nonlinear in a general sense. Conversely, parvo do respond well to positional displacements althoughunits are ideally suited to performing a linear afialysis of they are generally not highly direction-selective in theirluminance and color contours in relatively static images, motion responses (Felleman & Van Essen 1987), andwhich is arguably their major function (Marrocco et al. most are narrowly tuned for disparity, thereby indicating1982; Shapley & Perry 1986). that they are clearly capable of mediating local stereopsis.

The physiological characterization of the parvo system In contrast to Livingstone and Hubel's approach, thehas largely been confirmed by recent studies in which present one will infer the functions of the magno systemparvo neurons were selectively destroyed by injecting from the specializations of two regions in which it isthe retina with acrylamide (Merigan 1989; Merigan & disproportionately represented - the higher dorsal cor-Eskin 1986) or, on a more local basis, with ibotenic acid tical areas and the LVF. It is proposed that the magno(Schiller et al. 1989). Following parvo damage, chromatic system, like the LVF and the dorsal system, is criticallyvision, local stereopsis, and contrast sensitivity at low linked to the visual control of reaching and other manip-temporal and high spatial frequencies are severely dis- ulations in veripersonal visual space. This is reflected inrupted, but gross form perception and se.s~tivit-y at high its greater ability to perform a nonlinear analysis oftemporal frequencies remain largely intact. Although it transient and/or low-contrast inputs in the low-spa-has not been specifically investigate-', the loss of the tial/high-temporal frequency range, which parallels thesustained parvo system should also impair the ability to superiority of the LVF (Table i). Conversely, thoseretain images long enough to place them into a long-term functions not required of visuomotor coordination inmemory store. Indeed, the response of MT neurons peripersonal space (e.g., color processing) are poorlylargely ceases within 50 insec following the onset of the represented in both the magno system and the dorsalstimulus (Maunsell 1987), wich is far too short to com- visual system. Of course, the need for transient process-plete feature integration and other memory encoding and ing during visual search and scanning - in which imagesrecognition processes (see Coltheart 1983, Previc & must be rapidly erased so succeeding ones can be pro-Harter 1982). Whether this factor underlies the pattern cessed (see Breitmeyer 1980) - mandates that at leastrecognition impairments that have been observed after some magno cell. (especially those with spatially lineirdamage to the parvo-rich inferior temporal lobe is pres- but transient properties) be located in brain areas dealingently unknown. with extrapersonal space. It is obvious, however, that the



functional properties of the parvo system (sustained, more equivocal. Although the response properties of VPlinear, color-opponent responding within relatively small (ventral V3) suggest that this area receives a greater parvoreceptive field boundaries) are better suited to the search contribution than does dorsal V3 (Burkhalter et al. 1986),for and recognition of objects in extrapersonal space. this may be true only in a relative sense (i.e., the actual

Before I review the functional specializations of the size of the parvo representation may be as large in V3, buthigher cortical streams into which the magno and parvo the presence of additional magno-type cells could lead topathways feed, several important anatomical differences a smaller percentage of parvo-type neurons being re-between these systems will be presented. The biased corded in the latter's samples). In fact, a balanced parvorepresentations of the LVF and magno system in the contribution to the LVF and UVF representations wouldhigher dorsal pathways have already been briefly dis- be predicted on the basis of (a) the full vertical extent ofcussed and will be touched on again. Additional dif- far (as opposed to near) space, (b) the vertical symmetryferences relate to the extent of myelinization and the shown in many parvo-type perceptual functions (seetopographical precision of the callosal representation section 2.7), and (c) the full vertical representation of the(Burkhalter et al. 1986). visual field in parvo-rich inferotemporal cortex (De-

It is not clear exactiy where the dominance of the LVF simone et al. 1985; Grcs et al. 1985).in the magno and dorsal systems first appears, but there is In addition to the hemifield and neuronal biases, thecurrently no evidence of such a bias in the projections dorsal and ventral pathways are also distinguished byfrom the retina to the LGN, despite the greater overall their myelinization and callosal representation patterns.LVF representation in them. Following a re-analysis of The dorsal regions surpass the ventral ones in the extentdata from Malpeli and Baker (1975), Connolly and Van of myelinization, based on a comparison of V3 and VPEssen (1984) rioted a possible LVF representational bias (Burkhalter et al. 1986). This is consonant with the rapidin the magno layers of the LGN, but difficulties in conduction from V1 to MT and other dorsal areas. Con-anatomical mapping of the geniculate (see Livingstone & versely, the ventral regions may exhibit greater topo-Hubel 1988b) make the magnitude of this alleged graphical precision in their callosal representation of theanisotropy difficult to ascertain. Upon leaving the LGN, region surrounding the vertical meridian, as again re-the LVF and UVF radiations enter the dorsal and ventral flected in the differences between V3 and VP (Burkhalterregions of primary visual cortex (V1), respectively, and et al. 1986). Although greater topographical precisionremain isolated along this axis at least as far as the next would be expected of the more linear ventral system, thisthree cortical visual stages (V2, V3, and V4). An overall precision may also contribute to the callosal mediation ofLVF bias in the spatial map of VI appears to replicate that local stereopsis in the naso-temporal overlap region sur-found in the retina (Tootell, Switkes et al. 1988; Van rounding the vertical meridian (Mitchell & BlakemoreEssen et al. 1984), but whether this bias is greater for the 1970).magno-recipient zones is not known. Evidently, how- In summary, the dorsal cortical system's unique neu-ever, a uniform distribution of 2-deoxyglucose uptake roanatomical features, including its domination by magnooccurs within the cortical representation- of the central inputs, indicates an involvement in the processing ofeight degrees after pure color stimulation, which prefer- transient, nonlinear inputs during reaching and otherentially activates the parvo system (Tootell, Silverman et near vision behaviors. By contrast, the ventral system isal. 1988). largely fed by parvo inputs and exhibits the greater

It is in VI's output to the higher visual cortical areas topographical precision required of far visual perception,that a pronounced LVF bias in the magno system first along with the expected full vertical representation of theemerges. For example, magno layer 4B projects directly visual world. The next sections review the functionalto dorsal (LVF) area V3 and to LVF-dominated MT, but specializations of the dorsal and ventral regions and pointnot to V3's ventral (UVF) counterpart (Maunsell 1987; out their important relationships to near and far vision,Maunsell & Newsome 1987). The role of MT in various respectively. Relevant findings from both the animalnear vision activities has already been alluded to, but neurophysiological and human neuropsychological liter-many of the same finctional specializations are evidenced atures will be used to support the proposed functionalin dorsal V3; indeed, the neuronal response properties in specializations.the LVF and UVF representations of V3 differ so substan-tially that a separate label (VP) has been bestowed on the 3.2. The dorsal cortical visual systemlatter (Burkhalter et al. 1986). Since little vertical spe-cialization is observed in the second cortical tier (V2), 3.2.1. Neurophysiological findings. This review focuseswhose LVF/dorsal and UVF/ventral regions both receive on areas 7a and MT, the two most widely studied regionsdirect inputs from V1, the exclusive output of layer 4B to of the dorsal system. Although some controversy hasthe higher dorsal regions apparently represents the first arisen concerning the major specialization of the formermajor functional divergence of LVF and UVF processing region - for example, command functions in personalin the primate visual system. It may be speculated that space (Mounteastle 1976) versus visual perception/atten-the purpose of the direct dorsal pathways is to avoid costly tion (Robinson et al. 1978) - a unified perspective on bothtransmission delays in processing the rapidly moving areas 7a and MT may be obtained if one views them asinformation in peripersonal space. Indeed, latency delays primarily devoted to the perceptual needs of near vision.are considerable along the multisynaptic path from VI to The following represent some of the most distinguish-V4, but are quite negligible en route to MT (Maunsell ing visual properties of area 7a neurons. First, their1987). receptive fields are typically quite large and are biasel

Evidence for a biased representation of the parvo toward the lower contralateral quadrant (Figure S). Al-system in its projections to dorsal versus ventral cortex is though the fovea is adequately represented (Andersen



reaching movements, but primarily when the reaching isvisually guided and intended to obtain biological rein-forcement (Lynch et al. 1977; Robinson et al. 1978).Finally, these neurons can be influened by pure atten-

Z" tional shifts, in the absence of eye movements per se(Bushnell et al. 1981). [See also, NUtinen: "The Role ofAttention in Auditory Information Processing as Re-vealed by Event-related Potentials and Other Brain Mea-

- sures of Cognitive Function" BBS 13(2) 1990.]S-Many of the above-mentioned properties are also ex-

hibited by neurons in MT, which serves as an importantindirect source of input to the posterior parietal cortex(see Andersen 1987; 1988; Maunsell & Newsome 1987).MT neurons, however, are more restricted to processingstimulus motion per se, as their responses do not reflectarea 7a's many higher-order perceptual and motivational

IQ influences. Although MT receptive fields are generallyI smaller, they also exhibit a strong bias toward the inferior

contralateral quadrant (Gattass & Gross 1981; Maunsell &Figure 8. The receptive field map of area 7a, illustrating the Van Essen 1983a; 1983b; 1987; Van Essen et al. 1981).bias toward the contralateral LVF. Reproduced from Robinson MT neurons are capable of detecting high rates of stim-et al. (1978, Figure 12), with permission from the American ulus motion and exhibit the transient, short-latency re-Physiological Society and Dr. D. L. Robinson. sponsiveness characteristic of the magno system, which

projects heavily to this area (Maunsell 1987). They areinvolved in many global asj.ects of motion processing,

1987), as many as 40% of 7a's neurons do not include this including: (a) detecting whole-pattern, as opposed toregion (Motter et al. 1987). Perhaps even more intrigu- local-component, motion (Movshon et al. 1985; New-ing, many posterior parietal neurons appear to code some & Wurtz 1988); (b) coding the stimulus' speed asvisual space in terms of head-centered coordinates (An- opposed to its displacement velocity (Maunsell & New-dersen et al. 1985), so that their receptive fields remain some 1987); (c) enhanced responding to antagonistic mo-stationary relative to the animal's head rather than its tion of the background relative to the direction of motionfixation. As mentioned by Andersen et al. (1985), this in the classical receptive field (Allman et al. 1985); and (d)property would be especially useful in visuomotor coordi- sensitivity to changing-size contours (Saito et al. 1986), annation, and thereby indicates an emphasis on near vi- important aspect of stereomotion perception. Area MTsion. 7 Generally, area 7a cells are not fastidious about also appears to be involved in pursuit tracking, primarilystimulus properties such as shape, orientation or color, in its initiation (Newsome & Wurtz 1988), and in thebut are highly sensitive to various motion parameters. perception of "structure-from-motion" (Andersen 1988).Many cells are responsive to movement-in-depth, pre- Finally, MT neurons exhibit fairly broad disparity tuningdominantly away from the animal (Steinmetz et al. 1987) and, like those in MST (Komatsu et al. 1988), preferand are excited by the "opponent-vector" stimulation crossed-disparity stimulation (Maunsell & Van Essen(i.e., opposite motion in different meridians) that natu- 1983b). It should be noted in conjunction with MT'srally occurs during such motion. They also appear to be purported proximal bias that the complete triad of behav-sensitive to the rotation of an object in depth (Sakata et al. iors comprising the "near reflex" has been elicited via1985). cortical stimulation only in the posterior superior tern-

Area 7a neurons are also influenced by various extra- poral sulcal region (Jampel 1959) containing what are nowretinal inputs, particularly those from "body" senses known as MT and MST.(somatosensory, proprioceptive, and vestibular) whose In accordance with the theory presented here, Maun-recipient areas also reside in parietal cortex (Hyvarinen sell and Van Essen (1987) recently hypothesized that the1982). Many neurons are sensitive to where the animal is bias of MT neuronal fields toward the inferior con-fixating (Sakata et al. 1985) and may be either excited or tralateral quadrant and crossed disparities is related toinhibited by foveal fixation. Approximately half of the the control of the contralateral hand during reaching.fixation neurons in the experiments of Sakata et al. sig- These authors further support their argument by citingnalled the distance of the fixation in depth, with two- the progressively slower speed preferences of MT neu-thirds of them preferring near fixatios. it appears that a rons in moving from the lateral periphery to the verticalsmall minority of area 7a neurons responds prior to meridian (Maunsell & Van Essen 1983a), which wouldsaccadic gaze shifts (see Andersen 1987), but this rela- correspond to the decreasing visual angle traversed bytionship may largely reflect visual attentional influences the arm as it moves further in depth toward the fixatedsince the saccade-related firing rarely occurs in the dark object. This propensity could also explain the under-(Lynch et al. 1977; Robinson et al. 1978). Many neurons representation in MT of cells that prefer downward,are also active during pursuit eye movements (Lynch et oblique target motion toward the vertical meridianal. 1977) and can distinguish self-induced motion of the (Maunsell & Van Essen 1983a), given that the reachingenvironment during pursuit from actual background mo- arm rarely moves toward the fixated object from a highertion during steady fixation (Sakata et al. 1985). A signifi- position in the visual field.cant percentage of area 7a neurons also responds to The purported role of MT in visually guided ieaching



behavior may also account for much of its ability to with the fact that most ground objects during locomotionprocess global motion. Paillard (1982), for example, dis- he near the null-disparity region surrounding the verticaltinguishes between visual channels responsible for high- horopter, which slants along the ground from the base ofvelocity (global) motion analysis and positional (local) the animal to the horizon. Finally, it is difficult from andisplacement analysis, with the former responsible for ecological standpoint to understand why brain areas soguiding the arm and hand from the lower periphery to the obviously involved in reaching and eye movement con-central target area and the latter enabling the hand to trol should perform an optical flow analysis whose chiefgrasp the fixated object in central vision. MT's global value would be to maintain postural (i.e., leg) control. Itprocessing may also be important, however, in other may be concluded, therefore, that the global motiontypes of visuomotor coordination in peripersonal space. analyses performed by area 7a and MT are primarilyFor example, the sensitivity to shearing produced by related to the control of reaching, pursuit eye movementsopposite movement in the cell's center versus surround and other peripersonal behaviors, whereas flow patternsarea would be useful in triggering the ocular following during locomotion are more likely to be analyzed byduring object pursuit (Miles & Kawano 1987), whereas dorsomedial parietal areas that receive projections fromsensitivity to rotation-in-depth would be useful in object the extreme visual periphery (Allman 1977).manipulation (see section 2.4). In summary, most of the above-cited properties of

Comparative evidence also points to the major role of neurons in the higher dorsal structures can be related-hinsg brhavor in determining the functional proper- either directly or indirectly to near vision. While dorsal

tie. ef the middle temporal region, as illustrated by the neurons are not particularly involved in color or shapestriking transformation in the visual map of MT that processing, they do perform important visual analysesemerged with the higher primates. Whereas the MT map (e.g., global motion perception) that are crucial to thein the rhesus monkey is severely biased toward the visuomotor behaviors carried out in peripersonal space.inferior contralateral quadrant, the visual representations Even the involvement of some area 7a neurons within the nocturnal prosimian galago (bushbaby) and the saccadic eye movements does not contradict the pro-nocturnal owl monkey are basically symmetrical about posed relationship between the dorsal system and nearthe horizontal axis (Allman 1988; Maunsell & Van Essen vision, as many saccades obviously occur within the1987). This difference may be attributed to the fact that confines of peripersonal space, often in conjunction withprosimians (and, to a lesser extent, the owl monkey) smooth eye movements. Indeed, the poor spatial resolu-engage in more ballistic and stereotyped reaching behav- tion of most posterior parietal neurons renders themiors than do rhesus monkeys (Bishop 1962). For example, incapable of signalling the precise location of objects inthe bushbaby is primarily an insectivore that typically space (Motter et al. 1987), so they would be of marginalplaces its hand within a few centimeters of the insect value in the visual scanning of extrapersonal space.before striking, in marked contrast to the LVF trajectoryof the hand during object reaching in diurnal primates. 3.2.2. NeuropsychologIcal findings. A large of number ofAn even more primitive and vertically symmetric tenden- human clinical studies have attempted to define the rolecy displayed by many nonprimate mammals (including of the posterior parietal area in vision; thus, I provide onlyprosimians such as the lemur) is to pick up the food object a summary depiction here. Unfortunately, most clinicaldirectly with the mouth. [See also: MacNeilage et al.: investigations inherently lack the precise stimulus con-"Primate Handedness Reconsidered" BBS 10(2) 1987.] trol and anatomical localization characteristic of the neu-

Although the above findings illustrate the tremendous rophysiological literature. Furthermore, the close ana-importance of reaching and other peripersonal behaviors tomical proximity of the dorsal and ventral systems atin dorsal system function, other researchers have sug- some stages (e.g., V4's adjacency to area MT, as shown ingested that a sensitivity to high movement velocities, Figure 6) almost guarantees that both systems are at leastmotion shearing, and opponent-vector motion demon- partially damaged in a high percentage of clinical cases.strate an additional involvement of area 7a and MT in the Finally, the posterior parietal area itself is not a unitaryprocessing of optical flow information during locomotion structure, as it contains areas such as LIP (Andersen 1987;through the environment (e.g. Allman et al. 1985; Stein- 1988) that receive substantial input from V4 and maymetz et al. 1987). Although such a role is consistent with therefore be more closely aligned with the ventral systemthe disproportionate LVF representation in these re- (see Figure 7). Despite these reservations, neuropsychol-gions, five major observations weigh against it. First, the ogical evidence generally confirms the important role ofmost rapid flow rates during locomotion are found in the the posterior parietal area in the perceptual functions ofextreme LVF periphery, whereas most MT and area 7a peripersonal space.receptive fields are located within, or at least overlap, the Two of the most prominent symptoms of posteriorcentral 20 degrees of the visual field. Second, optical flow parietal damage are a disturbance of visually guidedpatterns during locomotion are almost exclusively ex- reaching and a constellation of oculomotor impairments,panding (because we almost never move backwards), both evident in a classic parietal disorder known aswhereas the majority of area 7a neurons respond to Balint's syndrome. Many studies have documented themotion away from the animal (Steinmetz et al. 1987). reaching difficulties (see Damasio & Benton 1979; Peren-Third, opposite motion in neighboring visual regions is in & Vighetto 1983), which also constitute one of thenever produced by egomotion through the environment, cardinal symptoms of posterior parietal lobe damage inso the preference of dorsal neurons for antagonistic mo- monkeys (Lynch 1980). As for the oculomotor impair-tion in the center versus surround must be related to ments, the literature review of Girotti et al. (1982) indi-other factors. Fourth, the preference of d-rsal neurons cates that pursuit and vergence movements are muchfor near disparities and/or near fixations is inconsistent more likely to be disturbed than are voluntary and spon-



taneous saccades, a distinction also noted by Pierrot- centered coordinates (Gazzaniga & Ladavas 1987,Deseilligny et al. (1986). Optokinetic and vestibulo- Kooistra & Heihnan 1989)' and biased toward the LVF,ocular (VOR) reflexes are also frequently impaired by both of which imply the disruption of a peripersonalparietal damage (Lynch 1980, Ventre & Faugier-Gri- attentional mechanism. It is conceivable that the spatialmaud 1986). confusions and neglect stemming from the loss of a body-

A number of perceptual disturbances have also been Lentered attentional sy stem also lead to the topographicalfrequently observed following posterior parietal damage, disorientation manifested by so man) parietal patientsespecially on the right side. These include global percep- (Le, inc et al. 1985). Although the body -centered coordi-tual deficits, visual perseveration, and contralateral ne- nate system may be intimately tied to the behaviorsglect. Included among theglobal deficitsare (a) an mnability perfmimed in peripersonal space, it can also extend intoto perceive objects in unfamiliar (including three-dimen- extrapersonal space, as would be expected given thatsional) rotations (Warrington & Taylor 1973), (b) poor information contained in far vision is important in main-recognition of fragmented -visual forms and random-dot taining posture and regulating egolocomotion (Brandt etstereograms (Rothstein & Sacks 1972, Vaina 1989, War- al. 1975). The saccadic exploration of extrapersonal spacerington & James 1967), (c) loss of stereomotion and other - necessarily tied to retinotopic coordinates - appears,global motion sensitivities (Vaina 1989, Zihli et al. 1983), however, to be much less affected by parieto-occipitaland (d) deficits in overall topographical visual orientation damage than by frontal lesions, for example (Holmes(Benton et al. 1974; Girotti et al. 1982, Levine et al. 1985). 1938; Karpov et al. 1968). Thus, parietal damage inter-Global perceptual and visuomotor impairments often feres not with extrapersonal visual functioning per se, butaccompany each other, as illustrated by the fact that rather its representation in a body-centered coordinatereaching disturbances were found in all 26 patients suffer- frame.ing from topographical disorientation (without memory In summary, neuropsychological evidence reinforcesloss) in the literature reviewed by Levine et al. (1985). the view that the posterior parietal region engages mainly

As noted by Vaina (1989), the basis of the global in the visual control over peripersonal space. A dramaticperceptual deficits exhibited by right parietal patients is illustration of this involvement is reflected in a symptomthe inability to solve the correspondence problem - i.e., known as teleopsia, which is usually associated withthe ability to construct a global form from spatially dis- occipito-parietal damage. Teleopsia refers to the illusiontributed local elements. This capability involves a funda- of objects and persons as being farther away than theymentally nonlinear set of computations that involve the actually are (Critchley 1953) and may be a natural percep-magno/transient pathways to a relatively greater extent tual consequence of the loss of a near vision attentional(Bonnet 1987; Peterhans & Von Der Heydt 1989a). By system. Although considered somewhat rare, it was alsocontrast, object, face, and color recognition are preserved reported by Newman et al. (1984) and in several patientsin parietal patients (Levine et al. 1985; Warrington & of Bender et al. (1968), and may turn out upon carefulJames 1967), along with local stereopsis (Rothstein & investigation to be more common than previouslySacks 1972). Zihli et al. (1983) reported an interestiag case believed.of a patient who apparently lost only the long-rangecomponent of the motion system after presumed damage 3.3. The ventral cortical visual systemto the middle temporal and/or occipito-parietal regions.Stereomotion, pursuit tracking, and RT deficits also char- 3.3.1. Neurophysiological findings. It has long been rec-acterized this patient's syndrome, although saccadic eye ognized that neurons in the ventral (occipito-temporal)movements were unimpaired. pathways engage in substantially different processing

Another parietal disorder that appears to directly relate than their dorsal counterparts. The following review -to the LVF and/or near vision specializations is visual focusing primarily on the neurophysiology of IT and areaperseveration (palinopsia). Both Critchley (1953) and V4 - will attempt to show how the ventral system isBender et al. (1968) concluded that perseveration is most specialized for the scanning and recognition of objects inlikely to occur after damage to the occipito-parietal areas, extrapersonal space.especially on the right side. While many different types of Consistent with several decades of monkey lesion evi-palinopsia have been reported, perhaps the most coin- dence (Mishkin 1972; Sahgal & Iversen 1978; Un-mon type involves the mere prolongation of images, gerleider & Mishkin 1982), IT neurons appear to besuggesting a disorder of the transient visual pathways. In highly involved in object recognition and visual memory.normals, the transient (magno) system is believed to Most of these neurons have large receptive fields (averag-reduce visual persistence by inhibiting the sustained ing 25 degrees in diameter) that virtually always include(parvo) system (Breitmeyer 1980), which may explain the fovea, even though they may be biased toward thewhy crossed-disparity and global stimuli (also predomi- contralateral hemifield. In contrast to the response ofnantly processed by the magno system) exert important many dorsal neurons, IT neuronal activity is clearly tiedinhibitory influences over uncrossed/local ones (Navon to the animal's gaze rather than to a head- or body-1977; Richards 1972). centered coordinate system (Gross et al. 1979). In-

As mentioned in section 2.5, one of the most prominent ferotemporal neurons often prefer highly complex shapesparietal symptoms is a disturbance of visual attention that or objects, including faces, and can respond to them inis especially evident in the hernispace contralateral to the any portion of their receptive field. Despite this "global"lesion site. This neglect syndrome has been exhaustively capability, IT neurons are highly sensitive to slightinvestigated, but its origins and manifestations are still changes in the local contours of objects and shapes (De-the subject of wide debate. Recent evidence has estab- simone et al. 1985, Gross et al. 1985), although they arelished that parietal neglect is framed primarily in body- not "linear" responders in a strict sense. Like their parvo



projection neurons, they also tend to respond in a sus- orientation-selective) neurons are to a greater extenttained fashion, for as long as several hundred milliseconds found in nonstained regions (DeYoe & Van Essen 1988;in some cases (Gross et al. 1974), which may account for Livingstone & Hubel 1988a). The greater staining fortheir involvement in long-term memory encoding double-opponent neurons may reflect the leading role(Miyashita 1988). that color plays in visual object search (Williams 196),

Although the above properties are consistent with the since such staining reflects a high degree of metabolicventral system's proposed emphasis on extrapersonal activity.space, a more specific illustration of this link is provided A particularly well-studied posterior region predomi-by recent analyses of "face-specific" neurons in the fun- nantly associated with the ventral system is area V4, thedus of the superior temporal sulcus (Perrett et al. 1984), major source of input to IT. V4 is clearly involved in formwhich receives important input from the inferior tem- perception, as many of its neurons exhibit spatial fre-poral lobe. Such neurons are highly responsive to ecologi- quency, color, and orientation selectivity and prefer thatcally valid spatial transformations in extrapersonal space the stimulus presented in their extensive background(e.g., a left facial profile along with a diverted gaze and field differ in some attribute (e.g., color or spatial fre-leftward movement), but are relatively unresponsive to quency) from the stimulus presented in the center of theperceptually rare transformations (e.g., facial inversions), receptive field (Desimone & Schein 1987, Desimone etApproximately 90% of the movement-in-depth neurons al. 1985). Area V4 has been the subject of some controver-in this area prefer movement toward the animal (Jeeves et sy among neurophysiologists, who are divided over,1l. 1983), which makes sense in that receding motion whether it is involved primarily in color constancy (Zeki(stemming from backwards locomotion) is rarely encoun- 1980) or in a preliminary analysis and/or selection oftered in extrapersonal space. visual forms (Desimone & Schein 1987; Moran & De-

Lesion evidence has consistently pointed to a major simone 1985). The fact that many V4 neurons do notrole of inferotemporal cortex in visual attention in pri- respond well to color (Schein et al. 1982) conflicts with themates (Butter 1969; Soper et al. 1975, Wilson et al. 1977), claim that their major purpose is to maintain color con-not surprisingly, therefore, the average size of IT recep- stancy. On the other hand, it would be useful for at leasttive fields closely approximates the extent of the extraper- some visual search neurons to have this ability, to ensuresonal attentional field. The involvement of IT neurons in the detectability of fruit and other colored objects despitevisual attention is illustrated by the finding that stimuli cyclical fluctuations in spectral illumination. Otherwise,passing through their receptive fields do not elicit a visual search would prove much less efficient, particu-strong response unless the animal is actually fixating and larly when noncolor form cues are compromise by shad-attending to them (Gross et al. 1979). Indeed, many ing and large egocentric distances.neurons are more influenced by task-related cues and V4's role in visual search is also reflected in the spatiallysequencing than by the actual physical profile of the selective attentional enhancement exhibited by its neu-stimulus (Fuster & Jervey 1982; Gross et al. 1979). As rons (Moran & Desimone ] 985), which enables the spatialmentioned earlier, a major purpose of the far attentional coordinates of searched-for stimuli to be precisely de-system is to "glue" features into integrated wholes, so as fined. The enhancement of V4 neuronal activity canto ensure that forms composed of identical features in extend as far as 15 degrees from the fixation point (Fischerdifferent arrangements are not confused. Temporal neu- & Boch 1981), which, as mentioned earlier, correspondsrons accordingly seem to be very sensitive to the overall roughly to the limits of effective visual search in humans.spatial arrangement of individual features (Perrett et al. Neurons in V4 are probably more involved in the prepa-1984). ration for (rather than the execution of) saccades during

Posterior portions of the occipito-temporal pathways visual search, as the attentional enhancement to relevantcontain neurons that are extensively involved in process- targets does not appear to require the execution of aing form and color information (Burkhalter et al. 1986, subsequent saccade to the target (Fischer & Boch 1985).Burkhalter & Van Essen 1986; Desimone & Schein 1987; The actual programming of saccadic eye movementsDesimone et al. 1985; Felleman & Van Essen 1987; during visual search is more likely to be controlled byMoran & Desimone 1985). Most disparity-sensitive neu- other areas to which V4 projects directly, such as therons in these areas are narrowly tuned, with their peak frontal eye fields (Van Essen & Maunsell 1983). Theresponse occurring when the stimulus is in the plane of involvement of V4 in visual search may explain why it alsofixation (Burkhalter & Van Essen 1986). As discussed receives some magno input, since it must engage inearlier, this property would be expected of neurons nonlinear (transient) response processing during rapidinvolved in far vision and contrasts markedly with the scanning yet still perform at least a rudimentary formcrossed-disparity preference of MT and MST neurons. analysis. The thinly stained CO pathways that project toBesides their involvemcnt in analyzing stimulus features, V4 are ideal for this purpose, since many of their inputsthese regions also appear to mediate the selection of originate in the interlaminar regions surrounding theindividual features (Braitman 1984; Gross et al. 1971; magno layers of the LGN (DeYoe & Van Essen 1988;Haenny & Schiller 1988; Manning 1971; Wilson et al. Hendrickson 1985).1977). Feature selection requires that the anatomical Based on the above neurophysiological review, it maysubstrate of each feature "channel" be somewhat inde- be concluded that the major function of the ventralpendent, as is confirmed by the pattern of cytochrome system is to engage in sc inning and recognition of objectsoxidase (CO) staining in area V2. Cells that signal color in extrapersonal space. Posterior areas such as V4 appear(e.g., double-opponent neurons) are more co; fined to to be more involved in visual search and feature selection,thinly stained CO regions, whereas nonchromatic (e.g., whereas IT and other anterior areas evidently perform

536 BEHAVIORAL AND BRAIN SCIENCES (1990) 13'3


the feature integration and memory scarch required for It is particularly relevant to the present theory that theobject recognition. No neurophysiological evidence di- above impairments are almost always associated withrectly links the higher stages of the ve-tral system with disturbed vision in one or both UVF quadrants (Damasiothe UVF, but then again neither has any single-neuronal & Damasio 1983; Meadows 1974a). The link betweenstudy to date specifically addressed the proposed hypoth- occipito-temporal damage and the UVF is invariablyesis that far attention, rather than far sensory processing attributed to damage to adjacent inferior striate andper se, is biased toward the UVF. prestriate occipital areas (representing the UVF) or to

damage to the UVF radiations that course beneath the3.3.2. NeuropsychologIcal findings. Although the visual temporal lobe on their way from the LGN to V1, via whatdeficits produced by ventral damage in humans are rarely is known as Meyer's loop. Several factors suggest, how-as precise as those exhibited in monkey lesion studies, ever, that the link between the temporal lobe specializa-clinical evidence generally supports the view of the yen- tions and the UVF may also arise from the UVF bias of thetral system outlined above. A listing of those symptoms far attentional system (see section 2.7). For one, excisionsspecifically produced by damage to the occipito-temperal limited to the temporal pole (well away from the radiationpathways in humans includes the following disorders: fibers) can result in transient UVF deficits (Van Buren &alexia (reading loss), color agnosia, visual object agnosia, Baldwin 1958), reminiscent of the transient LVF atten-prosopagnosia (impaired facial recognition), and topo- tional deficits that occur after parietal damage. Second,graphical memory loss (see Albert et al. 1979; Cummings most temporal lobe patients are not even aware of theiret al. 1983; Damasio & Damasio 1983; Damasio et al. UVF defects (Jensen & Seedorff 1976; Van Buren &1982; Larrabee et al. 1985; Levine et al. 1985; Meadows Baldwin 1958), which suggests that they have a field-1974a; 1974b). Recent evidence (Robertson et al. 1988) specific attentional deficit in addition to the actual senso-also suggests that temporal lobe damage in humans spe- ry loss.9 Third, UVF-specific memory loss and/or neglectcifically produces a disorder of local perception (e.g., the can occur after damage to prefrontal visual areas and caninability to-perceive the small E's in Figure 2a). Most of be unaccompanied by actual sensory loss (see sectionthe above disorders tend to correlate well among them- 4.2.3). Fourth, recent cerebral blood flow experiments inselves, and may be linked to more basic impairments. For humans have revealed that the inferior temporal andexample, prosopagnosia has been attributed to a general inferior (UVF) occipital regions compose a unified systemdisturbance of visual memory (Damasio et al. 1982; He- that is activated during visual imagery tasks (Goldenbergcaen & Albert 1978), while defective visual search and et al. 1989). Finally, it has been reported that pattern-scanning mechanisms may underlie object agnosia sensitive epilepsy - presumably associated with temporal(Bender & Feldman 1972; Kinsbourne & Warrington lobe dysfunction - is much more easily elicited by UVF1962). than LVF stimulation (Soso et al. 1980).

Rarely do the above symptoms accompany those re- The most plausible reason why no extrapersonal ana-sultiig from parietal lobe damage. For example, pros- logue of the parietal neglect syndrome has been reportedopagnosia correlates poorly with deficits in reaching, after damage to the temporal lobe (or, for that matter, anyvisuospatial orientation, and global form perception other area) in humans is that neuropsychological testing(Levine et al. 1985; Wasserstein et al. 1987), as illustrated generally permits free eye movements under continuousby the fact that not one of the 28 prosopagnosics in Levine viewing, effectively piecluding a retinotopically medi-et al.'s (1985) review showed a reaching defect. Although ated neglect from revealing itself. By comparison, thosedisorientation in relation to one's surrounding environ- animal studies demonstrating UVF and/or extrapersonalment may occur after either occipito-parietal or occipito- neglect have maintained precise fixational control.temporal damage, the latter is more typically followed by Nevertheless, the poor recognition of upper facial fea-a loss of topographical memory and the former by the loss tures in prosopagnosia (Gloning & Quatember 1966) andof a spatial coordinate system in which to place the other indications of distorted or absent upper-half formremembered landmarks (Levine et al. 1985). Yet, the perception (see Levine et al., Figure 2) point to a UVFparietal patient may be able to describe the place to be attentional neglect in many temporal lobe patients.visited without being able to describe how to get there, so In summary, the neuropsychological literature con-some spatial memory obviously remains intact. In fact, firms that the ventral portion of the primate visual systemtemporal lobe involvement in spatial memory is sup- is involved in the scanning and recognition of objects,ported by recent studies of patients with anterior tem- faces, and other images in extrapersonal space, but not inporal lobe damage (Goldstein, et al. 1989; Jones-Gotman the reaching, oculomotor, and global perceptual func-1986), although the integrity of the hippocampus (also tions performed by the dorsal system. In contrast to thedamaged in these studies) is considered more critical for latter's LVF representational bias, the anterior temporalthe nonegocentric representation of visual space (Jones- visual areas may be linked to the UVF only via higher-Gotman 1986). Given, however, that the hippocampus order attentional mechanisms. It is not known whetherreceives heavy projections from the temporal lobe and ventral damage leads to an exaggerated emphasis on nearthat the visual component of the hippocampal amnesic vision, mirroring the teleopsia associated with parietalsyndrome can be almost completely duplicated by bilat- damage. Although Penfield and Rasmussen (1950) re-eral destruction of the temporal lobes (Horel 1978), it may ported that distance illusions in either direction werebe presumed that an extrapersonal spatial representation common after temporal lobe stimulation, a map of theiris also contained in the temporal visual areas, as would be stimulation sites clearly indicates that some must haverequired of an area so intimately involved in visual search been located in the middle and superior temporal areas,and scanning. which are hypothesized to subserve near vision.



4. The origins of the near and far visual systems areas MT and MST. The involvement of the vestibularIn the primate system in parietal function has long been recognized, and

As argued previously, a major limitation of previous the deficits of the parietally lesioned patient are analo-dorsal-ventral theories is their failure to put forth plausi- gous to those of labyrinth-defective individuals (Barlowble ontogenetic and/or phylogenetic scenarios to explain 1970). Parietal symptoms such as the loss of a body-the origins of these specializations. Consequently, the fi- centered coordinate system, spatial disorientation, oculo-nal major section of this paper briefly explores the roles of motor (pursuit, OKN, and VOR) deficits, limb biases, andnear and far vision in shaping the primate visual system. contralateral neglect all resemble the symptoms of uni-

It is now widely accepted that visual experience plays a lateral or bilateral vestibular damage. Indeed, vestibularcritical role in the development of the visual system. Of destruction itself can produce unilateral neglect (Jean-course, some genetic influences may directly guide the nerod & Biguer 1987), whereas unilateral vestibular stim-formation of at least crude topographical representations ulation can alleviate the symptoms of parietal neglectand specializations, whereas others (such as those de- (Cappa et al. 1987; Rubens 1985).scribed in section 1.1) may determine the specific milieu The justification for assigning the vestibular system anwherein the neurodevelopmental shaping of the primate important role in the development of the dorsal system isvisual system occurs. Yet it is clear that visual experience based on three principal factors. First, the vestibularexerts a profound influence over the formation of higher system is an ontogenetically precocious system, withcortical visual maps and specializations (see Hyvarinen many vestibulo-ocular reflexes well-established by birth1982; Maunsell & Van Essen 1987; Von Der Malsburg & (Eviatar et al. 1979), so it is quite capable of influencingSinger 1988) and can even alter the visual representations the postnatal development of the higher visual pathways.at subcortical and retinal levels (Shatz & Stretavan 1986; Second, it has been hypothesized that left-right ves-Von Der Marlsburg & Singer 1988). Furthermore, the tibular asymmetry may be responsible for a further sub-experiential shaping of the visual system depends not division of near versus far visual perception into the leftonly on what is "seen," but also on what is "attended" and right hemispheres of most humans (see Previc, in(Singer 1985). preparation). Finally, disruption of dorsal visual function-

Thus, it is likely that differential experiences in the ing has been hypothesized to occur in at least two disor-near and far visual realms shape the primate visual sys- ders linked to vestibular dysfunction: strabismic am-tem. Even the most basic manifestation of the functional blyopia and visuospatial dyslexia (Previc, in preparation).link between the LVF and near vision - namely, the Regarding the second argument cited above, I recentlygreater ganglion density in the LVF, present in most theorized that the existence ofleft vestibi-ar dominancemamnmals (Skrandies 1987) - may be subject to experien- in most humans leads to a greater involvement of the righttial shaping, because it is not evidenced in the visual parietal lobe in vestibular processing (see also Penfieldperimetry maps of human infants (Schwartz et al. 1987). 1957). The specialization for vestibular functions may, inLikewise, the parvo representation in the LGN -'-n be turn, promote the right parietal area's greater emphasisreduced by refractive error during developmen. (Von on near vision, based on two principal observations.Noorden et al. 1983), which limits access to the fai visual First, recent neuropsychological studies have convinc-environment. In the following sections, it will be shown ingly established a dissociation between the hemisphereshow one developmental influence - the vestibular sys- in terms of global versus local processing (Delis et al.tem's role in reaching and oculomotor integration - may 1986; Vaina 1989), with the right hemisphere superior incontribute greatly to the development of near vision and global perception (e.g., the large S in Figure 2a) and thethe dorsal visual system. Conversely, important transfor- left in local perception (e.g. the small E's in Figure 2a).mations of the primate LGN and its ventral projection Second, the right hemisphere is believed to possess aareas, the superior colliculus, and the frontal visual areas more bilateral attentional system, whereas the left ap-can all be traced to the emergence of far vision, pears to be more exclusively concerned with the right

side of space (Heilman & Van Den Abell 1980). This isreflected in both the neglect of the left side of space4.1. Nar vision and the vestibular n,,stm following right-parietal damage and the left-right confu-

As noted by Ornitz (1970), the vestibular system provides sions following left-parietal damage (McFie & Zangwillone of the most important sources of information about 1960). The bilateral attentional system of the right hemi-the position of the body in space. Although it is by no sphere can be related to its putative emphasis on nearmeans exclusively concerned with peripersonal space, vision because the guidance of proximal arm movementsthe vestibular system does provide critical inputs for is more bilateral than is that of distal movements (Haaxmamany near vision functions, including pursuit eye move- & Kuypers 1975).lleUit. (LatItIa1Ii et al. -1u, 1 "agnusson Ct al. ,I n°Q and An .mportant contribution of the vestiblar system tolimb control (Fukuda 1959, Jeannerod & Biguer 1987). the development of the dorsal system is also suggested by

Despite decades of controversy, the most definitive its alleged role in strabismic amblyodia and viuospatialevidence to date has localized the major cortical yes- dyslexia, both linked to possible magno dysfunctiontibular projection to the posterior bank of the superior Regarding the former disorder, it appears that low-con-temporal gyrus (Friberg et al. 1985), in accordance with trast and low-luminance (i.e., magno) -isual performancePenfield's (1957) proposed locus based on cortical stim- is impaired in the amblyopic eye of strabismics, whereasulation data. Although the efferents from this region hae spatial resolution and suprathreshold (i.e., parvo) %isionnot been precisely mapped, a major vestibular projection are left relatively intact (Barbeito et al. 1987, Flom &stream apparently courses dorsally into the posterior Bedell 1985, Hess & Bradley 1980). The putative mragnoparietal lobe (see Hyvarmen 1982), in close proximity to loss coincides with the deficits exhibited by strabismics in

538 BEHAVIORAL AND BRAIN SCIENCES (1990) 13-3


velocity discrimination, contrast sensitivity, RTs, and i summary, the pre-eminent factor responsible for thepursuit eye movements (Hamasaki & Flynn 1981, Hilz et dorsal visual system's emphasis on peripersonal spaceal. 1977; Tychsen & Lisberger 1986b), including the may be its anatomical relationship with the cortical yes-failure to show the normal LVF advantage in pursuit tibular pathways. The contribution ofvestibular inputs totracking. In general, impaired global spatial percept.. -i i the specialization of MT and other doi sal areas for motionthe face of presened pattern vision and acuity charac- processing and %isuoinotor coordination is obviously notterizes the strabismic amblyopia deficit, whereas the limited to primates, however, as tLiese phylogeneticallyreverse may be true for the refractive amblyopias and older areas engage in similar functions in nonprimatesother "far" visual disorders (see Barbeito et al. 1987, (Schiller 1986; Tusa et al. 1989). Rather, the uniqueFlom & Bedell 1985, Hilz et al. 1977). Although a pc- behavioral interactions performed in peripersonal spaceripheral labyrinthine defect in most cases of strabismus is (see section 3.2.1) ultimately determine the specific LVFunlikely, various "vestibular" symptoms are frequently biases of the dorsal visual system in primates.observed in this disorder, including an abnormal VOR(Hoyt 1982; Schor & Westall 1984), abnormal caloric andpostrotary nystagmus (Salman & Von Noorden 1970: 4.2. The emergence of far vision and Its neuralSlavik 1982), and atypical postural biases (Niederlandova consequences& Litvinenkova 1973). It is not known whether the Although the emergence of far vision gave rise to wide-strabismic perceptual deficits are caused by the oculo- spread transformations in the primate brain, the organiz-motor imbalances also manifested in this disorder or v ice ing factors governing the neurodevelopmental shaping ofversa, but recent evidence indicates that relatives of the far visual system are more obscure than in the case ofstrabismics who do not actually show the oculomotor the near system. One possibility is that the ventral loca-deviations may nonetheless exhibit many of the "magno" tion of far vision may also be determined largely byperceptual deficits (Tychsen 1989). nonvisual influences. Perhaps the most important of

More convincing evidence links impaired vestibular these are the close proximity of primary auditory cortexpi ocessing with visuospatial dyslexia (see Previc, in prep- (since the auditory system is also concerned with extra-aration). For example, many reading-disabled individuals personal space) and the limbic system (whose involve-suffer from vestibulo-cerebellar oculomotor symptoms ment in emotional associations, cognitive maps, etc.,(Levinson 1988) and manifest severe postural problems implies an emphasis on the extrapersonal environment).when relying exclusively on vestibular inputs (Horak et Far visual structures such as the superior colliculus andal. 1988). Moreover, all postmortem aialyses to date have the frontal eye fields are not closey linked anatomicallyindicated prominent neuroanomalies in the posterior with the limbic region, however, and their auditorysuperior temporal gyral region of dyslexic brains (Ga- responses are more influenced by visual factors (i.e., gazelaburda et al. 1985). It has also been suggested that direction) than their visual responses are influenced byvisuospatial dyslexics suffer from a specific disorder of the auditory factors (see Bruce 1988). A second possibility istransient (magno) pathways (Martin & Lovegrove 1984), that far visual functions are assigned by default, andbased on evidence of increased visual persistence and reside only in those regions that have not already becomereduced low :patiai frequency sensitivity (Badcock & entwined with near vision. This hypothesis is somewhatLovegrove 181, Martin & Lovegrove 1984). Consistent consistent with developmental evidence, in that thewith this hypothesis are the many reports (e.g., Adler- crossed-disparity system does appear to develop firstGrinberg & Stark 1978, Bogacz et al. 1974, Pavlidis 1981) (Mustillo 1985) although the magno system may takeof pursuit deficits among dyslexics, although this finding longer to mature (Hickey 1977). The growth of the farhas not always been replicated (see Brown et al. 1983). visual sy..tem was not accomplished merely by enlarging

The vestibular hypothesis is not the only one that may those areas already outside the near visual system'sbe invoked to account for the posterior parietal lobe's bias sphere of influence, however, its development also en-toward near vision and the LVF, but alternative explana- tailed the active transformation of structures such as thetions do not suffice as easily. For example, the dorsal LGN and superior colliculus that once served mainly in alocation of the LVF projection in primary visual cortex near visual capacity (see below).could promote a near vision bias throughout the entire In contrast to its somewhat shrouded ontogeny, the fardorsal brain because of the confinement of peripers(,nal visual system's phylogenetic origins can be more clearlyspace to the LVF. This scenario is contradicted, however, traced to the ecological developments described in sec-by the enormous crossover of LVF and UVF projections tion 1.1. This is particularly true of the profound transfor-from V1 to the highest stages of the dorsal and ventral mations wrought in three important visual regions; thesystems, which suggests that higher-order visual field LGN and its parvo-projection areas, the superior col-biases dcpcid tfl-fio oL L,. L Uulrukit Jf olXt., spt- hculus, and the visual areas of the prefrontal cortex.cializations than on the location of primary visual corticalrepresentations. Second, the proximity of somatosensory 4.2.1. The lateral geniculate nucleus and the ventral sys-cortex to the posterior parietal area may influence the tern. The primate LGN differs in two important respectslatter's specialization for reaching, eye movements, etc. from that of other mammalian species. First, functionallyArea 7a is only circuitously linked with somatosensory distinct cell classes are much more segregated anatom-cortex (Pandya & Seltze 1980), however, and it is less ically, and second, the size of the parvo system is marked-evident that somatosensory disturbances can account for ly increased (see Guillery 1979). Both of these trends arethe gross reaching biases, disorientation, and inattention also observed within the primate order itself. Comparedto peripersonal space manifested in the parietal neglect to the LGN of diurnal monkeys, for instance, that of thesyndrome. prosimian galago contains additional laminae besides the

BEHAVIORAL AND BRAIN SCIENCES (1990) 13,3 539


magnn and parvo ones, exhibits much less striking func- shape, and other far visual analyses in the diurnal rhesustional segregation among laminae, and does not include than in the nocturnal owl monkey (see Burkhalter et al.as great a parvo representation (Norton & Casagrande 1986). Also, the presence of large numbers of "face-1982). sensitive" neurons in the inferior and anterior temporal

Many features of the LGN are formed during prenatal regions has apparently not been reported in any speciesdevelopment, but the final distribution of laminae and other than the rhesus monkey, as is consistent with thecell types depends on experience. Much of the postnatal enhanced role of facial expression as a means of socialshaping is brought about by binocular competitive in- communication in the higher primates.teractions, but attentional factors may also play a role It is difficult to account for the major expansion of the(Singer 1985) since a cortifugal attentional gating is be- parvo/ventral pathways in primates if they merely sub-lieved to operate at the level of the geniculate (Singer serve object perception, since there is no indication that1977). Thus, the tremendous expansion of the parvo cats and other mammals cannot learn object discrimina-system in the higher primates might have been influ- tions quite well. Rather, what differentiates primatesenced by the increased attention to far vision. This may from most other mammals is the significant enlargementalso be true for the "mixed" cells in the magno layers, of extrapersonal space, the striking increase in the abilitywhich have many properties in common with parvo units. to scan it, and the rexaarkable visuospatial memory forIn the cat, a small percentage of neurons possess both X- objects in it (Menzel 1973).like and Y-like properties, which apparently reflects tnesynapsing of geniculate X-cells onto Y-cell axons (Gar- 4.2.2. The superior colliculus. Another subcortical visualraghty 1985). Evidently, the percentage of such "mixed" structure that has been radically altered by the emer-cells substantially increases after disruption of the Y-cell gence of far vision in the primate is the superior col-pathways by monocular deprivation (Freidlander et al. liculus. This structure plays an important role in visual1982), suggesting an increased competitive advantage fui, search, saccadic eye movement initiation, and orientationthe linear X-cells under these circumstances. It may be toward visual and other stimuli. It has accordingly beenspeculated that the emphasis on far vision in the primate assigned to the far visual system in primates (Rizzolatti etproduces a similar bias against transient, nonlinear Y- al. 1985; Rizzolatti & Camarda 1987).cells, thereby increasing the tenacity (and survival rate) of Perhaps the most distinct feature of the primate col-parvo neurons in anatomically defined magno space. liculus is its adoption of a binocular visual coordinateAlthough the greater laminar segregation is believed to system relative to the animal's fixation. Unlike the colliculidecrease many of the competitive interactions between of other mammals, which receive inputs that emanatecell types in the primate geniculate (Guillery 1979), it has almost exclusively from the contralateral retina, the pri-nevertheless been shown that a disruption of far vision mate colliculus receives a substantial input from thecan indeed selectively reduce the percentage of parvo ipsilateral retina in initiating saccades confined to theunits (Von Noorden et al. 1983). contralateral visual field (Allman 1977; Goldberg & Robin-

It can be predicted, then, that parvo units should son 1978; Sprague et al. 1973). This transformation haslargely be confined to the visual search field (i.e., the been attributed to the evolution of frontally directed eyescentral 30 degrees), whereas cells with nonlinear recep- and good binocular vision (Allman 1977), but cats also havetive field properties should predominate in the more a largely contralateral retinal projection, despite theirperipheral LVF regions in which visually guided reaching good binocular vision. It is more likely that this changeand other more global processing occur. Parvo and magno represents an adaptation to the increased emphasis on farcells do, in fact, disproportionately represent the central vision, inwhichbinocularinputlimitedtothecontralateraland peripheral portions of the visual field, respectively visual field would be important in targeting saccades. By(Connolly & Van Essen 1984, Derrington & Lennie contrast, off-axisimagesofnearobjectsthatarenotdirectly1984), although these biases may be smaller than pre- fixated (and therefore the target ofpotentialfixation)are soviously believed (Livingstone & Hubel 1988b). The mag- disparate that only one eye's input is useful in directingno representation may also be slightly biased toward the saccades to the contralateral hemifield. Ordinarily, thisLVF (Connolly & Van Essen 1984), but a definitive would be the input from the contralateral eye, which isconfirmation of this anisotropy may be precluded by dominant in that same hemifield (Miles 1954).current anatomical mapping limitations. The above interpretation is consistent with the results

The major expansion of the LGN parvo system in of recent perceptual studies that have investigated theprimates is paralleled by a large increase in the size of the ability to ube eye-of-origin (utrocular) information. Utro-temporal lobe, in which it predominantly terminates. In cular discriminations are only possible for the transient,fact, Diamond and Hall (1969) argued that the expansion low spatial frequency stimuli that predominate in pe-of the temporal vismal areas represents one of the most ripersonal space (Martens et al. 1981), whereas eye-of-salient features of primate evolution. Although it has origin information cannot be used during target searchsince been shown that at least some rudimentary parallels (Wolfe & Franzel 1988), which entails a binocular scan-exist between cats and monkeys (Campbell 1978), other ning of extrapersonal space. The putative link betweenresearchers continue to doubt whether a clearcut homol- utrocular processing and peripersonal space is also con-ogue of the prnate inferotemporal areas is found in this sistent with the role of area MT - a near vision region - inspecies (Tusa & Palmer 1980). The primate occipito- ocular rivalry, discrimination, and dominance (Logothet-temporal pathways, however, are distinguished not only is & Schall 1989; Previc, in preparation).by their anatomical expansion but also by their unique The binocular representation of the contralateral hemi-functional specializations. For example, VP (the UVF field in the primate colliculus may be largely dependentportion of V3) appears to be much more involved in color, upon the descending effeients from primary visual cor-



tex. Indeed, coi tical cooling has been shown to abolish inferior regions). Rather, the general importance of thealmost all visual responses in the intermediate and prefrontal areas in attention and behavioral control - asdeeper layers of the primate colliculus (Schiller et al. well as their strategic ability to monitor processing in near1974). Although the corticotectal projection emanates and far visual space - may mean that they play a pivotallargely from 'lie magno pathways (Schiller et al. 1979), its role in regulating the balance between the two posteriordisproportionate representation of the central visual field visual systems. Perhaps the extensive interconnectious(Wilson & Toyne 1970) indicates that these magno inputs between the superior and inferior frontal regions (Gold-probably mediate the transient processing required in man-Rakic 1987) serve as the force that ultimately inte-the scanning of extrapersonal space. grates the primate's perceptual interactions within its

Other indications that the superior colliculus has been three-dimensional visual world.transformed into a far visual structure are the lack of The specialization of the inferior frontal lobe for objectdirectional sensitivity in the vast majority of its cells and recognition and memory parallels that of the inferiorthe inability to elicit combined head and eye movements temporal lobe, wheream the adjacent dorsal region sur-upon stimulation (see Goldberg & Robinson 1978). Both round.ig the principal sulcus may duplicate many of theof these properties contrast strikingly with the properties specializations of the posterior parietal area (Bruce 1988,of the cat colliculus, in which a high degree of directional Goldman-Rakic 1987). These differences may correlateselectivity is found and head movements are readily with the different eye movements elicited in the superiorelicited. Such a difference is expected, of course, given versus inferior portions of the frontal eye fields (areas 8that cats rarely move their eyes without a concomitant and 45, respectively). Whereas sa'cades elicited from thehead movement (Guitton et id. 1984) whereas primates superior portion are large in amplitude (extending wellonly make combined eye-head movements when the beyond the typical distance traversed during visualdistance to the target exceeds the 15 degree radius of the search), the much smaller amplitude of inferior saccadesvisual search field (Bahill et al. 1975). indicates that they are instrumental in scanning closer to

It is intriguing to note in connection with the proposed the fovea (Bruce 1988). Furthermore, eyc movementscollieular link with far vision that neglect of the UVF has elicited from the superior and inferior fields may bebeen shown to accompany collicular damage in many relatively biased toward the LVF and UVF, respectivelyspecies (see Sprague et al. 1973). Some researchers have (Bender 1980, Wagman 1964), despite the presence ofargued that damage to the colliculus itself is not crucial for multiple oculomotor maps in each region (Bruce et al.the production of this syndrome, but that the critical 1985).locus may be the pretectum (Pasik et al. 1969) or intertec- It has yet to be firmly established whether the UVFtal commissures (Matelli et al. 1983). UVF receptive field neglect noted by Goldmnan-Rakic (1987, Figure 3) arisesbiases do, in fact, exist in the superior colliculi of many from a more ventral location in prefrontal cortex thanmammals (Drager & Hubel 1976, Sprague et al. 1973, does LVF neglect, or whether it corresponds to the locusFigure 9), just as stimulation of thecollicilus (alongwitha of the IT projection area. Nor is it clear whether thehost of other midbrain structures) results in a prepon- prefrontal UVF neglect is related to the extrapersonalderance of upward, divergent eye movements (Sprague neglect noted by Rizzolatti et al. (1985), as these authorset al. 1973). Ironically, no clear vertical anisotropy has have not distinguished between the role of the superiorbeen reported in the primate colliculus, although pre- and inferior arcuate regions in attending to extrapersonalvious topographical mapping studies have used only anes- space. In producing the far visual neglect in theirthetized animals. monkeys, however, Rizzolatti et al. (1985) appear to have

Despite the negative topographical results in primates, damaged primarily the ventral arcuate region. Perhapsit is tempting to speculate that the role of the primate this is also why Latto and Cowey (1971) observed acolliculus in far vision is somehow linked to its involve- deviation of the eyes into the ipsilateral LVF followinginent in a midbrain system oriented toward the UVF. In unilateral damage to the frontal eye fields.fact, an excess of convergence accompanies the UVF As knowledge concerning the functional specializationsneglect produced by pretectia lesions in monkeys (Pasik of prefrontal cortex continues to grow, the above issueset al. 1969), in striking opposition to the impaired con- will undoubtedly be settled. For now, it can at least bevergence and LVF neglect produced by parietal lesions hyputhesized that the ventral arcuate and midbrain UVF(see section 3.2.2). The greater convergence associated attentional systems work together in directing attentionwith UVF neglect also parallels the increased con- to the realm of extrapersonal space, paralleling theirvergence produced by descent of the eyes in humans (see critical combined role in the initiation of voluntary sac-section 1.1). cadic eye movements (Schiller 1986).

4.2.3. The preirontai visual areas. it is now believed thatthe dorsal-ventral division of the primate visual pathways 5. Conclusionscontinues into the frontal lobes, with the parietal andtemporal visual areas mamntainint xtensive and re- The foregoing theoretical review of the specializations ofciprocal connections with the sur .r and inferior frontal the LVF and UVF and their respective neural systemsregions, respectively (Bruce If , Goldman-Rakic 1987). indicates that both were greatly enhanced by the in-As with the temporal visual areas, the frontal lobe has creased segregation of near and far visual space thatundergone a tremendous expansion im primates (Gold- occurred during primate evolution. The LVF has beenman-Rakic 1987), but this expansion may not merely shown to be more important in the perceptual processesr..iect an enhanced emph.sis on extrapersonal space required of visuomotor coordination in peripersonal(which would have led piimarily to an expansion of the space (largely performed by the dorsal pathways of the


Commentary/Previc" Visual field specialization

primate visual system), whereas the UVF has been linked representation of motion-in-depth perception. Although UVF-with the visual search and recognition mechanisms di- LVF differences were not mentioned by these authors, it stillrected toward extrapersonal space (primarily controlled appears that the LVF may predominate in stercomotion detec-by the ventral system). In contrast to previous theories, tion for most subjects (D. Regan, personal communication),this one has adopted an ecological perspective in relating particularly those who possess no major stereomotion scotomas.neural functioning to the perceptual processes required 5. It should be noted that Richards and Lieberman's finding

was challenged by Bradshaw, Frisby et al. (1987), who found nofor successfully operating in tihe greatly expanded three- near-far asymmetry in the ability of their subjects to detectdimensional visual world of the primate. It has also structure-from-motion. Two factors may underlie the discrep-emphasized the role of experiential rather than genetic ancy between the two results. First, the tasks in the respectivefactors in the actual shaping of the higher visual areas of studies were different (volume estimation vs. shape discrimina-the primate brain. tion), with the shape discrimination task in Bradshaw et al.'s

The comprehensive theoretical perspective put forth study yielding extremely high overad correct detections (90%),in this paper offers the possibility of unifying a wealth of thereby raising the possibility that a ceiling effect was intro-diverse and previously unexplained findings regarding duced. Second, the stimuli in Bradshaw et al.'s study werethe nature and origins of human visual perception. Al- presented in the UVF only, a potentially serious flaw in view ofmost all of these findings have been, or could be, the the purported UVF bias against crossed disparities.

6. I personally have failed to observe an elimination of thesubject of individual review and debate. Regrettably, it corridor and Ponzo illusions at equiluminance in my ownmust be conceded that many important points of discus- laboratory.sion were necessarily glossed over in the attempt to 7. Since the monkey's head was restrained in Andersen etpresent as comprehensive a theory as possible. There is al.'s study, it is possible that their "head-centered" cells actuallyundoubtedly danger in such a decision, which Martin signal spatial location in a body-center d frame of reference,(1988) has described as going "a bridge too far." Indeed, which would arguably be more useful ir &nitoring the po. 4tionMartin's argument is supported by the revelation that a of the limbs during reaching. Since Gazzaniga and Ladavaslargely neglected set of findings concerning differences (1987) demonstrated that parietal nelect is not framed in abetween the LVF and UVF may provide an important head-centered coordinate system, a boly-centered explanationclue in piecing together the origins of the primate visual seems more plausible. (See also DiZio & Lackner 1989)

The intent of this theory, however, is not to 8. Gazzaniga and Ladavas (1987) actually suggest that par-system. t ietal neglect may be framed in gravitational coordinates, but theconstrain the interpretation of existing data, but to ex- corporeal and gravitational axes were aligned in their experi-pand the scope of future research. Its essential message to ments. It is difficult to understand why visual control of thefuture visual research is that the primate visual system motor system- which is organized corporeally (i.e., the left andcan only be understood in relation to a richly three- right hemispheres control the right and left sides of the body,dimensional visual world and the behavioral interactions respectively) - would be framed in nonegocentric gravitationalengendered therein, coordinates.

9. Critehley (1953) noted the much greater awareness ofvisual field defects after anterior (e.g., optic tract) as opposed to

ACKNOWLEDGMENTS cortical damage. Since visual attention operates even at theI would like to acknowledge the contributions of: (a) those level of primary visual cortex (Ilaenny & Schiller 1988), it mayindividuals who reviewed early drafts of this manuscript (Pres- be assumed that all posterior cortical damage is accompanied byton Garraghty, Fran Greene, John Maunsell, Lawrence some degree of attentional impairment.Tychsen, and Denise Varner); (b) the BBS reviewers whocontributed important suggestions and criticisms; and (c) themany other researchers who provided preprints and correspon-dence during the course of this endeavor. Also, thanks are givento the many technical contributions provided by personnel atthe USAF School of Aerospace Medicine, especially those of' Open Peer CommentaryJoseph Franzello (computer literature searches), RaymondBlancarte (illustrations), and Janet Trueblood (technical edit-ing). The writing of this manuscript was supported in part by agrant from the Air Force Office of Scientific Research. Commentaries submitted by the qualified professional readership of

this journal will be considered for publication in a later issue asContinuing Commentary on this article. Integrative overviews and

NOTES syntheses are especially encouraged.1. The projection of the visual fields to opposite hemispheres

is certainly consistent with the basic left-right division of thevisual world. It has also been proposed that a weaker subcorticaldorsal-ventral brain axis mirrors the up-down split of the visual Does visual-field specialization really have.rld ....... 9 but h implications or coordinated visuai-motor

superseded the primordial vertical one in determining thestructure of the primate higher visual cortical pathvays. behavior?

2. It is not clear why Yasuma et al. did not find A visual fieldeffect, but the relatively small stimulus ('34) compared to those Richard A. Abramsused in the other studies could have been a factor. Department of Psychology, Washington University, St. Louis, MO 63130

3. One potentially serious flaw, for example, involved the use Electronic mall: c39823 [email protected] extremely brief (30-mscc) stimulus presentations that proba- Given the imp rtanee of visual-imotur coordination, it is notbly prevented the more sluggish Lolor-oppnent liainls from Surprising that distinct br,un necdianisms might underlie thebeing activated (King-Smith & Carden 1)76). processing of visual information related to reaclng in periper-

4. At the time of this paper's acceptance, Hong and Regan sonal space dower visual field) and the visual information(1989) published additional findings concerning the % isual fitld needed tosearmh furandevaluatu obJLcts m xtraprsonalspatc

542 BEHAVIORAL AND BRAIN SCIENCES (1990) 13 3

C , nentary/Previc: Visual field specialization

(upper visual field). What is fascinatingabout Previc's analysis is researchers have characterized eye and limb movements asthat there exists considerable evidencc from a variety of do- consisting of preprogramnined bursts of activity in agonist mus-mains in support of this possibility. Indeed, his theory nte- cles with well-defined force-time relationships (Abrams et algrates a wide range of seemingly disparate neurophysiological 1989, Abrams et al. in press, Meyc. et al. 1988, Meyer et al.and psychological findings into a cohesive framework Nc. r- 1982, Schmidt ct al. 1979). According to these views, move-theless, despite all that it has to recommend it, the theory falls ments may be programmed on the basis of the perceivedsomewhat short because it is incomplete. In particular, it is not distance required of the movement. Other workers have er-at all clear what the implications of this perceptual specialization phasizd the position-seeking aspects of c: e and limb move-are for the control of actual reaching behaviors. ments. In sonic situations, commands to the muscles may

Previc's analysis rests on the premise that vsual specialization directly specify the final desired, d location of the movementarose, in part, from the evolution of visuomotor manipulatory (Abrams et al. in press, Mays & Sp rks 1980, Polit & Bizzi 1979)skills (i.e., "eye-hand coordination"). He shows, quite con-vinc- These alternative views of motor programming and productioningly, that the visual functions that are presumably needed to are not necessarily inconsistent with each other (Abrams ct al inmonitor such limb movements are best performed by parts of press), and may reflect the operation of specialized mechanismsthe visual system that work best in the lower visual field, and the sensitive to different types of visual information.visual functions needed to perform scanning, visual search, and Prey ic's theory presents a considerable challenge - raising asobject recognition are best performed by parts of the brain that r.iany new questions as it has answered. The task that remains isare linked to the upper visual field. While considerable evi- to extend the theory and link perceptual specialization with thedence is presented to support this perceptualspecialization, the control of motor behavior. The theory provides a good frame-link to actual motor behavior is somewhat tenuous. The problem work in which to accomplish that goal.can be stated as follows: The fact that the visual-field specializa-tion seems to be just what is needed to accomplish reaching andscanning in the world does not in itself constitute evidence thatthis is why the specialization exists, or that this is why the Seeing double: Dichotomizing the visualspecialization evolved as it did.

Unfortunately, little evidence is offered to indicate that the systemcontrol of reaching depends on visual-field specialization at all.For example, Previc argues that "an ability to attend specifically R. Martyn Bracewellto the contralateral, proximal LVF would allow the trajectory of Department of Brain & Cognitive Sciences, Massachusetts Institute ofthe reaching hand to be more accurately monitored" (sect. 2.5, Technology, Cambridge, MA 02139para. 2). This may indeed be true, but what is the evidence that Electronic mall: [email protected] is true? What is needed to complete the theory is an analysis of Previc has put forward a bold hypothesis, drawing on an im-the extent to which principles of motor control are consistent pressive array of experimental data. Many of his ideas arewith the properties of the mechanisms that Previc assumes are interesting. Unfortunately, I feel that the attempted synthesis isinvolved in that control. in places inconsistent and is ultimately unconvincing. In this

Although it is not completely understood precisely what kinds commentary I wish to deal with three topics. the question ofof visual analyses are actually performed on images of moving specialisation within the visual cortex, the use of the termslimbs, or how people might actually use visual information to "linear" and "nonlinear," and some of the evidence cited, bothprepare and produce limb movements, there has been consider- behavioural and neurophysiological.able work on these issues. Unfortunately this research is not Cortical maps. Previc surveys the parcellation of function inaddressed in the target article. In what follows, I outline a few the visual cortex and discusses some of the problems of the strictsuch issues that may inform, or be informed by Previc's theory. "isolationist" position (e.g., MT "does" motion, V4 "does"

Perhaps nne of the most relevant results is the finding that a colour) well. However, he goes on to ask "Why, for instance,person's ability to localize an object in space depends to a great should the processing of the features of an object be divorcedextent on the nature of the response the person must produce. from the processing of its relation to other objects?" (sect. 1.2Subjects often make large errors when providing perceptual para. 4). He appears to find "such divisions - . . contradicted byjudgments about the locations of objects, but they can accu- the unity of our phenomenological experience" (sect. 1.2, para.rately localize the object if a motor response is required 4). Why then do primates (and other mammals) have more than(Bridgeman et at. 1979, Hansen & Skavenski 1977, Honda 1985, one visual cortical field at all? It has been argued that functionalMatin et at. 1969). These results suggest that brain mechanisms iarcellation is an efficient way of organizing neuronal hardwareresponsible for the production of reaching movements have for the complex operations involved in visual information pro-access to information about the environment that is unavailable cessing (e.g., Barlow 1986). Moreover, such parcelation is notto the perceptual/cognitive syste.. This is essentially what necessarily at odds with the perceptual unity we experiencePrevic has proposed. special mechanisms evolved specifically (Cowey, 1981). [See also Ebbesson. "Evolution and Ontogenyfor the purpose of coordinating visual input with motor output. of Neural Circuits" BBS 7(3) 1984, Prechtl & Powley. "B-Someinsightintothenatureofthesediflerencesisprovidedbya afferents. A Fundamental Division of the Nervous Systemrecent study by Abrams & Landgraf (in press), who concluded Mediating Homostasis?" BBS 13:2 1990]that the apparent difference between perceptual and motor And yet Previc himself clearly believes in parcellation ofsystems might be explained in part by a difference in the type of function. The dorsal/ventral dichotomy is at the heart of hisvisual/spatial information used to produce the two types of paper.responses. Such differences may be the result of specialization Local/llnear versus global/nonlinear visual perception. Previcin the visual system that causes different brain structures to be defines "linear" and "nonlir -Ar" "in accordance with theirsensitive to different types of visual information, much as Previc neurophysiological usage (E ,roth-Cugell & Robson 1966)"suggests. (sect. 1.3, para. 2). The 1crsal pathway is supposed to be

An alternative approach resear hers have taken to learn about characterized by noninear mechanisms iand the ventral path-details of motor behavior has been to evaluate the moivements way by linear ones. It scems clar, however, that the majority ofthemselves to gain some insight into the information used to cortical visual neurons, especially be) ond the striate cortex, areproduce those movements. This approach has revealed the nonlinear according to the criteria of Enrth-Cugell and Rob-operation of two very different types of control processes that ion. In what sense arc "fa,,c selctivc" ncuruns in the temporalunderlie the production of reaching movements. First, some lobes representative of a linear system?


Comnentary/Previc: Visual field specialization

Previc equates local with linear processing and global with for peripersonal spatial representation (since, as Previc pointsnonlinear processing. It is not clear to me how he defines "local" out, saccades are frequently directed to targets in extrapersoaland "global." The Barlow & Levick (1965) model of directional space).selectivity posits a nonlinear mechanism, at what one might well 3. Previc states that "tile poor spatial resolution of mostconsider a "local" level kthe receptive field of a single retinal posterior parietal neurons renders them incapable ofsignallingganglion cell). the precise location of objects in space" (sect. 3.2.1, para. 9).

Fig. 2a illustrates Previc's confusion over this issue or per- Computational studies have revealed that it is perfectly possiblehaps nt; confusion over his v.ew). The small E's are supposedly for a population of neurons with broad but overlapping tuning"perceived using contour-dependent local processing", where- curves to represent spatial location quitc precisely (Hinton et al.as the large S "requi-e[s] global perceptual processing" (sect. 1986). The notion of coarse coding is gaining increasing pr ni-1.3, para. 3). But th,. detection of an "E" - tile extraction o, a nonce in neuroscience (e.g., Sejnowski 1988). Also, it wouldfeature - is a nonlinear process. indeed be strange ifparietal neurons - at least in ensembles -

Two attentional systems? It is clear that there are differences were incapable of accurate spatial localisation, given tile clearbetwe n the upper and lower visual fields (UVF and LVF). deficits in such furctions that follow parietal lesions (CritchlcyHowever, I must confess that I do not find that the evidence 1953).marshalled compels me to accept Previc's ,ypothesis. For 4. Previc stresses the role of MT in visually-guided reachingexample, let us consider the suggestion that there are two behavior. He cites Paillard's (1982) suggestion that global mo-countervailing attentional systems, one mediating body-cen- tion analysis suppo, ts the ability of monkeys to guide the handtred, peripersonal attention (which favours the LVF) and an- from the LVF to an object in foveal vision. Local analysisother mediating retinotopic, extrapersonal attention which (presumably mediated by Previc's ventral system) then allowsfavours the UVF). Damage to the parietal lobe produces ne- manipulation of the object. And yet surely manipulation is parglect, which is more pronounced in the LVF and the periper- excellence a peripersonal activity?sonal space. This neglect "appears to be framed largely in terms 5. Kendrick & Baldwin (1987) reported that 40/561 neuronsof body-centered rather than rettnotopic coordinates" sect. 2.5, recorded in he temporal cortex of sheep respond preferentiallypara. 3). However, evidence from patients with neglect syn- to faces. This percentage is similar to those reported for monkeydrome and normal subjects suggests that attention may be tied temporal cortex (e.g., 8-9%, Perrett et al. 1985). Previc shouldto one of several frames (e.g., gravitational, retinotopic, head- perhaps be more cautious in claiming "unique functional spe-centered, see Jeannerod 1987). Moreover, the work of Bisiach cializations" (sect. 4.2.1, para. 4) for the primate ventraland colleagues suggests that parietal neglect may be manifest at pathway.a purely ideational level (Bisiach & Luzzatti 1978, Bisiach et al. Summary. I have raised these issues not as points to carp over,1979). but rather to illustrate the difficulties and dangers in making

Previc also contends that there is a UVF advantage in attend- such a sweeping hypothesis. Previc has attempted a bold syn-ing to retinotopic, extrapersonal space. He cites the work of thesis of many eperimental findings. It is pleasing to see anScarisbrick et al. (1987), which showed that most normals bisect ecological perspective being taken. Certainly, some of thea vertical line above the midpoint. Yet might one not interpret evidence may be taken to support the hypothesi:,, but I feel thatline bisection as an example of a task carried out in peri-, not much of it is equivocal. The proposed dichotomy in functionalextra-, personal space? specialization is not strongly supported.

Neurophysiologlcal evidence. In his discussion of the M and Ppathways in monkey visual cortex, Previc makes a number of ACKNOWLEDGMENTSequivocal claims and raises several points I wish to address: It is a pleasure to thank the Whitaker Htealth Sciences Fund for its

1. Previc states that "the ventral system . . . exhibits the support and Robert J. Snowden and Catherine Cooper for comments ongreater topographical nrecision iequired of far visual percep- an earlier draft of this commentary.tion" (sect. 3.1, para. 12). This does not appear to be true in thehigher ventral areas such as V4 and the inferior temporal cortex,where visual receptive fields are large and topography is disor-derly (see van Essen 1985).

2. Work from this laboratory on the posterior parietal cortex The benefits and constraints of visualhas not shown that "receptive fields remain stationary relative to processing dichotomiesthe animal's head" (sect. 3.2.1, para. 2). Anderson et al. (1985) infact demonstrated that receptive fields of area 7a cells remains Julie R. Brannanretinotopic, although the strength of responses depends on the Departments of Neurobiology and Neurology, Mount Sinai School oforbital position of the eye. It is on the population level that such Medicine, New York, NY 10029cells may code space in head-centrei coordinates. We (An- Electronic mall: [email protected] et al., in press) have recently confirmed and extended Visual science has a long history of creating processing di-this result to another field of the posterior parietal cortex, the chotomies (X/Y, sustained/transient, "what"/"where," par-lateral intraparietal area (LIP). vo/magno) which, although ultimately not truly independent,

Previe suggests, reasonably enough, that a head-,.cntred stimulate a great deal of productive research. Previc's up-representation would be "especially useful in visuomotor coor- per/lower dichotomy is equally problematic, but may be equallydination" but he goes on to state that this "indicates an emphasis fruitful in providing an alternate framework for directing re-on near vision" (sect. 3.2.1, para. 2). We have shown that search. Iwillpomntoutsomepotentialdifficultmsforthisnewestalthough presaccadic cells are rare in area 7a, they are common dichotomy, however.in area LIP, and they are active for saccades in the dark Do the data suggest any real division? Previc notes (sect. 1.2)(Bracewell et al. 1989, Gnadt & Andersen 1988). Moreover, that previous dichotomies are not absolute and are subject topresaccadic activity is modulated by eye position in a fashion many counterexamples. Unfortunately, this also applies to thesimilar to that of the visual activity of area 7a cells (Andersen et upper/lower dichotomy. As Previc points out, the functionalal., in press). It is possible that such eye position modulation of specializations of upper and l,,wcr lhcmifields overlap more tianpresaccadic activity underlies a head-centred representation for they differ. In temporal and spatial resolution, for example, thesaccades (see Robinson 1975). Thus the presence of a head- magno/parvo dichotomy shows a clearer division than the up-centred representatici per se should not be taken as evidence per/lower one does. More troubling, in section 1.3 Previc states


Commentary/Previc: Visual field specialization

that short-range motion is a local process. However, even in Ups and downs of the visual field:short-range motion it is often necessary to solve the correspon- Manipulation and locomotiondence problem (Chang & Julesz 1984), which requires globalprocessing (see sect. 3.2.2). Perceptual phenomena are often Bruno G. Breitmeyerdifficult to dichotomize along a local/global axis. A further Department of Psychology, University of Houston, Houston, 7X 772045341example is that contrast thresholds are determined by local Electronic mall: psycm9uhupvm.bHtnetspatial inhomogeneities (Wilson & Giese 1977), yet they alsodepend on a global process of spatial probability summation The hypothesis that the lower and upper visual fields in humans(Robson & Graham 1978). Finally, in section 2.5 Previc suggests are functionally specialized for near and far vision, respectively,that retinotopic attention associated with visual search is biased is attractive and allows a conceptual integration of a wide rangetoward the upper visual field. But in a recent visual search task, of diverse findings. However, as with unitary explanations ofKrose and Julesz (1988; 1989) found that performance was better most complex behaviors, Previc's version of the hypothesisalong the horizontal dimension, diminishing when targets were either leaves out some cogent findings or appropriates findingsin either the upper or lower visual fields. These counterexam- which could be explained equally well by alternate schemes.pies illustrate that the tipper/lower and local/global dichotomies For example, in his discussion of the relationship between visualare less than distinct. search and extrapersonal space, Previc correctly states that

Can this theory explain the reeding proceu? The reading visual search in humans typically begins in the upper visual fieldprocess, taking place in near, peripersonal space, is somewhat (and preferably from left to right). However, this alone cannotdifficult to reconcile with Previc's near/far dichotomy. Althongh be takc , as evidence supporting the author's thesis of functionalreading requires some global, transient activity (which Previc specializat;on in the tipper hemifield for far vision and visualties to near vision), it is primarily local. In particular, reading search. After all, in subjects living in literate cultures suchrate increases with field size only up to 4 degrees, and is limited scanning strategies could be acquired from extensive exposureby a window only below 4 characters (Legge et al. 1985). Other to structured textual material which, by convention, placesaspects of reading contradict perceptual processes which are spatial and temporal sequential constraints on how they scansupposed to take place in near vision, For example, attenuating visual displays. What is required here is a convincing argumenthigh spatial frequency content by blurring below a bandwidth of that such culturally biased scanning strategies are based on,2 cycles/character reduces reading rate (Legge et al. 1985); rather than the basis for, functional differences between upperPrevic suggests that low spatial frequencies dominate near and lower visual field scanning performances. The fact thatvision. Reading rate falls off sharply below 10% contrast (Legge lower animals also tend to initiate scanning in the upper part ofet al. 1987), which is more consistent with parvo processing (tied the visual stimulus (Hebb 1949) would lend credence to Previc'sby Previc to far vision). Although Previc states (sect. 3.2.1) that hypothesis.near vision involves visual pursuit while far vision uses volun- Previc's unitary hypothesis further requires eliminating alter-tary eye movements, it is well known that voluntary saccades are nate explanations for functional differences between upper andan important part of the reading process (Rayner & McConkie lower hemifields. Previc's hypothesis of specialization of the1976; Rayner & Pollatsek 1981). Finally, Williams & Brannan lower field for function in near peripersonal space is based on(1986) found that children who were poor readers were much reasonable considerations of reaching and manipulative behav-slower than good readers in reaction time to the local informa- ior serving the largely frugivorous diet of primates such as thetion in Navon's (1977) global precedence task. Differences were monkey. Human evolution, however, did not get hung up innegligible in judging the global aspect. Together, these data trees or bushes. Besides evolving into gatherers of fruits, ber-suggest that local processing is crucial to the reading process - ries, nuts, and so forth, humans also evolved the capacity towhich takes place in near, peripersonal space. explore and move across terrain in order to hunt animals for food

A look at the reading process leads one to question whether and other needs. As with other predators, this relies ofcourse onany dichotomy is relevant when complex, real-world perceptual locomotion on a largely horizontal terrain. Except in earlyevents take place. Although local information is fundamental, infancy, locomotion by humans is usually performed bipedally,global information in the form of attentional activity which with upright posture. As noted by Breitmeyer et al. (1977), suchpreprocesses parafoveal information is also necessary for effi- locomotion or posture on a horizontal terrain already provides acient reading skill (Brannan & Williams 1987). basis for establishing biases for crossed and uncrossed dis-

Conclusions. Previc's upper/lower visual field dichotomy parities, that is, for near and far vision, in the lower and uppermay not explain visual processing any better than existing hemifields, respectively. Previe's thesis that preferences fordichotomies. However, it is always interesting to speculate on crossed and uncrossed disparities in these respective hemifieldsthe possible evolutionary benefit of any apparent parallel pro- are related to reaching and manipulating in near space and tocessing in the human visual system. Recently Brannan and exploration of far space is not contradicted by this explanation;Camp (1987) noted that many of the visual changes associated rather it is complemented.with the aging process (e.g., reduced temporal sensitivity) Moreover, searching, hunting, and other locomotor behaviorresult in a reliance on more sustained information. This is also rely heavily on postural control guided not only by theconsistent with certain cognitive capabilities, such as "wisdom," vestibular system but also by visual kinesthesis (Gibson 1958).that are associated with aged people. We suggested that it would To support his particular thesis, Previc would like to establish abe beneficial to the human species to have two types of pro- close link between near vision and the vestibular system. In-cessors: younger, quicker ones as well as older, slower, "wiser" deed, whilc thc vcstibular sstcm does pro-vidc input to nealones. Given older adults' tendency to develop presbyopia ("farvisual functions such as visual pursuit, it also provides a clearsightedness") perhaps this theory is not inconsistent with Pre- input to the control of locomotion through the environmentvies notion. (Dichgans & Brandt 1978). Visually guided locomotion depends

on centrifugal optical flow patterns (Regan & Beverly 1979),because humans as well as other ground dwelling creatures mostfrequently move forward rather than backward. Recently,Rauschecker et al. (1987) found a centrifugal organization ofdirectional preferences in the motion sensitive units of thelateral suprasylvian cortex of the cat. Moreover, these authorssuggest that homologous areas in the monkey ought to be found


Cominentary/Previc: Visual field specialization

in or near area MT. Such areas are part of the dorsal cortical quantified, however, the issue is necessarily a statistical onepathway whose function, according to Previc, is to serve near because each area represents both the UVF and the LVF evenvision as opposed to visually guided locomotion, though neither has a simple visuotopic map. Previe's target

The upshot seems to be that an inclusive explanatory scheme article may prompt such a quantitative analysis, but his reviewwhich incorporates not only Previc's hypothesis based on of the extant neurophysiological literature is somewhat slanted,near/far visual function but also one based on locomotion both in evaluating evidence regarding the UVF/LVF thesis andthrough the environment is more valuable to fully explain the also in discounting the central/peripheral distinctiveness of ITfunctional differences between the lower and upper hemifields and PP visual responses. With respect to the UVF/LVF hypoth-and the corresponding specializations of the dorsal and ventral esis for ITand PP, the evidence simply is not there. The figure ofcortical streams of visual processing. In fact, the presence of Robinson et al. (1978) that was reproduced (Fig. 8) seems tovisual structures supporting locomotion on a liorizo,tal surface show a preponderance of LVF representation for its sample ofmay be a general feature of most land-dwelling mammals. These -30 PP receptive fields (RFs), however, one tends not to noticestructures would already be in place prior to the evolution of some of the upper right quadrant RFs because their far bordersbipedalisin and the consequent freeing of the upper limbs for apparently exceeded the screen border and hence were notreaching and for manipulating objects. Hence, functional dif- drawn. The published figure showing the most PP data perti-ferences between the upper and lower hemifields supporting nent to the UVF/LVF issue may be Figure 9 of Mountcastle etnear/far vision as envisaged by Previc may in some measure al. Aj84. Of the 90 quantitatively analyzed "asymmetrical radi-already have been established on the basis of prior needs for al" types of PP cells comprising this figure the slight prepon-ground-based locomotion. derance, if any, is of cells with stronger responses from the

UVF. Similarly, the published figure showing the most IT RFsis probably Figure 4 of Desimone & Gross (1979), and the vastmajority of the -100 IT RFs shown therein are comprised of

Response field biases in parietal, temporal, comparable portions ofUVFand LVF. In summary, the hypoth-esis of an enhanced LVF representation in PP cortex and

and frontal lobe visual areas enhanced UVF representation in IT cortex is simply not sup-

ported by the neurophysiological literature.Charles J. Bruce and Martha G. MacAvoy Second, Previc reasons that because most visual RFs in bothSection of Neuroanatomy, Yale University School of Medicine, New Haven, IT and PP are "large," a foveal versus peripheral distinction isCT06510 absent. However, in reaching this conclusion he neglectsElectronic mall: [email protected] important aspects of IT and PP visual responses. Not only doA functional dichotomy of "two visual systems" was originally nearly all IT RFs include the center of gaze, but even IT cellsproposed by Trevarthen (1968) and Schneider (1969), and has with very large RFs invariably respond most intensely to fove-been refined over the past two decades. One visual system, ated visual stimuli. In striking contrast, many "light-sensitive"originally termed "ambient" by Trevarthien, has also been given PP neurons have "roveal sparing" or even foveal inhibition whileadjectives such as spatial and motion-sensitive, and is associated giving strong excitatory responses to movements in all parts ofwith processing coarse peripheral visual field information. The the far visual periphery (e.g., Mountcastle et al. 1984).4 Theseother system is specialized for pattern and form recognition, foveal/peripheral distinctions point to function and have beencolor discrimination, and foveal/central field processing of fine emphasized by many researchers. They are both stronger anddetail, and was termed "focal" by Trevarthen. In the primate, more abundant effects than a possible statistical preponderanceboth systems depend critically upon visual information relayed of LVF representation in PP cortex, or of UVF representation inthrough V1;1 beyond VI the primate's dorsally situated parietal IT cortex.lobe areas are generally most important for ambient/spatial Previc aligns the smooth pursuit and saccadic classes of eyevision whereas ventral/temporal lobe areas subserve focal/ pat- movement with the ambient/LVF and fucal/UVF groupings,tern vision. This dichotomy has been extended back towards the respectively. We disagree with associating the smooth pursuitvisual periphery. to the metabolic activity patterns in V1 and V2, (SP) type of eye movement with the ambient half of the visualto the parvocellular-magnocellular layer specializations of the dichotomy or with the LVF. (We think the near-far distinctiongeniculate, and to the X-Y distinction of retinal ganglion cells, does not apply either, as saccades and SP can both serve to trace

Here, Previc makes a strong case for differences in visual far and near stimuli.) Previc repeatedly associates SP with theprocessing abilities in the upper and lower visual hiemifields vestibular ocular reflex and the optokinetir reflex, however, SP(UVF & LVF). In general, the differences lie cites reflect a is not another mechanism to reduce overall retinal image slip-relative specialization of UVF for the focal (form-color-central- page. Instead, SP subserves the primate's desire to foveateparvocellular-X) set of functions and of LVF for the ambient particular, often small, moving visual stimuli. To support a SP-(spatial-mnotion-peripheral-mnagnocellular-Y) set. 2 " However, it LVF connection, he cites Tyclisen & Lisberger's (1986a) findingwould be a mistake to supplant the more traditional functional that in the step-ramp/Rashbass paiadigm human SP acceleratesdichotomy with these postulated specialiations of the upper faster in response to peripheral LVF motion than to peripheraland lower visual fields or the far/near terminology, in part UVF motion. However, this study also found that the retinalbecause these UVF/LVF specializations are relative or statis- eccentricity of stimulus motion was critical, with the fastesttical rather than functional nd in part because some of the accelerations to motion being in the central visual field, whichevidence ms disputable. Wt will review physmological vidence supports a focal vision interpretation of SP. Moreover.from visual association cortex, behavioral evidence concerning Lmbtcrger & Pavelko (1989) concluded that in the monkeyeye movements, and also new data regarding the repreentaton target motion in the superior and inferior visual hemifields isof eye movements in the frontal eye fields, equally effective for the initiation of pursuit." Thus, although

First, consider the status of the UVF/LVF thesis with respect there exists some LVF superiority for peripheral motion analy-to neurophysiology of the monkey's visual association cortex. 6is, presumably based in part on the large LVF representation inParietal lobe visual areas, especially the posteriorparietal lobule area MT, the SP eye movement system is concerned with(PP or 7a), are strongly aligned with the ambient vision group- continued foveatiun ofmvmg stiulih and does not necessarily

ing, and temporal lobe visual areas, particularly inferotempural parallel all aspects of tlt. bram's motion analysis systemcortex (IT) with the focal vision grouping. The relative impor- Finally, do the prir. ate's Frontal Eye Field. (FEF) belong intance of the UVF and LVF for IT and PP neurons has not been the focal/UVF/far visik n grouping? Previc places FEF in this



grouping primarily because of its traditional association with Twisting the world by 900saccadic eye movements; however, recent data indicate that SPeye movements are also represented in the monkey FEF: M. P. Brydena and Geoffrey UnderwoodbMicrostimulation ventral to the small saccade representation of "Depatment of Psychology, University of Waterloo, Waterloo, Ovtaro N2LFEF elicits smooth eye movements at thresholds as low as 10 3G1, Canada, and bDepartment of Psychology, University of Nottingham,1LA, and obtaining velocities as high as 25-500/sec (MacAvoy et Nottingham NG7 2RD, Englandal. 1988). Neurons in this area respond in association with Electronic mall: [email protected] andstimulus motion and SP eye movements, agreeing in preferred [email protected] with the elicited movements (Gottlieb et al. 1989); Previc has argued for functional distinctions between LVF anddeficits in SP eye movements follow removal ofthis ventral FEF UVF on the basis ofvery limited data. His Table 1 lists a numberregion (Keating et al. 1985, Lynch 1987; MacAvoy & Bruce of distinctions between upper and lower visual fields that are1989). The FEF can still be viewed as a primary cortical poorly supported, in particular those involving differences be-mechanism for foveation, and thus as an "output" module ofthe tween crossed and uncrossed disparities, saccadic eye move-focal visual system, ifone accepts our functional characterization ments, and global and local processing. We would agree thatof SP. these distinctions are premature and poorly grounded.

However, we hesitate to further categorize the FEF as an Previc's Table 1 shows some remarkable similarities to otherUVF structure because the FEF receives visual information, summaries of the differential functions of the two cerebraldynamic and static, from the entire visual field and represents hemispheres, as manifested by differences between the left anddownward, as well as upward, saccades and SP. Some UVF right visual fields (cf. Bryden 1982; Sergent 1983a; Underwooddominance in FEF is possible, and in this regard we note that 14 1976). In particular, the distinction between local and globalofthe 16 cells comprising the scatter plot ofFigure 4 ofBruce et processing (or analytic and holistic processing) has often beenal. 1985 have obliquely upward response fields and elicited applied to visual laterality studies, with the left hemisphere (ormovements. On the other hand, Figures 8 and 9 of the same right visual field) being seen as better at detecting local featurespaper show both upward and downward elicited saccadic move- and the right hemisphere (or left visual field) better at morements. As is the case for other visual association cortex, the issue holistic processing (Semmes 1968; Bradshaw & Sherlock 1982;of an asymmetrical UVF/LVF representation in FEF awaits Martin 1979). Likewise, Sargent (1983a; 1983b) has proposedquantitative neurophysiological analysis. that the left cerebral hemisphere is better at making use ofNOTES relatively high spatial frequencies, as demonstrated in her

1. Although striate cortex is extremely critical in the primate, it is also studies of laterality effects in face recognition. Previc has twisted1. lthug stiat crte i exremlycriica i th pimae, t s aso this by 90' , and given us some elegant evolutionary and physio-clear that the primate superior colliculus is the primary structure I bresponsible for the considerable visual abilities (collectively termed logical arguments for making functional distinctions between"blindsight") that survive lesions ofstriate cortex (Feinberg et al. 1978), the lower and upper visual fields. By his logic, one might expectespecially the remaining oculomotor abilities (Mohler & Wurtz 1977). to find many visual laterality effects to be replicated in the upper[See also Campion et al.: "Is Blindsight an Effect of Scattered Light, and lower visual fields.Spared Cortex, and Near-threshold Vision?" BBS 6(3) 1983.1 Moreover, The left-right differences are now clearly established. It is notthe visual activity in both the superior temporal polysensory cortex clear whether or not Previc would see upper-lower differences(Bruce et al. 1986) and in area MT (Rodman et al. 1986) that survives as independent of hemispheric asymmetries and additive tostriate lesions critically depends on the superior colliculus. Although them. Should one expect to see left VF global effects exagge-cortical cooling abolishes some collicular visual responses, as stated in rated in the lower visual field and attenuated in the upper visualthe target article, the retinotectal projection evidently maintains raed in the ow eri feld a te u d nte u isalenough visual activation to support visual responses elsewhere and field? If so, then the ordering ofeffects should move diagonallysome visual behavior, including visually guided saccades, from LL to UR.

2. Origins of the dorsal and ventral systems: Even though relative, For example, if the sequence of processing proceeds fromPrevic's hypothesis, together with the visual topography of striate global to local, then Previc should argue that scanning eyecortex, could still help explain the overall dorsal-ventral grouping of movements begin in the lower left and move to the upper right.extrastriatevisualspecializationsanterior toVl/V2in theprimatebrain: In fact, eye movement studies suggest that the pattern isThe UVF representation of VI (and of V2) lies in the ventral occipital generally orthogonal to this, proceeding from upper left to lowerlobe whereas the LVF representations are in the dorsal occipital lobe. right (Brandt 1945).Perhaps early in primate evolution the relatively greater importance of Despite the plethora of research on visual field effects, verythe LVF for spatial/motion/ambient functions dictated specializingdorsal cortex closer to the LVF representation of VI/V2 for these few researchers have paid much attention to possible top-functions, and, conversely, specializing ventral extrastriate cortex for bottom differences. It is true that one of the earliest visual fieldform/color/focal types of finctions where the UVF was of equal or studies, that of Mishkin and Forgays (1952) found not only agreater importance, right visual field superiority for the identification of words, but

3. We omitted the linear versus lonlinear distinction because it also a lower visual field superiority. If word recognition involveshardly applies beyond simple cells in the vistal cortex and probably no more analytic or local processes, this finding is counter to thathigher order visual area, regardless ofvisual field or function, is linear in predicted by Previc's model, for this sees the upper visual fieldthe original, operational sense applied to the retinal ganglion cells. As as being more "local." Unfortunately, very few researchersused in the target article, linear/nonlinear is suspect jargon and suggests using visual ficid paradigms have paid much attention to top-associations like "linear thinking."

4. The visual receptive fields of Area 7a (or P1') neurons do not bottom differences: Liederman et al. (1985) and Liederman and"remain stationary relative to the ammal's head rather than its fixation." Meehan (1986), for instance, have presented words at different

Even though area la responses are modulated by direction ot gaze, the corners ofan imaginary square, but they collapsed data acrossneurons still have retinotopic RFs and thus possess neither head- different conditions so that top-bottom differences cannot becentered nor body-centered coordinates. Unfortunately, Previc is not ascertained. Previc's target article certainly opens the door foralone in confusing neural-iietwork modeling hypotheses with piysiolog- the replication of a wide variety of visual laterality studies asical data. comparisons of UVF and LVF; whether or not such studies will

be fruitful remains a matter of conjecture.Previc also suggests that low spatial frequencies are more

frequently encountered in peripersonal space. It is true thatbringing an object closer reduces the spatial frequencies of themajor contours. However, in peripersoual space it is the details



that become relevant, while in extrapersonal space it is the more developing the ability to move our eyes rapidly to bring theglobal contours. We need to discriminate whether that onrush- power of the retina to bear upon the available light then whying object is friend or foe, but in peripersonal space we pick out should we need to develop new asymmetries in neurophysiolog-the fine details of the fabric or the specific characteristics of the ical processing? The very ability to search and inspect finelytext. In many ways, then, peripersonal space is local and analyt- would obviate the necessity for specialization.ic, while extrapersonal space is global and holistic. Previc has adopted a position free ofthe encumbrances of data

Although Previc also sees differences between UVF and LVF to support the distinction between UVF/LVF processing spe-in disparity detection, this has not proved to be a consistent cialization, and it is surprising to see such an argument prior tofinding. Manning et al. (1987), for instance, failed to find any seeing the collection of data. Previc seems to have lost hissignificant top-bottom differences for the detection of either overall, global view of visual field asymmetries, as well as hiscrossed or uncrossed disparities using dynamic random dot focal, local view of the importance of the relationship betweenstereograms. theory and data.

A further source of evidence for the functional specializationof the upper and lower visual fields involves a UVF advantage ACKNOWLEDGMENTfor saccadic movements, but there are problems with this view. This commentary was prepared while MPB held a Bilateral ExchangeMost important, the data do not give good support for such a Grant from the Natural Sciences and Engineering Research Council ofconclusion. The paper by Heywood and Churcher (1980), which Canada and The Royal Society of London, at the University ofis cited in support of a UVF saccadic movement advantage, Nottingham.reviews four earlier studies of up-down movements. Only two ofthese found an advantage for upward saccades, although theexperiment reported in the Heywood and Churcher paper alsofound that saccadic latencies to targets in the UVF were shorter Functional specializatIon in the visualthan those to LVF targets. Targets were single point sources of system: Retinotopic or body cente:ed?light which were displayed for 2 see while the subject main-tained fixation on a central point. When the target brightness Charles M. Butterwas incremented, the subject made a saccade to it. The latency Department of Psychology, University of Michigan, Ann Arbor, MIof this saccadic movement was 31 msee longer for LVF targets 48104-1687

than for UVF targets. The fact that subjects knew the location of Eletronic mail: [email protected] target before the onset of the signal which triggered the In the exposition and defense of his view concerning the naturemovement suggests that the difference is a function of saccadic of functional specialization in the dorsal and ventral corticalprogramming or control, rather than target detection. This visual pathways, Previc draws on a large number of findingsprovides some support for the UVF/LVF distinction, but from a wide range of fields, including neurophysiology, neuro-Heywood and Churcher reported no effects of target distance psychology, psychophysics, and the evolution and ontogeny ofand there is no evidence of differences between UVF and LVF the nervous system. Although, as lie admits, there are clearlymovements in accuracy of fixation following a saccadic problems in fitting all the findings he discusses into the moldmovement. required by his hypothesis, there is indeed an impressive sweep

Further evidence in support of a distinction comes from visual to his theorizing.search tasks in which eye movements are monitored. Hall My main reservation about accepting Previc's view is that he(1985), for example, found preferences for initial inspections in has not considered the possibility that near and far vision maythe UVF when subjects searched for a target picture in a set of not be related so much to visual field differences as to dif-three UVF and three LVF pictures. Not all the available mea- ferences in body-centered visual space. One reason for enter-sures support the asymmetry, however. Findlay and Harris taining this possibility is that when the eyes are directed to(1984) monitored the movement from a central fixation point to objects of interest, whether one is scanning them in far space orone of eight "clockface" locations and then to either of the two manipulating them in peripersonal space, it seems unlikely thatadjoining locations. They found no directional differences in in either of these situations the objects of interest are preferen-saccadic amplitude. Thus, the evidence is at best equivocal for tially located in upper or lower fields. Furthermore, in many ofany UVF/LVF difference in the control of saccadic eye the studies the author cites in which the eyes are centered,movements. retinotopic location is confounded with upper and lower visual

Rather than a general principle of "UVF advantages for space. On the other hand, if the eyes are free to move, as in thesaccadic eye movements" we need more analysis of the aspects examples given above and in some of the visual search studiesofsaccadic movements which show differences and those which Previc mentions, it is likely that any differences in performancedo not. There are differences in preferences for the direction of with stimuli located at different heights are attributable to theirsearch, an effect of saccadic latency (which is not unequivocal), location in body-centered space.and no effect of saccadic amplitude. Perhaps the preference for Two recent experiments Previc refers to that deal with al-UVF searches is simply an indication of learned probabilities of titudinal neglect (Butter et al. 1989, Rapcsak et al. 1988) makethe locations of objects for which we are likely to have to search. this point. In both studies, the patients, who had bilateralThe kinds of objects in the LVF are not those we have to search lesions of the parietal cortex (a cortical region that in Previc'sfor, and perhaps most of the things that require inspection of the 'view preferentially processes lower field stimuli), neglected theenvironment (principally objects in extrapersonal space) are lower halves of 'vertically oriented rods, when asked to bisectlikely to be in the UVF. This would not be to say that there is these rods visually, they pointed too lugh. lixation was notfunctional specialization of the two visual fields, so much as controlled, furthermore, as the patients were required to pointthere are different probabilities of occurrence of objects in our to the top and bottom of the rod before poimting to the center, itvisual field and that these probabilities will be learned over seems unlikely that differences in upper and lower visual fieldtime. Saccadic onset latencies, which also varied according to processing accounted for their abnormal bisections. These pa-some reports for upward and downward movements, would tients also pointed too high when bisecting rods using onlythen also be a function of practice. tactile/kinesthetic cues and, in one case when bisecting the

A more general problem with the argument concerning func- perceived distane between two sounds, one located above thetional specialization of saccadic movements is that at the stages head, the other below the head. Thus, their neglect is clearlyof evolution at which saccadic movements vvcrc de eloping, not retinUtopiadlly based, rather, it is more likely to be related toPrevic also sees UVF/LVF differences emerging. Now, if we are body-centered space.



Another procedure that can be used to disentangle the effects puts forth a near-jar dichotomy, which he relates to non-of retinotopic and body-centered location in the vertical dimen- linear/global and linear/local perceptual mechanisms, respee-sion is to require subjects to direct their gaze to points above and tively. The near-far dichotomy is also related to ("biased to-below the centered eye position while they perform tasks in ward") the lower and upper visual fields, respectively. Hence,which stimuli are presented in upper and lower visual fields the link is made between functional subdivisions of visual(perhaps using the chin or neck as the dividing line between the processing and specializations within the upper and lower visualtwo halves of space). A similar procedure, involving lateral fields.fixation of gaze, was used in a recent study to show that what A new improved theory of visual processing would be wel-appeared to be a hemianopia in a patient with a unilateral lesion comed by those who spend a substantial portion of the day tryingwas actually hemispatial (lateralized) inattention (Kooistra & to unravel how the visual system works. The key question is:Heilman 1989). Thus, the separate contribution of field and Does the treatment offered by Previc represent an improve-space factors needs to be investigated more thoroughly before ment over what is already available in the literature? An evenone can with confidence attribute the functional difference more basic consideration is: To what degree is his view sup-between dorsal and ventral visual pathways to one or the other ported by the evidence?of these factors. If there were linear/nonlinear differences in visual informa-

tion processing between upper and lower fields as Previc postu-lates, there should be differences in the response properties ofcells with receptive fields in the upper and lower visual fields.

Visual information In the upper and lower Specifically, cells with linear attributes should predominate in

visual fields may be processed differently, the upper visual field, whereas cells with nonlinear attributesshould predominate in the lower visual field. There is no reasonbut how and why remains to be established to believe that this is the case. Furthermore, the attempt by

Leo M. Chalupa and Cheryl A. White Previc to relate upper visual fields to the parvo stream and lowerof Psychology and the Neurobiology and Physiology Graduate fields to the magno stream is rather tenuous. In the primate

Department of clonde Davis, A 95616 lateral geniuclate nucleus, for instance, the upper and lowerGroups, University of Calif ornla, Davis, CA 95616 visual ficlds ar' represented about equally in the parvo and

Even casual inspection of a flat-mounted mammalian retina magno layers. While a case can be made that the "dorsal corticalreveals a nonuniform distribution of cells. The nonuniformity is system" processes different aspects of visual information inrelatively modest in rodents (animals with poorly developed comparison to what is processed by the "ventral cortical sys-focal vision), quite pronounced in cats (more visual animals), and tem," the relevance of this dichotomy to the upper and lowerstriking in primates (the most visual of all mammals). In addition visual fields is obscure. Even where there is evidence for someto the variations in overall cell density across a retina, there are disproportionate representation of the hemifields in visual cor-differences in the distribution patterns of specific cell types. tical areas, the relevance of this inequality for visual informationPerhaps the best known example is the pronounced dis- processing is far from straightforward. For instance, there are nosimilarity in the distribution of cones and rods in the pho- clear upper/lower visual field differences on several behavioraltoreceptor layer. tasks involving processing of velocity information, although this

The functional significance of such retinal regional variations attribute is thought to involve the MT visual area, where thehas been investigated by psychophysicists for more than a lower field has greater representation (see discussion incentury. Most of these studies have been concerned with Murasugi & Howard 1989).relating the sensory/perceptual capabilities of the visual system This is not to say that the processing of visual information fromto retinal eccentricity (i.e., central-to-peripheral regional varia- upper and lower visual fields is necessarily identical. Recenttions). The reason for such an emphasis seems obvious. The immunocytochemical studies provide evidence that there is anonuniformity in the density of retinal cells, particularly gan- difference in the organization of the upper and lower retina.glion cells, is most pronounced when the central region of the White et al. (1988a; 1988b; in press) have shown that there areretina is compared with the periphery. For instance, in the adult two types of somatostatin-immunoreactive neurons in the catcat the density of ganglion cells in the area centralis is about 80 retina: a small cell type, thought to be a wide-field amacrine cell,times greater than at the margins of the retina. In the human and a large cell type that resembles the alpha class of ganglionretina, the central-to-peripheral difference is even more strik- cells. Both the small and large cells are found preferentially ining; ganglion cells around the fovea are several hundredfold the inferior retina, with the small cells in highest concentrationmore dense than in the far periphery (Stone 1983). With the at the retinal margin (see Figure 1). Somatostatin-immunoreac-exception of the nasotemporal decussation pattern, other fea- tive processes, however, are distributed at all eccentricitiestures of retinal regional specialization have received relatively within the inner plexiform layer of the retina. A similar distribu-little attention. Thus, Previc's attempt to explain differences in tion of somatostatin-immunoreactive cells has also been ob-functional specialization between lower and upper visual fields served in humans (Sagar & Marshall 1988) and rabbits (Rickmanis certainly quite novel. & Brecha 1989; Sagar 1987). The functional significance of the

Another prevalent theme in the visual sciences has been the somatostatin-immunoreactive neurons is unknown. However,attempt to subdivide visual processing into separate compo- C. A. White and colleagues suggest that "this peptide could benents. These functional specializations have usually been relat- involved in raising the signal-to-noise ratio in neurons across theed to specific pathways (in modern parlance, "streams") within retinal surface in response to input fron. the upper visual field."the central nervous system. For exanmple, 1in often cited func- Ilis speculation is in line with the physiological observationtional subdivision from the 1960's stems from the "two visual (Zalutsky & Miller 1987; 1988) that infusion of somatostatin insystems hypothesis," which advocated that the retino-geniculo- the rabbit eyecup preparation increases visually evoked dis-cortical pathway dealt primarily with answering the question charge levels."what is the stimulus," whereas the retino-collicular pathway As more detailed information becomes available about thedealt with answering the question "where is the stimulus" functional and morphological organization of the mammalian(Schneider 1969). Perhaps the most influential scheme of the retina, the intriguing questions raised by Previc will be an-1980s has been that of Livingstone and Hubel (1988), linking swered. We suspect, however, that the answers will differ fromglobal visual processing to the magno system and stimulus those provided by Previc.identification to the parvo system. Previc finds the functionaldistinctions proposed by others unsatisfactory. lie accordingly



Sma1l1c*s, G activities in cortical brain areas which causes our ov\'rall visualpercept. What new dimension does the near-far hypothesis addto this statement?

Previc hypothesizes that the demands of near and far space fora primate are essentially compatible with the lower visual field

, N T (LVF) and upper visual field (UVF) respectively, and that thisT T .division is in turn supported by specialized magnocellular and

- >).W~, ', .% .; .parvocellular projections. Perhaps he pushes this hypothesis a" -,. • little too hard. Whereas a restriction of peripersonal space to

LVF is reasonable for the human, at least considering modernS*"lifestyle, it does seem surprising that the peripersonal space for

Ulm ," primates should be restricted to LVF- considering that thereare many species of frugivorous arboreal monkey whose prog-ress through the trees requires the hands to be elevated above

Largo cal, GCL the head most of the time, to pluck fruit as often from UVF asfrom LVF. The strength of Previc's argument depends to a large

.. , .extent on the interpretation of the role of the dorsal cortical•:. .."ri'.: •areas 7a and MT. This requires rejecting the originally postu-

"'' ' Nlated role of MT in the processing of optical flow information"" . • : ,'. during locomotion in favour of a role incorporating peripersonal

* i, " ' : .... "reaching. In arguing against the role of optical flow processing," ... " 'Previc suggests that the most rapid flow rates would be found in

. .A. ' .the extreme LVF periphery, a region not well represented in- :. .. ,. :.: -,: :'..:i:" - either MT or area 7a. However, in determining the path taken

by an animal during visually guided locomotion, which requiresattention to the interpretation of obstacles and terrain, pe-ripheral LVF vision would not be important. Second, the role of

Figure 1 (Chalupa and White). Maps of a pair of adult cat reaching during locomotion is important for actions such asFreas showing (hlcatndW of every potai r a ct leaping, incorporating high surround velocities as well as antag-retinas showing the location of every somatostatin-immunoreac- onistic centre-surround motion during the grasping of a target.tive small cell (above) and large cell (below) in the ganglion cell Third, with the neurons (especially those of area 7a) sensitive toi., cr (GCL). Some immunoreactive cells of the small type are change in a large number of different inputs - retinal,also found in the inner plexiform and inner nuclear layers, somatosensory, vestibular, proprioceptive, and attentional - itwhere they are distributed preferentially in the inferior retina is highly likely that the stimulus-space for the neurons of these(not shown). AC, area centralis; T, temporal; N, nasal. The areas has not been fully elaborated. Thus, while the argumentlarge, filled circle in each retina represents the optic disk. for reaching as a role for dorsal cortical areas is quite compelling,(Adapted from White et al., in press.) whether it is for the static, peripersonal activities or for reaching

during locomotion is less obvious.The argument for a "far" visual role for the ventral corticalThe ups and downs of visual fields areas also causes some difficulty, notably in the comparison of

David P. Crewther the properties of areas V3 and VP. While the argument forspecialization of the magno projection to the LVF (and hence

School of Optometry, University of New South Wales, Kensington, NSW, peripersonal) part of V3 was quite convincing, the evidenceAustralia presented for a "far" visual role of VP and indeed for the absence

Previc has made a brave attempt to combine anatomical, physio- of a parvo-generated function in V3 was far less so. Perhaps it islogical, behavioural, neuropsychological, ecological, and evolu- the term "far" vision that provides more problems than thetionary information to produce a cohesive explanation for the reported evidence. Although the neurons of the ventral system,functional specializations of the lower and upper visual fields of with their relatively higher proportion of parvo inputs, haveman in particular, and primates in general. I use the word brave disparities fairly narrowly distributed around the fixation planebecause when one gets down to the details of what is known of (cf. the dorsal system with a bias towards crossed disparities) thethe "actual" higher functions of visual areas of the brain, at best use of the parvo system during near fixation is basically ne-we rely on speculation. A less courageous approach would first glected. While Previc uses the distribution of disparities as anconsider objective fact (such as visual field maps, reaction times, argument for "far" vision, perhaps the term "fixation" visionreceptor densities, spatial and temporal properties of evoked would be more appropriate. A monkey grooming another (es-potential traces) separated from the more speculative imputed sentially a peripersonal task) would almost certainly call uponfunctional roles (such as associating the function of area 7a with the parvo system in the search for and identification of nits.reaching, or ascribing the role of the magnocellular pathway to There is a danger inherent in Previc's inferring the "function"near" vision and the parvo pathway to "far" vision), leaving us of the magno system from the specializations of the two regionswith a more restricted statement: Certain types of visual field- in which it is disproportionately represented." The associationrelated functional specializations are exhibited by primates but uf the inagnu system with the visual control of reaching andnot by most lower mammals. None of these specializations are, other peripersonal visual operations may neglect its role in aon most measures, highly evident in the retina, the LGN, the host of other visual attributes, especially those involved insuperior colliculus, or even the primary visual cortex V1 and the global visual perception. It also implies that the so-called highersecond cortical tier V2. The specializations are present because visual cortical areas, such as area 7a, MT, IT, V4 and so on, aresome of the higher cortical areas, espei ' in occipito-parietal the ultimate generators of visual function, causing the reader toand occipito-temporal cortex, are res ced to the upper or relegate regions such as primary visual cortex, VI, to the role oflower visual field and at the same time have a very different a neural relay station, just as in the past, researchers havedistribution of inputs in terms of LGN magnocellular or par- described the role of the lateral geniculate nucleus.vocellular cell types. The vertical differences in visual perfor- In summary, I believe that the case for the near-LVF-magno-mance are seen as biases because it is the combination of celluar link has been well argued and should be included in the



distinguishing properties of the dorsal cortical pathway, along thesis. However, the result appears to be a secondary conse-with global motion perception, in a description of the dor- quence of the latency difference referred to in the previoussal/ventral cortical dichotomy. The far-UVF-parvocellular link paragraph since the centre of gravity effect is less marked foris less strong; in particular, I find the idea of far vision less saccades with longer latency (Ottes et al. 1985; Findlay 1985).compelling than the concept of the use of the parvo system in Previc is surely right about the importance he assigns to theobject identification in the fixation plane (which of course local/global distinction. However, attempts to relate this dis-includes far vision). tinction to visual hemifields, whether up/down or left/right, fail

because adequate vision demands the integration of both formsof processing throughout the whole visual field.

Ecology and functional specialization: Thewhole is less than the sum of the parts Pigeons, primates, and division of labor in

the vertebrate visual systemJohn M. FindlayDepartment of Psychology, University of Durham, South Road, Durham M. A. Goodalea and J. A. GravesbDHI 3LE, England *Department of Psychology, University of Western Ontario, London,Electronic mall: [email protected] Ontario, Canada N6A 5C2 and bDepatment of Psychology, University ofPrevic has performed a useful service in collecting and assem- St. Andrews, St. Andrews, Fife, Scotlandbling a diverse literature relating to differences in psycho- Electronic mall: [email protected] and [email protected] and physiological measures between upper and lower Although one might wish to dispute Previc's characterization ofvisual fields. He has also presented an admirable survey ofsome the differences between the tipper and lower visual fields inof the different strands of parallel processing known to be primates, it is important to acknowledge that functional spe-present in the visual system. He uses this evidence to support a cialization of different regions of the retina occurs in a wideposition of "functional specialization," which suggests that visu- variety of nonmammalian vertebrates. Indeed, it would beal material is processed in different ways in the two hemifields. surprising if the same were not true of primates as well. WhileThis difference is alleged to have arisen as a consequence of the the visual system of birds is different from that of primates fromdifferent visual ecology the two fields encounter. The thrust of the retina on up (Donovan 1978; Hodos 1976), Previc's con-the target article is to suggest that these differences have centration on primates appears to have led him to ignore theoccurred during the course of primate evolution (see e.g., evidence for similar adaptations in other species, even though itsection 5). would strengthen ecological foundation of the distinction he is

The case for functional specialization is made by the ac- trying to establish. In the common pigeon (Columba livia), forcumulation of small pieces of evidence. The critical question example, there is plenty of evidence to suggest not only thatseems to be whether this evidence can really support the strong there are two functionally distinct regions of the visual field, butclaim for functional specialization as a phylogenetic process, that the nature of the specialization within these two regions isparticularly since there may be an alternative ontogenetic expla- remarkably similar to that described by Previc in the primatenation for many of the findings. The visual system is sensitive to (for review, see Goodale & Graves 1982 and Graves & Goodalevisual experience during development, as Previc notes in see- 1979).tion 4. If the visual system is genetically programmed to have The visual field of the pigeon, like that of many birds andcentral symmetry and isotropy, we might still expect tipper- nonprimate mammals, is largely monocular and panoramic.lower visual field differences to emerge because of the differen- Only a limited portion of the field immediately in front of andtial ontogenetic experience of the two fields. The difficulty with below the bill is binocular (Martinoya et al. 1981; Nye 1973).Previc's thesis is that it has not addressed the question of why The retina of the pigeon is quite different from that of thedifferences between the visual fields are generally so small and primate or other mammals in that many of the cones contain oilwhy, for many measures, isotropy and circular symmetry is in droplets that can be either clear or colored. The binocularfact so impressive (Rovamo & Virsu 1979). portion of the visual field corresponds to a region in the upper

Turning to detailed evidence, I feel qualified to make only a temporal quadrant of the retina, the so-called "red area" or "redfew comments relating to my area of specialization: saccadic eye field", where many of the cones contain large red oil droplets.movements and visual attention. As Previc notes (sect. 2.1), The remaining portion of the retina consists of the "yellow field"saccadiceye movements to visual targets in the lower visual field in which few or none of the cones contain the large red oilshow longer latencies than to targets elsewhere. This has been droplets characteristic of the red field (for review, see Emmer-found in several studies, although as tteywood and Churcher ton 1983a and Goodale & Graves 1982). The red area has a(1980) indicate, there is one counterexample (Miller 1969). A relatively high tectal magnification factor and a ganglion cellrelated finding is that when two targets are presented simul- density comparable to that of the central fovea and visual streaktaneouslyinupperandlowervisualfields, the tendency to move in the monocular yellow field (Clarke & Whitteridge 1976;the eyes to the upper target is very strong (Findlay 1980; Levy- Galifret 1968; Yazulla 1974), all of which suggest an area spe-Schoen 1969). No explanation for these differences is known, so cialized for acute vision comparable to the fovea itself (Clarke &Previe's suggestion that they arise because of the importance of Whitteridge 1976). The near point of accommodation for thisthe tipper visual field for visual scanning cannot be rejected. portion of the visual field, however, is much closer than that ofNonetheless, this link seems too remote and untestable to form the upper frontal and lateral fields (Nye 1973). Thus, just asa very satisfying explanation. Previc has described in the primate, there appear to be two

Previc places strong emphasis on the distinction between areas of specialization within the visual field of the pigeon; (1) alocal and global processing (sect. 1.3). Target-elicited saccadic lateral fovea within the large monocular field of each eye foreye movements also demonstrate an interesting form of global viewing distant objects, and (2) another area of acute binocularprocessing. When two targets are presented simultaneously in vision, corresponding to the red area of the retina, for viewingneighbouring positions in the visual field, the first saccade is stimuli located only a few centimeters away from the bill. Thisregularly directed at some "centre of gravity" position (Findlay near/far distinction is also supported by a variety of other1982). Unpublished experiments in our laboratory have shown psychophysical and neurophysologcal evidence, including dif-that this integrative effect is actually less marked in the lower ferences in the spectral sensitivity functions of the two fields (forvisual field, which, in a sense, runs directly counter to Previc's review, see Emmerton 1983b).


ComtnentarylPrevic: Visual field specialization

Moreover, like the primate, the pigeon appears to use the Attention to near and far space: The thirdpart of the visual field special-zed for near vision, in this case tile dichotomybinocular field, to control grasping movements directed atstimuli in peripersonal space (Goodale 1983a; 1983b). Even aKenneth M. Heilman, bDawn Bowers, and cPaul Shelton'though the pigeon uses a beak rather than fingers to pick up Department of Neurology, University of Flonda College of Medicine,objects, the problem it faces is much the same as that facing the Neurology Service, Veterans Affairs Medical Center, Gainesville, FLmonkey - to grasp objects in near space as efficiently as possible. 32602,08b and Section of Neurology, University of Manitoba, Winnipeg,A high-speed cinematic analysis of pecking in the pigeon Manitoba, Canadacshowed that in both key-pecking for food reward and normal Studies of spatial neglect have revealed that the brain is right-feeding, the decision to peck is made during a briefhead fixation left heinispatially organized, and this helinispatial organization isthat occurs some 80 mm from the surface on which the target is important in the control of behavior. For example, the rightlocated (Coodale 1983a, 1983b). Once the decision to peck is hemisphere appears to attend and l:repare for movementsmade, a second and even briefer fixation occurs at a distance of (intend) in and toward left contralateral egocentric hemispace.55 mm, which presumably allows the bird to calculate the size, Patients with right hemisphere lesions not only bisect linesdepth, and location of the target. This behavior is very stereo- toward the right but their bisection error is also more severetyped and during both fixations the target falls within the when a line is placed in left hemispace than when it is placed inbinocular portion of the field corresponding to the red area. right hemispace (Heilman & Valenstein 1979). This right-leftThus, while the actual solution to the problem is different, it dichotomy may be only one of three attention-intention spatialwould appear that in both the primate and the pigeon particular dichotomies, however. Not only horizontal but vertical neglectareas of the visual field are specialized for the control of grasp- has been reported. For example Rapscak et al. (1988) reported aing. In the case of the monkey (and man), the control ofgrasping patient with biparietal lesions who neglected the lov. er half ofinvolves on-line modulation of the reach trajectory on the basis vertically presented lines, and Shelton et al. (1990) reported a(in part) of visual information about the position of the moving patient who had bilateral inferior temporal lobe lesions andlimb (and sometimes the target itself) that is largely provided by neglected the upper part ofvertical lines. Brain (1941) posited athe lower visual field. As Previc points out, that information will third dichotomy when lie suggested that the region in extraper-consist of"optically-degraded and diplopic images" (particularly sonal space within grasping distance may be of special signifi-if the subject is foveating a more distant aspect of the environ- cance, in this target article Previc proposes that the visualment while making the grasping movement). As a consequence, system may be organized so that the lower visual field islie argues, the primate brain has evolved mechanisms to handle specialized to process near stimuli and the tipper visual field isthese stimuli - mechanisms that presumably involve the mag- specialized to process far stimuli. We have made some observa-nocellular pathway - that use specialized "nonlincar/global tions and reported a patient who provides partial support for aprocessing." Pigeons, whose pecking is largely "visually bal- down-near, up-far dichotomy (Shelton et al., 1990).listic," have solved the problem a somewhat different way by The patient is a 66-year-old, right-handed man who devel-having an area of high visual acuity, the red area, which can oped a bilateral inferior temporal lobe infarction that was proba-provide detailed binocular information about the nature and bly secondary to etobolic disease. His clinical picture is complexlocation of the target to be pecked, while at the same time (for details of his case and experimental procedures used toanother two foveae, one in the monocular field of each eye, are study him one should refer to our original report). Whenavailable for -viewing distant stimuli ofpotential interest, such as initially examined he showed a bilateral tipper hemivisual fieldpredators. defect with a preserved ability to detect and localize light in the

It is interesting to note that when a pigeon is flying, its head is upper fields. This upper field defect eventually improved. Heheld so that the bill is oriented well below the horizon with an also had a visual agnosia and, as briefly mentioned, verticaleye-center to bill-tip angle ofaround 390 (Erichsen et al. 1989). neglect. For example, when asked to bisect a vertically orientedThis posture insures that the ribbon-like visual streak that line lie set his mark below the actual midline. He also demon-extends from the lateral visual fields to the upper frontal visual strated upper vertical neglect on cancellation and drawing tasks.field (as marked by relatively high ganglion cell density) is In addition to testing line bisection in the vertical position (e.g.parallel to the horizon, which, together with the maintenance of intersection of the midsaggital and frontal planes), we also testeda constant orientation of the semicircular canals, may be ain for radial neglect by having the patient bisect lines presented atimportant requirement for accurate v isual control of flying. the intersection of the transverse and midsagittal planes. Radial

Thus, the pigeon, like Previc's monkey, has specialized visual lines were presented to this patient and to controls in threestructures for the control ofdiffereiit behavioral functions, some locations, with the line adjacent to tile body surface (near), withof which require processing of visual stimuli in near space and the line approximately 30 cia from the surface of the bodyothers which require processing of visual stimuli in far space. (middle), and with the line midpoint 60 cin from the body (far).The more we learn about the organization of vertebrate visual The patient consistently misbisected radial lines towards hissystems, the more evident it becomes that the visual system is body at all three distances. However, performance in far spaceorganized into a number of relatively independent visuomotor was worse than it was in near space. In far space lie misbisected"modules", each ofwliich has a special role to play in the visually radial lines by a mean of 7.10 cm and in near space lie erred by acontrol of behavior (Goodale 1983c, Goodale 1988). Previc's mean of 2.34 cm. His line bisection errois were greater than 4thesis, right or wrong, is a commendable attempt to explore this standard deviations from the mean of the normal controls.possibility in detail in the primate visual system, and, as such, is To learn whether this neglect of far space was modalitya welcome departure fromt the tledenity of iMany visuai sLIk- specific we also tested our patient and controls with a tactiletists to remain fixated on monolithic accounts of visual radial line bisection task where the blindfolded subjects ex-processing. plored the entire line and attempted to bisect the line. Again the

patient misbisected the line toward his body.There are at least two mechanisns that caii induce a systemat-

ic error in line bisection tasks: inattention or a directionalhypokinesia (hypometria) (Ileilnan et al. 1985). Extinction tosimultaneous stimuli or a failure to detect stimuli cannot beattributed to a directional hypokinesia or hypomictria and, in theabsence ofa primary sensory defect, is thought to represent anattentional deficit. This patient demonstrated a vertical visual


CommentarylPrevic: Visual field specialization

extinction. He was able to detect finger movements in either flexible orientors like monkeys. In any case, any correspon-upper or lower visual fields when presented alone but when dence of magno versus parvo with the LVF versus UVF is quiteupper and lower stimuli were presented simultaneously in the subordinate to magno correspondence with peripheral andsame coronal plane, there was a failure to detect movements of parvo with central vision. In contrast to the Y system, the greatthe upper fingers. However, on some occasions an upper field majority of cells in the parvo-related X system represent centralstimulus was no longer extinguished when it was presented 30 vision (Kaas 1989).cm closer to the subject's face than the simultaneous stimulus in Why then did the foveal object-related processing stream andthe lower quadrant. Taken together, these observations suggest the peripheral spatial-relational processing stream (Ungerleiderthat this patient's bilateral inferior occipital-ventral temporal & Mishkin 1982) become anatomically segregated in monkeys?lesion induced inattentiveness to far stimuli. Because this inat- We suggest that these are not, as Previc suggests, alternativetentiveness was both in the visual and tactile modality, the modes, but complementary ones. The reason it is not adaptivedefect was either polymodal or supermodal. Although one must for monkeys to specialize at a fixed differential visual field is whybe cautious about generalizing from one patient to a population the two visual systems are separate. Their very mobility compli-or deducing normal function of a brain area based on behavioral cates the task of placing perceived objects in a spatial frame-deficits associated with destruction of that brain area, our work, necessary as this is for effective foraging. According to theobservations of this patient suggest that the inferior oc- functional cerebral distance principle (FCDP; Kinsbourne &cipitotemporal region may be specialized to attend to upper and Hicks 1978) parallel disparate but complementary mental oper-far stimuli. ations are best served by keeping their respective neural sub-

Leinonen et al. (1979) and Leinonen and Nyman (1979) strates well separated in functional cerebral space - that is, inrecorded from a population of neurons in area 7b of the parietal the neural network. Dorsal/ventral separation of object andcortex of monkeys and demonstrated that their activity is en- spatial processing presumably minimizes crosstalk betweenhanced only by visual targets approaching the cutaneous recep- these concurrent complementary processing modes, which es-tive field or by stationary stimuli within 5-10 cm of it. Most of tablish distinctive objects within a spatial framework. It isthese light-sensitive cells did not respond at all if the target was precisely when simultaneous functioning is required that neu-further than one meter from the monkey. Area 7b projects to the ronal segregation is needed. When it is not, the same neuronalposterior bank of the arcuate sulcus; Rizzolatti et al. (1981a; population might be used at different times for different func-1981b) showed that arcuate neurons have properties analogous tions (Duffy 1984).to those in area 7b, the majority responding to visul stimuli That the near-far dichotomy is of secondary importance con-only if less than 10 cm away. Rizzolatti et al. (1981a; 1981b) pared to the peripheral-central dichotomy is also shown in termsintroduced the term peripersonal space to denote this region of of function by closer delineation of the fundamental neuronalextrapersonal space and showed subsequently that post-arcuate response properties which characterize the two streams. In theand area 7b ablations induced visual and tactile neglect of dorsal stream, the receptive fields of neurons in area MT arecontralateral pericutaneous stimuli (Rizzolatti et al. 1985). large, being an order of magnitude larger than striate cortical

Although patients with biparietal lesions have not been stud- receptive fields (Maunsell-& Van Essen 1983a; 1983b). In MSTied for the neglect of near peripersonal space and monkeys with the fields are still larger by another order of magnitude, ofteninferior temporal lesions have not been studied for the neglect of more than 100 degrees in diameter (Komatsu & Wurth 1988).far peripersonal space, our observations together with those of Similarly, large bilateral receptive fields are characteristic ofRizzolatti and his co-workers would suggest that there is a third neurons in parietal area 7a (Motter & Mounteastle 1981). Depthdichotomy. Even though the inferior temporal region is impor- preference is relative to the plane of the fixation point. Tuning totant for attending to far space, the parietal areas are important binocular disparity has been seen in MT (Maunsell & Van Essenfor attending to near space. 1983a; 1983b) and in MST (Komatsu et al. 1988). In MST the

NOTE response to stimuli depends on whether they arc nearer or1. Send correspondence to K. M. Heilman, University of Florida further than the fixation point, with no clear preference forCollege of Medicine, Department of Neurology U.236), Gainesville, FL "near" or "far" (Roy & Wurtz 1989). Area 7a neurons are also32610 depth sensitive. Two-thirds prefer close fixations and one-third

far fixations (Sakata et al. 1980).High order response properties of dorsal stream neurons are

specific for moving stimuli. The vast majority of MT neurons areThe role of dorsal/ventral processing movement-selective and strongly direction-sensitive (Dubnerdissociation In the economy of the primate & Zeki 1971). In MST specific cell populations respond to

complex, rotational, and radial movements (Tanaka & Saitobrain 1989) and are particularly sensitive to computer-generated sim-

ulptions of optic flow fields (Duffy & Wurtz 1989). Similarly, inarea 7a the direction-sensitive neurons demonstrate opponent'Behavioral Neurology Laboratory, Shriver Center, Waltham, MA 02254, vector organization suited for flow field analysis (Motter et al.bLaboratory of Sensorimotor Research, National Eye Institute, National 1987).Institute of Health, Bethesda, MD 20892 In the inferior temporal section of the ventral stream, inPrevic has added a welcome input emphasis to the functional awake monkeys during attentive fixation, the receptive fieldsdifferentiation of the dorsal and ventral extrastriate streams in are small, 5-10 degrees in diameter (Richmond et al. 1983). VAprimates. However, his suggebted causal sequence, from al- neurons are color, orientation, and shape semitive (Zeki 1973,titudinal visual field specialization to differentiation of related 1977). Their strikingly face-selective responses are invariantextrastriate processors, is perhaps the wrong way around. Dif- with changes in the size of the face, contrast reversal of theferential specialization of UVF and LVF implies a somewhat picture or reduction in the 3-D cues available (Rolls & Baylisfixed relationship between the position of the visual fields and 1986). They respond differentially when the face stimulus isextrapersonal spatial coordinates. The reverse is the case for the rotated from profile to full face (Perrett et al. 1985).agile canopy monkey, whose eyes, head, and body are con- Thus dorsal stream neurons are oriented toward analyzingstantly changing position relative to each other, and who is as global visual movement parameters. They may support posturaloften hanging upside down as right way up, It is species that can stability, guide locomotion, and encode the three dimensionalmove neither head nor eyes relative to the body (especially relationships between features of the environment. Most offishes) that show fixed differential visual field specialization, not these goals may be readily achieved through the analysis of optic


ConmentaryIPrevic: Visual field specialization

flov, fields (Gibson 1986), a task to which these neurons seem therefore be performed by some other system. Indeed, Un-ideally adapted. As such, their function is to some extent gerleider and Mishkin (1982) have argued that the parietal lobesconsistent with near rather 'han far vision, but it certainly are spccialized for spatial perception whereas temporal areas areinvolves the full field rather than just the lower peripheral field, specialized for object perception.

The response properties of the ventral stream neurons sug- Previc challenges this description of functional specializationgest that they extract information about locally cohesive le in the two visual streams on teleological grounds by asking whyments (Desimnone & Schein 1987). Facility for object recogni- the processing of object feaures should be separated from thetion, notably of faces, uses multisen',ory informatiou and processing of spatial relations. Rueckl et al. (1989), using smemory as integrated in polyser.sory and hnbic temporal c,.r computatiunalapproach, ha c prom ded a partial answer. Thesetices (Desiinne & Gross 1979). The final discriminations, for researchers show that a divided neural network (with one set ofexainiole, of faces mediated by the ventral system, occur in near hidden nodes processing shape information and another setrather than far vision, and at fixation iather than in the U'F. processing location information) is computationally more efil-

Why then the altitudinal anisotropies in man? According to cient than a single, undivided network of the same size inthe functional cerebral distance principle (FCDP), each visual encoding both shape and location information. Thus, it mayhalf field is to some extent under the influence of the processing simply be more efficient for the brain to allocate its resources somodes of the more adjacent (more "connected") segment of that different systems encode these different properties.extrastriate cortex. Just as activating the human left hemisphere To offer physiological evidence inconsistent with the Un-favors verbal processing in the right visual field and activating gerleider and Mishkin (1982) distinction, Previc also cites find-the right hemisphere favors spatial processing in the left visual ings that parietal lobe receptive fields overlap the foveal regionfield, so activating dorsal cortex by posing a spatial problem and IT receptive fields are very large, averaging 25 degrees inactivates the more connected LVF and vice versa. No additional diameter. At first glance these findings seem inconsistent withpost hoc adaptive rationalizations for altitudinal visual field the view that spatial perception (using much information fromanisotropies are needed. the visual periphery) occurs in the parictal lobes and object

perception (using information mostly from central vision) occursin IT. O'Reilly et al. (in press) used a computational approach todiscover nonobvious functional properties of the dorsal systemconsistent with these receptive field properties and with Un-

Why the computations must not be ignored gerleiderand Mishkin's (1982)characterization. They first noted

Chad J. Marsolek that parietal lobe receptive fields are also very large (Anderseniet al. 1985; Motter & Mounteastle 1981), which helps to explain

Department of Psychology, Harvard University, Cambridge, MA 02138 why these fields commonly overlap the fovea. To shed light onElectronic mall: [email protected] the different functional properties of these two areas, O'Reilly et

In an interesting description of the apparent segregation C.' al. (in press) next looked for other characteristic differences.primate visual processing in peripersonal and extrapersonal They noted that the distribution of receptive field peak locationsspace, Pre% ic hypothesizes that this processing segregation may was different in area 7a (they are more evenly distributed acrosshelp to explain why the dorsal and ventral visual streams are the field) than IT (they are almost always found on tne fovea,functionally separated. In a sense, Previc has added a new Gross et al. 1972, Motter & Mountcastle 1981). To discoverdichotomy to the pile, detailing evolutionary interpretations for possible effects of this variable, O'Reilly et al. (in press) variedthis new characterization and citing neurophysiological findings the number of off-center receptive field peaks that were "hard-consistent with it. The problem, as Previc is aware, is that the wired" into the receptive fields of input layer nodes of a three-previous characterizations of the functions of the two visual layer neural network. The coordinates of a dot in a matrix werestreams have also been supported with physiological evidence, specified far more easily when enough of these receptive fieldMore important, not one of these distinctions seems to be peaks were off-center. In fact, when all the regions of peakincompatible with the others, How do we correctly characterize response were located on the center, the network could notthese functional differences? I propose that a computational accomplish the mapping. It never computed the location of theapproach (Marr 1982) is necessary. inputl

The initial complication is the broad overlap among the These results indicate that the distribution of receptive fielddifferent functional descriptions of the dorsal/ventral dichoto- peaks is an important neural characteristic for encoding spatialmy. This overlap makes it hard to generate different predictions location, and clearly area 7a neurons are much more suited forto support one or another of the alternatives. Evidence support- this processing than IT neurons. Regarding this functioning,ing one scheme is likely to support the other. Previc cites the conclusion that since most individual parietal

A computational approach provides additional constraints and neurons have poor spatial resolution, they cannot process thesuggests nonobvious functional properties of brain systems. If precise spatial location of objects (Motter et al. 1987). However,we regard neurons in the parietal lobe (especially in area 7a) and when parietal neurons are looked at as a system performing aneurons in the inferior temporal lobe (IT) as performing differ- certain computaion, the system as a whole (having neuronsent computations (in terms of a system's neuronal input, an with overlapping receptive fields and well distributed peaksoperation performed on this input, and the output produced), a covering all of the visual field) can process precise spatialpicture different from Previc's emerges. locations, regardless of the spatial resolution of individual cells.

For example, an adequate description of the different com- Previc, in a broad attempt to unify many findings under oneputations being performed by these two systems must account conceptual roof, fails to take advaitage of tile Hoiiputdtiuilfor "stimulus equivalence across retinal translation," the ability approach and falls victim to interpreting individual neuronalof neuronal systems in IT (inferior temporal cortex) to identify an properties without consideration of the computations beingobject regardless of where its image strikes the retina (Gross & performed by larger systems of neurons. Ile qualifies this effortMishkin 1977). Given that objects can be identified through by concluding that his intent is not to constrain how physiologi-processing in the ventral system no matter where they appear in cal data are interpreted, but to call for the expansion of futurethe visual field, or even whether they are in peripersonal or research, primate vision must be understood in relation to ourextrapersonal space, IT neurons seem to be performing certain diverse, three-dimensional environment. Perhaps the mt ssagecomputations (involving object recognition) that are different actually communicated is the need for expanding future in-from those needed to encode the spatial location of objects. terpretations of brain system functioning to include the com-These other computations (involving spatial location) must putations being performed by these neuronal systems.



ACKNOWLEDGMENT visual input has to be attended during reaching and otherI wish to thank Stephen M. Kosslyn and Christopherj. Azorson for their peripersonal activities.valuable comments and assistance in preparing this commentary. The LVF's advantages (arising from greater sensitivity to

motion, luminance, and texture gradient) to the greater opticalmotion flow below the horizon during forward locomotionwould diminish with restriction of the peripheral LVF. Thus,according to Previc's hypothesis, visual space in a restrictedfield should become increasingly extrapersonal.

Peripheral lower visual fields: A neglected This is not what was found by Previc, and many perceptualfactor? misjudgments and performance difficulties were experienced in

Dolezal's observations; observers judged familiar objects asNaoyulci Osaka appearing smaller and nearer to themselves and their perceivedDepartment of Psychology, Faculty of Letters, Ryoto University, Kyoto 606, point of observation came close to the ground level because ofJapan the absence of their seen body image in the peripheral LVF. AElectronic mall: b52046"/ tansel.cc.utokyo./une.UTokyo, similar kind of experience was found among scuba divers [email protected] :[email protected] astronauts whenever peripheral LVF information was reduced

Previc's view on the functional specialization in the LVF and (Dolezal 1982).UVF in the human visual system appears quite interesting in This apparent shrinkage of peripersonal space during fieldconnection with magno/parvo (dorsal/ventral) dichotomy and restriction should be taken into consideration in connectionthe related neurobiological differences. The hypothesis that with the dynamic functional interdependency between near/near/peripersonal and far/extrapersonal visual space represent peripersonal and far/ex~rapersonal visual space and the distine-functionally distinct systems did not seem clear enough, tion between them.however. The last comment is related to manual RT (reaction time)

First, where is the functional boundary between the two differences between LVF and UVF. I agree that the RT to mosttypes of visual space? Regarding the near/far dichotomy, Previc stimuli is shorter in the LVF. My own results indicate that thestates that the boundary ofperipersonal and extrapersonal space RT to a flashed target is shorter by about 23 ins in the LVF thancould be specified as within and beyond the edge of an arm's in the TJVF along the vertical meridian; it is even some 10 insreach, respectively. However, as he pointed, the boundary is faster in the temporal hemifield (Osaka 1976; 1978). Thesenot functionally clear enough, even though easy enough to results would agree with Previc's assumption that th.; LVFdefine physically. Thus, he allows for an interdependency be- facilitates visuomotor coordination in peripersnnal space. Ittween two types of space by saying that the body-centered should also be noted that the temporal L.VF corresponds to thespatial coordinates used in peripersonal visuomotor activities quadrant where the hand appears in the LVF.could also be used beyond the arm's reach, just as local percep-tual analyses directed toward extrapersonal space can also beperformed on objects within reach. This view clearly weakensthe proposed dichotomy.

Second, what kind of factor influences the peripersonal Properties of neurons In the dors.,; ,isualspace? If, as Previc argues, the body-centered visual attention pathway of the monkeysystem for monitoring visuomotor activities in peripersonalsoce could be countervailed with a retinotopic attention system Ralph M. Siegelfor visual search and scanning in far/extrapersonal space, the Thomas .J. Watson Research Lzboratory, International Business Machines,LVF is therefore critically linked to the visual control of reach- Yorktown Heights, NY 10598ing in near/peripersonal space. Visually guided eye-hand coor- Electronic mall: [email protected] in peripersonal space is critical for reaching an object The target article presents a particularly thorough review oi'theand also critical for monitoring the reaching hand in the pe- major differences between the dorsal and ventral visual path-ripheral LVF. ways in the primate. There are still open questions, however,

The question is whether the observer's peripersonal space about the electrophysiological support for the functional segre-could be maintained as it was before when reaching in the gation into upper and lower visual fields that is proposed toperipheral LVF is prevented during visuomotor coordination coincide with near/far space, or nonlinear/linear processing.task. In his "tube study," in which a field of view was restricted This is because there are errors and controversy about some ofto the central 12 degrees by wearing long (33 cm) narrow tubes, the deductions in the target article made from the knownDolezal (1982) found that, in such a , estricted peripheral visual properties of cortical neurons in the darse1 -athway.field, observers tended to report obe.A shrinking and a reduc- I would first like to point out a misiitierpretation of experi-tion of the distance from themselves. Thub, the observer under- mental results of Andersen et al. (1985). In the discussion of thereached for objects and overreached the ends of the tubes. The dorsal visual system, Previc writes (sect. 3.2.1., para. L); "Per-observer experienced disorientation, loss of stability, and even haps even more intriguing, many posterior parietal neuronsdifficulty in walking. This evidence from field restriction sug- appear to code visual space in terms of head-ceqtered coordi-gests that peripheral LVF information is important. Peripheral nates (Andersen et al. 1985), ,uch that their receptive fieldsinformation is critical for monitoring the movement of the hand remain stationary relative to tLc animal's head rather than itsfrom LVIF, it also allows the hatnd to be guided fromi the fixaio."peripheral LVF to the fixed target, even though it cannot be Our experimental results showed that the receptive fieldviewed directly. Consequently, the body part image appearing remained stationary relative to retinal coordinates, not the fixedin the peripheral LVF has a role in keeping and stabilizing head coordinate system. A change in the animal's gaze angle lednear/peripersonal visual space. to a multiplicative modulation of the amplitude of the response.

Third, how can the central.'pripheral distinction be related We suggested that the representation of head coordinates wasto the peripersonal/extrapersonal dichotomy? Previc also notes distributed among a population of neurons which was consistentthat the central/peripheral differences cari be incorporated into with the neurological notion that tle inferior parietal lobule isthe peripersonal/extrapersonal (near/far) dichotomy, in that the site of the representation of"extrapersonal space" (Critchleymost far vision is limited to the central 30 degrees because of the 1953).poor spatial resolution of the peripheral retina, while peripheral Two different theoretical approaches (Siegel & Andersun


Commnentary/Previc: Visual field specialization

1987, Zipser & Andersen 1988) indicate that a population of subdivision LIP (Cnadt & Mays ", .). MSThas disparity tunedthese angle-of-gaze cells could be used to signal the location of cells (Roy & Wurtz 1989), but not enough data is available toan object in head-centered coordinates within 10 in spite of the determine whether thcre is a l.- ference for near or far.large receptive field size. Both these studies relied on the "Fifth, it is difficult from an ecological viewpoint to under-summing of the activity of these cells to achieve behaviorally stand why brain areas so obviously involved in reaching and eyereasonable precision. Thus Previc's suggestion that parietal movement control should perform an optical flow analysisneurons are "incapable of signaling the precise location in whose chief value would be to maintain postural control."space" is incorrect. The angle-of-gaze cells of parietal cortex The key point in this statement is the assumption that thecould be used to locate an ob,ect that is not an immediate "near inferior parietal lobule is involved only in peripersonal behav -space" (e.g. a monkey in a tree 10 meters away), which contra- iors. While it is true that motion flow fields can be used fordiets Previc's thesis. reaching and eye movement control, they are also useful for a

Previc also argues against "an involvement of area 7a and MT number of other visual-motor tasks (see Ullman 1979, Longuet-in the processing of optical flow information during locomotion liggins & Prazdny 1980). Some examples are locating thethrough the environment." (sect. 3.2.1., para. 8) His view is that boundaries of objects, determining the three-dimensionalthese regions are used for "reaching and other peripersonal structure of an object, locating oneself in a moving environ-behaviors, " and five "major observations" are made to support ment, as in moving through trees for arboreal primates, andthis view. All of these "observations" are, at the least, subject to indeed postural control. There is no a priori rea -in to rule outdispute. Previc's points kseet. 3.2.1., para. 8) alongwith relevant parietal involvement in these processes, particularly in consid-comments follow: ering some of the effects of parietal lesions (e.g. disorders of

"First, the most rapid flow rates . . . ai e found in the periph- movement, spatial perceptions, etc., in particular, see Critchleycry whereas most MT and 7a receptive field are located with- 1953, Chap. 5).in . . . the central 20' of the visual field." In summary, the dorsal visual pathways, which include MT

Area 7a neurons have quite large visual receptive fields, and 7a and a host of other regions (Maunsell & Van Essensometimes up to 400 in size (Andersen et al. 1985). The pe- 1983c), can indeed be used for visual flow field analyses as wellripheral MT fields can albo be large (Albright & Desiinone 1987, as for determining precise spatial position. I have attempted toMaunseil & Van Essen 1983a). The result that these receptive describe the controversies that exist for some of the argumentsfields overlap with the center of the visual field does not used to support Previc's contentions. Although a weakness inpreclude their use for optical flow analysis in the periphery or any one supporting proposition does not completely negate theacross the visual field. Indeed, a model has been proposed conclusion, the many ostensible shortcomings of the target(Siegel 1987, 1988) that uses the properties of MT neurons in a article suggests that we should exercise some caution beforeparallel processing scheme to extract flow field information. embracing the idea of an upper and lower visual field

"Second, optical flow patterns during locomotion are . . . cx- dichotomy.panding and the majority of area 7a neurons respond to motionaway from the animal." NOTE

The study cited (Steimnmetz et al. 1987) used unnatural stimuli 1. By which I assume the author means "center surround motion."

(small moving squares) to test for optical flow properties. Recentphysiological studies in the awake behaving monkey (Siegel1989) using more natural and controlled stimuli (random dot Different regions of space or differentfields) (Siegel & Andersen 1988) to test for optical flow selec- spaces alogether: What are thetivity have found an equal number of cells selective to bothexpanding and compressing stimuli. Furthermore, this study dorsal/ventral systems processing?suggests that there are highly nonlinear interactions betweensubregions of the receptive field, making it difficult to predict Gary W. Strong

the response to a full field motion stimuli from either local College of Information Studies, Drexel University, Philadelphia, PA 19104

patches of motion or small moving squares. Region MST, Electronic mall: [email protected] or [email protected]

pointedly missing in the discussion of the dorsal pathvay, also Previc's target article represents an impressive effort to inte-has neurons selective for real optical motion flow (Tanaka et al. grate previous work, but I remain of the opinion expressed in1986, Saito et al. 1986, Duffy & Wurtz 1989). Both the 7a and section 1.1 ofthe target article, that "the LVF-near and UVF-farMST neurons in the dorsal pathway can be used for motion flow links are far from absolute." Previc prematurely dismisses asfield analysis that can occur in both the upper and lower visual "not altogether satisfactory" Ungerleider and Mishkin's (1982)field as well as in near or far personal space. distinction between spatial/peripheral and object/central per-

"Third, opposite motionl is never produced by egomotion ception (section 1.2). However, Ungerleider and Mishkin'sthrough the environment, so the preference . . . for center s. placement of spatial/peripheral perception in the parietal, orsurround must be related to other factors." dorsal, pathway coincides with Kesner et al.'s (1989) identifica-

Center-surround motion cells were first proposed for a tion of the parietal cortex as involved in the processing ofnumber of etholugical and physiological reasons by Allman allocentric spatial information, that is, information "based on(1977). Center-surround motion can be obtained vhren an oh- memory for specific stimuli representing places or relationsserver moves through an environment, the analysis of such between places that are independent of one's body orientationsmotion is essential for localization in the environnent. Thus in space" (p. 956). To the other pathway Ungerleider & Mishkin"never" is clearly an overstatement, assigned the function of object/central perception. This is sim-

"Fourth, The preference of dorsal neurons for near disparity ilar to what Mack (1978) calls the "object-relative mode" ofand/or fixations . . . perception. This distinction being made, it appears that a more

Physiological data are not nearly so complete as to permit accurate identification of the functional difference between thesuch a blanket statement for the five or more dorsal visual areas dorsal and ventral ptthwvays may not be based so much on a(e.g. IT, MST, FST, LIP, 7a, VIP, etc.) The preference for partitioning of the regions of space as perceived by the subjectnear disparity is suggested only for area MT kMaunscll & Van (Previc's thesis) as on different spatial coordinate systems. TheEssen 1983b). Some data has been collected for a broadly basic question I wish to pose is whether the dorsal/ventraldefined area 7a (Sakata et al. 1980), but there appear to be distinction is based on a differentia! processing of LVF/UVFfurthe, complications in the interpretation of such studies in the "regions of visual space" as Preic suggests, or on the spatial



"types of information they process" (a view Previc dismisses in frontal cortex as well as other areas. This could even be a moresection 1.2). plausible characteristic structure for primate cortex than one

The impressive amount of data Previc reviews suggests, at tile based on the UVFILVF distinction since hand-eye coordinationvery least, a distinction between an object-based tracking sys- is a distinctive feature of primate behavior.tern for visuomotor coordination in peripersonal space (section2.7) and a retinotopic visual system. It is one thing to suggestthat these two systems must exist, another to say that theycorrespond to the LVF and the UVF, respectively. Consider, asexamples, the acts of tracking a distant motion, such as that ofa The primary visual system does not carerunning animal, and reading a book. The former involves both about Previc's near-far dichotomy. Why not?the UVF and far perception, in opposition to what Previc wouldsuggest, and tile latter involves both the LVF and retinotopic, Robert W. Williamssaccadic vision, also in opposition to Previc's thesis. Department of Anatomy and Neurobiology, University of Tennessee, School

Furthermore, ifparietal patients cannot solve the correspon- of Medicine, Memphis, TN 38163dence problem (as reported in section 3.2.1), then Previc must Electronic mail: ,williams@utmem3,bitnetexplain how this relates to or differs from the statement in Can a umajor division exist in the processing and representationsection 3.3.1 that "a major puqose of tie far attentional of upper and lower visual space without a corresponding divi-[temporal] system is to 'glue" features into intebrated vloles, so sion in the structure and function ofthe retina, lateral geniculateas to ensure that form. composed ofidetical features in differ- nucleus, and striate cortex? An, information theorist would sayent arrangements are not comifused. " I hold that botu the parietal yes k.Marr 1982). The program and the processor are separate,and the temporal system can play the role of "tag-asognment' and the same processor can run many different programs. But anand that there is no reason to believe that it can't be done by evolutionar biologist would say no. The program and thefoveal, saccadic vision (see Strong & Wltehead 1989). What processor coevolve, and if Previc's dichotomy is vital - asdiotinguishes one pathway from the other probably depends on defined by its effects on fitness - then in the course of sevceralwhether the so-called "glue" kor tag) is based ol spatial location million years of intense selection this dorsal-ventral rift should(dorsal/parietal) or object pattern kventral,'temporal). The fact have left marks along the entire visual system. In otherwords, athat tile hippocampus receives "heavy projections from the single processor cannot be optimized for two different and %erytemporal lobe" (Section 3.3.2) and does not itself have an complex tasks. "Previc takes this evolutioiiary-ecological ap-6.vious global spatial mapping (Eichenbauin et al. 1989) sup- proach and, following in the parallel processing tradition, looksports tile view that the temporal lobe is the site of integration for substrates of the near-far dichotomy in the retina, lateralbased on something other than space, such as "objecthood." genmculatc, and visual cortex. Here anatonists and physiologists

The reader may suppose that Pre% iL holds his particular thesis let him down. There is really % cry little c idence ofasy inmnetryin part as a way ofnalang sense ofvvhat lie calls the "teleological in the representations of upper and lovve; visual space, particu-challenge," which is, "Why . . .should the processing of the larly in primates.features of all object be divorced from the processing of its One counterargument is that structural modifications haverelation to other objects . . Such divisions arc contradicted by not kept pace with changes in the algorithm. This is umnlikely.the unity of our phenomenological experience..." (section Over a period of merely 25,000 years, the visual system of the1.2). Current research suggests, however, that there is no cat lineage has undergone wholesale structural change (Wil-reason to suppose that the "unity of experience" is related to the hams et al. 1989), and these changes are obvious in the retinalocality of representation in brain tissue. Aiple and Kruger and lateral geniculate nucleus. Furthermore, in several species,(1989), Eckhorn ct al. (1988), Gray et al. (1989) and Gray and upper and lower visual space is treated differently by theSinger (1989) show clearly that perceptual unity is tied to primary visual system. There can be no more dramatic examplesynchrony of activity rather than to locality in the cotex. Unity than the tropical fish Anableps (Walls 19,12) with two pupils perthrough synchronization completely changes the way one looks eye - one for vision in air and one for vision in water. Verticalat the brain, in that there is no longer any need to consider that a asymmetry has also been discovered in a few mammals, butrepresentation is in one neural site, or that one type of infornia- unfortunately for Previt, in all the wrong species (see accom-tion is processed in one area. Locality arguments now have to be panying commentary by Chalupa & White). For example,made on other grounds, such as the need for nonspecific peri- retinal ganghion cells are more heavily concentrated along thecolumnar inhibition that has b-en identified between cortical dorsal vertical meridian in herbivores, dogs, and bush babiesminicolumns, "tile most basic units" of the primate cortex (Hughes 1977, Stone 1983), and Hughes argues that in her-(Mounteastle 1979). bivores this vertical streak subserves the grazing field betweenThe processing distinction between the dorsal and ventral face and forefeet. But in monkeys and humans almost all we can

pathways is perhaps that of between-object relationslhps and say at preseit is that (1) rods are iiore closely packed in tilewithin-object relationships, respectively. Between-object rela- duisal hmiretina, forinig a rod hot spot about 5mm iabouv tiletionships are much more dynamic than within-object rela- fovea ,Wckler et al. in press), and that (2) somatostatin-inimu-tionships and probably rely on a different coding method in tile norcactme cells are located almost cxclus v ely in tile ventralbrain. Whereas within-object relationships could conceivably hemiretina (Sagar & Marshall 1988). Both of these features arebe coded by redundant populations of neurons (such as the face- probably related to differences in mean illumination of upperrecognizer populations of Rolls at al. 1989), between-object and lower fields, not to near and far vision.relationships require a more dynamc, cons truted representa- So why doesn't the primary visual system of primates earetion such as temporary synchrony among a population ofnonrc- about PreviC's near-far dichotomy? I have three suggestions.dundant columns. 1. Previc is exaggerating the segregation of near and far

In conclusion, I agree wvith Prev ic that "It must be conceded functions in lower and upper visual space. To some degree,that the near-far dichotomy falls short as a complete explanatory experiments vvith mn ertig prisms bear this out (Harris 1965).scheme" (section 1.2, para. 8). Furthermore, thereare probably As a leuristic, a little exaggeration Is useful, but lure Previc me ya large number of simultaneous processing pathways in the have gone too far. The near-far dichotomy may actually be abrain, some not tied to the ventral/dorsal distiction. Fur subtle gradient. This would be the simplest explanation for tileexample, Goldman-Rakic (1988) proposes the existence ofpar.d- abounce of near-far specialization in the primate primary visuallel hand/eye circuits, each involving parietal, temporal, and system.



2. Other more important constraints and processes dictate time it was thought that this phenomenon might be in some waythe design of the primate visual system. At most, Previc's related to ecological pressures resulting from the left to rightdichotomy is a relatively unimportant, although interesting, order of reading used' y several forms of alphabetic script. It haswrinkle. Since there is minimal evidence for near-far segrega- now been established, however, that right visual hemifieldtion at any level between the photoreceptor mosaic and area 18, superiority for word recognition is found even in languages thatwe are obliged to accept that other processes have had the are read from right to left (Carmon et al. 1976), and that thedominant role in the evolution and parcellation of the primate effect is less clearly shown by left-handed people than by right-visual system (Diamond & Hall 1969; Hassler 1966, Polyak handed ones (Bradshaw et al. 1981) even though any ecological1957). One of the most interesting recent ideas is that the pressures created by reading direction will be the same forevolution of stereopsis in primates was driven not by the selec- right- and left-handers. Thus it is clear that the primary determi-tive advantage ofdepth discrimination per se (there are plenty of nant of visual hemifield differences in word recognition ability isother cues) but by its utility in revealing well-camouflaged food cerebral asymmetry rather than ecological pressures per seand foes (Frisby 1980, p. 155). (though this is not to deny that these may exert some modifying

3. Previc's dichotomy is real and possibly important, but he is influence).looking for structural substrates and functional correlates in the Since information processing differences between the left andwrong parts of the brain. The visual system provides a wealth of right visual hemifields do not seem to be primarily determinedinformation to disparate systems in parietal, frontal, and tem- by ecological pressures, why should we think that matters willporal neocortex. Each of these external systems requires an beany different with respect to upper(UVF) versus lower (LVF)interface, and each interface requires custom wetware for filter- visual hemifield differences? Convincing arguments areing, mixing, and re-representing this hybrid information. These needed. Previc tries to provide two, one of which I find convinc-interfacing requirements have molded a specialized belt of ing, one unconvincing.paravisual areas that surrounds the relatively uniform visual The convincing argument is that because so much visuallycore. These are the regions in which the split in the visual guided reaching takes place in the LVF, it has become speciallys; stem, whatever it may mean, is most obvious. If there is a adapted for this purpose. Although the literature on uppercrisp dichotomy between near and far, as Previc argues, then its versus lower visual hemifield differences is not extensive, Pre-proximate causes will be found in rostral parts of neocortex and vie demonstrates that it is consistent with this idea.not in the particular design of the retina, geniculate, or visual Much less secure is Previc's attempt to argue that the UVF iscortex. Another way of saying 'his is that the near-far dichotomy specialised for object search and recognition in "far" (extraper-is not a property of the visual system at all, but simply an sonal) visual space. The ecological argument is unconvincing. Asoutgrowth of diverse requirements of nonvisual neocortex. I look around me now, I can see plenty of out-of-reach objects

that fall either above or below fixation, and once I have identi-fied an object of interest I refixate to bring it into central vision.Similarly, when I am moving araund, especially outside thehouse, new objects of interest are often hidden in small de-

Only half way up clivities or obscured behind other foreground objects. As theyare revealed, these will mostly be found initially in the LVF.

Andrew W. Young Previc in fact concedes this point, and admits that "the rela-Department of Psychology, University of Durham, Science Laboratories, tionship between far vision and the UVF is not nearly asDurham OHI 3LE, England exclusive, as both vertical hemifields represent the extraper-Electronlc mall: [email protected] sonal portion of visual space" (section 1.1, para. 5). Moreover,

In his insistence on the importance of an ecological perspective his review actually demonstrates that "virtually no UVF advan-

for understanding neural functioning, Previc is surely right. We tages exist in sensory processing per se" (section 1.1., para. 5).

should not consider functional specialisation without any refer- Instead, Previc arrives at the conclusion that there is a "link

ence to the demands of the world to which the functions have between the UVF and an extrapersonal attention system that

adapted. But the argument can also be misleading if taken to facilitates object search and recognition" (section 2.7, para. 2).extremes. Every aspect of cerebral organisation does not reflect Evidence for this link, however, is no more than suggestive at

this type of constraint, present.Consider, for instance, the extensively documented involve- Why is Previc so concerned to establish a specific role for the

ment of the left cerebral hemisphere in language abilities. Many UVF? Why isn't he simply happy with the view that for the

types ofexplanation have been offered for this phenomenon, but reasons he gives the LVF plays a particularly important part innone of those now taken seriously are in ecological terms. visually guided reaching, whereas search mechanisms involvedIndeed, it is difficult to see what ecological pressures could in locating and identifying objects must operate efficiently in

conceivably result in left rather than right hemisphere spe- both upper and lower visual fields? Part of the reason seems tocialisation for a particular mental al ility. [See also Corballis & be that he wants to develop a theory in line with the widely held

Morgan: "On the Biological Basis w, Human Laterality." BBS (but seldom articulated) dogma of complenmentary specialisa-1(2) 1978 and Bradshaw & Nettleton: "The Nature of Hemi- tion, which holds that if one part of the brain subserves aspheric Specialization in Man." BBS 4(1) 1981.] If anything, left particular function, then another part of the brain (preferablycerebral involvement in language must confer the slight disad- the opposite part) must subserve the opposite function. Thisvantage of making cople marginally less able to respond to dogma has already generated numerous unhelpfil dichotomies

speech coming from the left. Yet in everyday life we interact in theories of left versus right cerebral hemisphere asymme-without discomfort with people occupying this spatial position tries, it would be a shame if it were to be too readily imported(left of us). The underlying pressures ultimately responsible for into our thinking about the upper and lower visual fields.left cerebral involvement in language, and any overall advan-tages it confers, would seem to be more realistically sought ininternal, organisational factors (such as avoiding unnecessary orcomplex duplication of function).

One of the consequences of left cerebral specialisation forlanguage is that most right-handed people are better able torecognise words presented in their right visual hemifield (seeBryden 1982; Bradshaw & Nettleton 1983, for reviews). At one


ResponselPrevic: Visual field specialization

Author's Response quence of up-down visual field inversion. According toDolezal, a perceptual reversal occurred as to "where inthe FV [field-of-view] near and far surfaces appeared tobe. Specifically, with spectacles, physically distant places

Visual processing in three-dimensional appeared at the bottom of my FV, whereas places closerspace: Perceptions and misperceptions to me appeared closer to the top of the FV" (Dolezal 1982,Fred H. Previc pp. 246-247).

The target article clearly suggested that the empiricalCrew Technology Division, USAF School of Aerospace Medicine, Brooks support for the relationship between the dorsal pathwaysAFB, TX 78235-5301Electronic mall: prev'c%[email protected] and near vision and the LVF is much stronger than for tie

relationship between the ventral pathways and far visionTihe BBS trevtment of my article generated many diverse and the UVF. Some commentators (e.g., Young andcommentaries a:bout the proposed relationship between Crewther) accordingly accept only the former's exis-LVF-UVF processing differences and the different per- tence. Yet Young's suggestion that the latter link wasceptual requirements of peripersonal and extrapersonal partially contrived in order to create an aestheticallyspace. It also unveiled important new or overlooked satisfying dichotomy ignores the fact that attentional andempirical findings relevant to the various claims of this oculomotor imbalances in the direction of far vision aretheory. In the pages that follow, I review the commen- created by parietal damage, which clearly implies thetators' observations in basically the same order as fol- existence of a countervailing far attentional system (seelowed in the target article: I begin with a treatment of section 3.2.2 of the target article). The existence of twosome basic conceptual issues before moving on to a attentional mechanisms that are normally perfectly bal-discussion of specific behavioral, neurophysiological, and anced in the vertical and sagittal dimensions is alsoneuropsychological findings, and conclude with a brief congruent with recent evidence by Heuer and colleaguesmention of those comparative issues that relate to the (Heuer & Owens 1989) that resting vergence is close toLVF-UVF and near-far visual processing dichotomies. the edge of peripersonal sr , (' one meter) only when

the eyes are vertically centered. (Otherwise, vergenceincreases during lowering of the eyes and decreases

1. Conceptual Issues during ocular elevation.) Moreover, the existence of farvisual neglect in Heilman et al.'s bilateral temporal

1.1. The validity of the near-far distinction patient clearly demonstrates that a far attentional systemSeveral commentators (particularly, Young) object to my does exist in humans. Heilman et al.'s crucial findingdichotomization of near and far visual space, while others belies the claim of Kinsbourne & Duffy that the link(e.g., Williams) argue that I exaggerated the interdepen- between temporal lobe functions (e.g., visual search) anddencies of vertical (LVF-UVF) and sagittal (near-far) the UVF does not require the postulation of a specialprocessing. Still others (e.g., Kinsbourne & Duffy) object relationship between the UVF and far vision, a point toto the application of the near-far distinction to the spe- which I will return later.cializations of the dorsal and ventral neural pathways. In Although I demonstrated in the target article that theresponse, I must first point out the many visual scientists distinction between near and far visual space cannot be- including Helmholtz and Gibson - who have noted the easily defined according to the physical extent of the arm,important ecological links between the UVF and far two commentators (Brannan and Strong) nonethelessvision and between the LVF and near vision (see Heuer reiterate the fact that reading - a parvo/ventral function -et al. 1988). Indeed, one would have difficulty in in- often takes place within the confines of peripersonalterpreting even the most basic LVF-UVF differences space. Another commentator (Breitmeyer) further noteswithout recourse to these overarching relationships. the difficulty in categorizing the visual processing thatHow, for example, could one fully explain the UVF bias in mediates our ability to locomote and orient in the en-saccadic eye movements yet ignore the fact that UVF and vironment, since such processing utilizes far visual inputsLVF saccades also lead to greater divergence and con- (Brandt et al. 1975) yet is presumably mediated by thevergence, respectively (Enright 1989)? Second, I reiter- magno/dorsal pathways.ate that I am by no means the first investigator to argue As regards the first counterexample, I argue that read-that the near-far distinction may have a fundamental ing requires the processing and output mechanisms thatneurophysiological and neuropsychological basis. Al- are ordinarily applied to infomation in extrapersonalthough I cited Rizzolatti et al.'s (1985) near-far theory, the space, even though it typically occurs within arm's reach.neurophysiological distinction between near and far vi- More specifically, reading makes use of a retinotopicsion put fottui by Pettigiew and DLiehe (i987,). belis an scalillihg systen that is (a) largely free of head and limbeven closer resemblance to my own, as will be discussed movements, (b) foveally centered and located within thelater. plane of fixation, and (c) capable of perceiving local

Williams implies that a fundamental ecological rela- contours. This same characterization holds true for rflat-tionship between the LVF and near v ision and the UVF ed functions such as facial recognition, albeit to a moreand far -vision is not borne out b) in-erted pr;sm studies. limited extent. Whether faces are recognized inside theYet the perceptual adaptation experiments described b, boundar) of peripersonal space (e.g., in photographs) orDolezal (1982) directly contradict his assertion, as the be)ond its confines (as is the case for almost all real-lifedistortion of near and far space and of the size and faces that we encounter in our daily existence), they aredistances ofobjects within three-dimensional space actu- almost invatiably subjected to the Visual scanning andally proved to be the most dramatic perceptual conse- local contour analysis that are used in extrapersonal

BEHAVIORAL AND BRAIN SCIENCES (1990) 13'3 559

Response/Previc: Visual field specialization

space. Thus, facial recognition can hardly be considered a notopic coordinate system (given that Heilman ct al.'s"near" perceptual task, as Kinsbourne & Duffy propose. temporal patient partially manifested far visual neglect inRather than discarding the concept of a physically well- body-centered space - i.e., in a tactile bisection bias).demarcated near and far visual space, I prefer to view the Before concluding this discussion, I wish to respond toabove counterexamples as supporting a distinction Le- a specific remark concerning the ecological interdepen-tween body-centered space (whose sagittal midline lies at dency of vertial (LVF-UVF) and sagittal (near-far) spacearm's reach) and oculo-centered space (whose sagittal in arboreal primates. Crewther and Kinsbourne & Duffymidline lies in the plane of fixation). While the two speculate that visuomotor and other processing is notmidpoints occasionally deviate during activities such as confined to a particular vertice i.emifield in these spe-reading, a similar divergence of body-centered and oculo- cies. On the contrary, even in these species the head willcentered space can also occur in the context of left-right virtually always remain dorsal to the proximal limb (e.g.,vision (Kooistra & Heilman 1989). shoulder) joints - as will the eyes relative to the mouth -

The second objection of Breitmeyer can be met by along the body axis. As long as the animal maintainsfurther differentiating between the "far" space used in fixation on a more distal food object, therefore, thevisual orientation - the "visual background" according to retrieval and ingestion of it will predominantly occur inGrusser's (1983) scheme - and that synonymous with the the LVF regardless of the position of the head and/orvisual search and scanning field. The maintenance of body relative to the ground. ThL. major difference be-visual orientation, for example, is regarded as an "am- tween the ecological interactions of arboreal and ter-bient" process that traverses virtually the entire visual restrial primates would be the latter's relatively infre-field and does not require visual attention (Post & quent exposure to totally inverted images of extraperson-Leibowitz 1986), as compared to the more limited spatial al space. This would explain why, contrary to Kinsbourneextent and greater attentional demands of visual search & Duffy's assertion, facial neurons in the anterior tem-and scanning. (Indeed, it is common practice in vection poral lobe of macaques typically respond poorly to invert-experiments to instruct the subjects to divert their atten- ed faces (see Perrett et al. 1984, Figure 7) and why thetion away from far vision in order to maximize the perception of human faces is so poor for 180-degreeorientational effects of large-fieli image motion.) The far rotations in the frontal plane (Yin 1969).visual background likewise influences convergence statewithout requiring conscious awareness, as indicated by 1.2. Local-global perceptionthe "empty-field" myopia that accompanies loss of a Several commentators challenged the dichotomy be-textured background field (Whiteside 1952). This phe- tween linear/local and nonlinear/global perception andnomenon could explain the greater perceived "nearness" its relationship to parvo and magno function, despite theof objects when only central vision is present (see Dolezal fact that other theories (e.g., Bonnet 1987) have put forth1982, cited by Osaka).2 It may therefore be concluded similar distinctions. Bracewell and Bruce & MacAvoythat the ambient use of visual background information by argue that the link between linear neuronal processingparietal regions in maintaining visual orientation and and higher-order shape processing is dubious in that allconvergence does not invalidate the proposed segrega- complex shape recognition processes are necessarily non-tion ofperipersonal and extrapersonal attentional mecha- linear. While this argument is superficially valid in thenisms into the parietal and temporal cortices, respec- sense that a neuron that responds to a face in varioustively. This is especially true since visual orientation portions of the visual field clearly does not exhibit linearmechanisms housed in the parietal lobe undoubtedly rely responsiveness, it overlooks the fact that the "global"on a body-centered coordinate system that is more closely shape recognition carried out by the ventral system isaligned with peripersonal visual activities than with extra- based on the output of parvo (e.g., linear) contour pro-personal visual information processing. cessing and can be impaired by degradation of contour

In summary, peripersonal and extrapersonal visual perception (Desimone et al. 1985). It is also true that, asspace are best viewed as reasonably well-demarcated Brannan notes, "local" spatial processing is capable ofdepth sectors that can be defined according to different being space-averaged at a higher stage of processing. Butcoordinate frames, all of which are probably centered the global shape recognition that is achieved via a hier-near the reach of the arm in the resting state. By recogniz- archical assemblage of linear outputs is quite unlike theing that the division of near and far visual space can be processes involved in the perception of illusory contoursdrawn in different coordinate systems that do not always and other distributed forms, which are findamentallycoincide, one can preserve the near-far dichotomy with- nonlinear in nature (see Chang & Julesz 1984). An excel-out resorting to a "gradual transition" zone (see Williams) lent example of such nonlinearity is the response ofthat is uncharacteristic of the other two visuospatial certain neurons to an illusory contour even when there isattention dimensions (Hughes & Zimba 1987). One can absolutely no luminance; contrast located in their classicalalso dismiss the attempt of Strong and Crewther to dilute receptive field (Peterhans & Von Der Heydt 1989b). Asthe near-far neuropsychological distinction still further for Bracewell's argument that some nonlinear visualby suggesting that the dorsal and ventral systems may be processes also depend on local spatial interactions, hisbetter distinguished by their different processing modes example of directionally selective motion responsiveness(i.e., the ventral pathway is portrayed as a retinotopic is somewhat unfortunate in that direction-selectivity is"fixation" system) rather than by their emphases on relatively rare in the parvo-rich ventral (UVF) prestriatedifferent regions of space. This is because the parietal areas that clearly process local contour information (Fel-lobe does not have only a body- or head-centered system leman & Van Essen 1987, Figure 17).(see section 2.5 of the target article and Bracewell), nor It should also be pointed out that Brannan's belief thatdoes the temporal lobe maintain an exclusively reti- local motion perception can produce a solution to the

560 BEHAVIORAL AND BRAIN SCIENCES (1990) 13'3


correspondence problem is evidently based on a raisin- tion is not always yoked to our eye and limb positions) intoterpretation of Chang and Julesz's (1984) study. Chang a unified representation of three-dimensional visualand Julesz actually investigated the cooperative processes space arguably represents one of its greatest achieve-involved in detecting apparent motion in random-dot ments (as well as greatest mysteries!). As Abrams pointedcinematograms, a global percept. The task used in their out, much of this integration must lie outside consciousexperiments is in fact quite similar to one that has re- awareness, given the relatively poor visual localizationcently been shown to require the neuronal output of performance of our verbally mediated perceptual system.magno-rich MT (Newsome & Wurtz 1988).

Finally, the assertion by Bryden & Underwood thatvisual details are more important in near vision than in far 2. Issues related to LVF-UVF behavioral findingsvision is without an ecological basih precise spatial infor-mation is not only unnecessary in prforining most visu- Most commentators acknowledged that processing in theally guided limb movements in peripersonal space; it is vertical hemifields is probably not identical. Crewtherprobably nonavailable as well. The movement of the arm and Young perhaps reflected the majority opinion inin space is far too rapid (not to mention diplopic) to allow arguing that the data provide more support for specificfor the perception of local details, as evidenced by the LVF advantages than for UVF ones. But Findlay raisestremendous degradation of high-requency spatial con- the more general question as to why LVF-UVF process-trast sensitivity at high temporal velocities (Kelly 1977). ing differences, even when reliably present, are appar-

ently so slight. The obvious reply is that such differences1.3. Spatial coorclInate systems are minimized by the fact that (a) the vertical hemifieldsRecent evidence from visual neglect and neurophysiolog- align only partially and asymmetrically with near and farical studies indicates that multiple spatial coordinate visual space, (b) the segregation of near and far visualsystems are used by the primate visual system in per- space is itself somewhat imperfect, and (c) the visualforming various visuomotor tasks (see section 2.5 of the processing differences in peripersonal and extrapersonaltarget article). The exact nature and functions of these space are offset to a great extent by the large amount ofdifferent coordinate systems remain unclear, however. I common processing performed in these two sectors.originally argued that egocentric (head- and/or body- The proposed specializations of LVF processing re-centered) systems are used by the dorsal pathways to main unchallenged, for the most part. Bryden & Under-achieve precise visuomotor coordination in peripersonal wood suggest that the case for a LVF global processingspace, whereas the ventral system employs an oculocen- advantage could be greatly strengthened if some of thetered (retinotopic) one in its scanning of extrapersonal paradigms u.scd to evaluate left-right differences in localspace. Upper-lower differences could then be expressed versus global perception could also be applied to thein terms of visual spaces using the former coordinate investigation of LVF-UVF differences, a pointwith whichsystem and in terms of tsualfields using the latter one. I certainly agree. Although Chalupa & White claim thatThis argument attempted to explain why visual neglect - the failure to find differences in velocity judgments be-which is at least partially framed in egocentric spatial tween the LVF and UVF casts doubt on the interpreta-coordinates since it occurs even when free eye move- tion of other LVF motion superiorities, it is not clearments are permitted - is most frequently found in the whether the study in question (Smith & Hammond 1986)lower as opposed to upper visual quadrants. involved global or local motion processing. Finally,

Although Strong seems to have accepted this view to Bruce & MacAvoy claim that the LVF superiority insome extent, Butter raises the possibility that processing pursuit initiation has not been found in monkeys, butsuperiorities in both vertical hemifields are framed more their interpretation of Lisberger and Pavelko's (1989)in terms of body(hemi) space rather than retinal space. study is somewhat incorrect. While the greater ocularButter's view is partially confirmed by the deficit found in acceleration to LVF pursuit targets was less dramatic inHeilman et al.'s temporal-lobe patient, who apparently monkeys than in hum,"ns, Lisberger and Pavelko's resultsexhibited the classic symptoms of visual neglect in upper, did point to a LVF bias in the late interval of pursuitfar visual space. The presence of a body-centered coordi- initiation to targets located along the vertical meridian.nate system in the temporal lobe could also explain why As stated by the authors, "target motion in the inferiorhemispatial neglect can be expressed ideationally - i.e., visual field was consistently more effective at initiatingin the patient's imaging of extrapersonal space (Bisiach et pursuit than was motion in the superior visual hemifield."al. 1979), which involves primarily the ventral cortical (Lisberger & Pavelko 1989, p. 181) As for Bruce &pathway (Goldenberg et al. 1989). Conversely, the visual MacAvoy's claim that the ventral portion of the frontalneglect manifested by parietal patients clearly includes eye fields is involved in pursuit control, it should be notedboth space-specific and field-specific components (Kooi- that pursuit gains in frontally lesioned monkeys are ap-stra & Heilman 1989), in line with evidence that both parently most reduced for targets moving at low velocitieshead-centered and oculo-centered coordinate frames (Lynch 1987, Table 1). This finding, points to the spe-may be used by parietal neurons (Andersen et al. 1985). cialization of the frontal pursuit system for processing

The best conclusion at this time is that each cortical distant (i.e., slower-moving) targets.pathway maintains both types of coordinate frames, al- One commentator (Osaka) provides further support forthough the egocentric one may be weighted more heavily the superiority of the LVF in manual RT performance.by the dorsal system and the retinotopic one by the Osaka not only further documented the LVF RT latencyventral pathway. The ease with which the primate brain advantage (- 10-20 msec in his study), but also arguedcontinuously integrates the outputs of body, head, that the additional latency decrement for stimuli present-ocular, and even attentional coordinates (since our atten- ed in the lower temporal visual field relates to the fact that



this region processes information from the monocular In summary, the evidence for a specific role of UVFregion of the LVF periphery, in which the arm com- processing in visual search and scanning continues tomences its motion during reaching. His interpretation mount. Bruce & MacAvoy's reported bias of frontal eye-would, of course, mesh nicely with that offered in section field saccade neurons toward the UVF - albeit based on2.2 of the target article. the limited sampling of Bruce et al. (1985)- confirms the

The largest group of comments concerned the pro- results of frontal eye-field stimulation experiments inposed specialization of the UVF for visual search and humans (see Godoy et al. 1990) and provides the firstsaccadic scanning. Some commentators alluded to nega- glimpse of a possible neuronal correlate of this link.tive findings in this regard (e.g., Findlay & Harris 1984and Mishkin & Forgays 1952, both cited by'Bryden &Underwood; and Krose & Julesz 1989, cited by Bran- 3. Neurophysiological Issuesnan), but such findings must be interpreted with consid-erable caution. Findlay himself not only acknowledges 3.1. Retinal LVF-UVF differencesthe UVF saccadic bias in his commentary but cites evi- Recent evidence concerning the greater density ofdence (Findlay 1980, Levy-Schoen 1969) that further somatostatin-stained fibers in the inferior (UVF) retina issupports it. Also, Krose and Julesz's target recognition cited by both Williams and Chalupa & White. Thisexperiments used target and distractor stimuli that were intriguing finding represents the first evidence ofa specif-subjected to varying amounts of frontal-plane rotation, a ic retinal bias favoring the inferior retina, although it doessituation not normally found in extrapersonal space. Fur- not appear to be limited to primates. The greaterthermore, Krose and Julesz included only two subjects in somatostatin staining in the ventral retina has been linkedtheir experiments, one of whom actually showed a slight to both luminance and color processing (Sagar & MarshallUVF bias in some conditions (see their Figure 2). 1988), but the relatively poor luminance sensitivity of the

I believe it can be tentatively concluded that a UVF inferior retina and the lack of trichromatic vision insaccadic latency advantage does exist. At least four stud- nonprimate species that also show the somatostatin asym-ies have clearly documented such a superiority (Hackman metry render both of these interpretations somewhat1940; Heywood & Churcher 1980; Levy-Schoen 1969; problematic.Miles 1936), while one researcher (Findlay 1980) found a For unstated reasons, Williams dismisses previousnonsignificant UVF advantage and another obtained the findings of a higher ganglion cell density across theUVF latency bias in unpublib'sed findings (P. Hallett, superior retina (Perry et al. 1984; Van Buren 1963) as wellpersonai communication). By contrast, only one pub- as the rod-cone-density asymmetry reported by Oster-lished study has reported even a marginally nonsignifi- berg (1935). I find these findings quite convincing, evencant LVF advantage (Miller 1969). Addition,' -'-scale though their presence in other species makes them diffi-experiments involving a substantial number ui buojects cult to interpret within the context of the origins ofand controlling for or manipulating various stimulus pa- primate vision. Since there still exists no evidence of arameters (including the size and eccentricity of the target) specific difference between parvo (3) and magno (or)are clearly mandated, however. ganglion cells in their vertical distributions, it remains

There also appears to be a greater involvement of the unclear whether the LVF-UVF asymmetries observed atUVF in saccadic search and scanning. In addition to the the retinal level in any way relate to those observed in thefindings cited in the target article, those of Brandt (1945) higher-order visual cortical regions.and Hall (1985) - both cited by Bryden & Underwood -reinforce the conclusion that visual search commences in 3.2. Magno-parvo differencesthe upper quadrants, whereas a study that was published My synopsis concerning the distinctions between magno-after the completion of the target article further docu- cellular and parvocellular processing in the primate LGNments the UVF bias in target recognition performance received little mention. Kinsbourne & Duffy continue to(Yund et al. 1990). At least two observations should be maintain that the fundamental spatial distinction be-made concerning the former tendency. First, its alleged tween the two systems involves the central and pe-presence in preschool children (Hall 1985) and non- ripheral biases of the parvo and magno systems, respec-human species (see Breitmeyer) indicates that it is proba- tively, yet they offer no new empirical evidence. As Ibly not a trivial consequence of the top-to-bottom reading argued in the target article, there probably does exist aexperience in Western cultures. Second, Bryden & Un- slight difference in the mean retinal eccentricity of thederwood's use of the "global" precedence effect in per- two cell types, but the marginally more uniform retinalception (Navon 1977) to infer the spatial direction of distribution found in the magno layers cannot justify thevisual search is highly misleading, as I consider visual postulation of a basic central-peripheral dichotomy.search to be primarily a "local" process. Bryden & U nder- In the context of the inagno-partvu differences, i againwood's position is especially tenuous given that not all note the very similar theoretical analysis put forth byfindings concerning the lateral pattern of visual search Pettigrew and Dreher (1987). Building upon a previoussuggest a uniform left - right gradient (and hence a view of Levick (Levick 1977, cited in Pettigrew & Dreherglobal --9 local progression). While most research sup- 1987), these researchers postulated a greater role of theports the notion that scanning begins in the upper left X- and Y-systems of cats in far and near vision, respec-quadrant and proceeds rightward in the UVF, the scan- tively. Since Y-type cells are confined exclusively to thening direction in the LVF is less agreed upon, compare magno layers in primates whereas X-cells are biasedBrandt (1945) and Chedru et al. (1973). Moreover, Hall's toward the parvo layers (see section 3.1 of the target(1985) results question whether even UVF scanning al- article), an extension of Pettigrew and Dreher's analysisways commences on the left. to primates would predict that the magno layers mediate



transient visual processing in both the near and far sectors mates can only detect a monkey at ten meters with a one-of visual space, whereas the parvo layers contain neurons degree resolution (to use Siegel's example), how couldinvolved only in far vision (including the plane of fixation). the dorsal system detect a particular fruit object or deter-

mine its edibility at such a modest distance? Fourth, the3.3. Dorsal-ventral differences entire resolution issue is largely moot in that only a smallThis analysis appears to have generated more controversy fraction of area 7a neurons are responsive prior to saccadicthan any other section of the target article. For the most eye movements, as Bracewell concedes. Although this ispart, the criticisms concerned the portrayal of neuronal not as true of neurons in area LIP, this latter parietal areavisual processing in the dorsal visual system. At least one is less clearly linked to the "classical" dorsal systemminor criticism (noted by Bracewell, Bruce & MacAvoy, leading from MT --* MST --+ 7a (Maunsell & Newsomeand Siegel) was valid, as evidence of parietal neuronal 1987). Moreover, LIP's involvement in saccadic eyeinvolvement in a head-centered coordinate system has movements does not automatically entitle it to be consid-indeed been found only at the population level so far (see ered a voluntary extrapersonal scanning center, sinceAndersen et al. 1985). Conversely, two other minor many saccadic mechanisms are reflexive or orientationalcriticisms of my characterization of ventral neuronal func- in nature - e.g., the "express" saccade system (seetioning were without justification. Contrary to Bruce & Fischer & Breitmeyer 1987) and the peripheral oneMacAvoy's assertion, I clearly stated in the target article located in dorsolateral prefrontal cortex (see Bruce 1988).(section 3.3.1) that no evidence of an UVF bias among Rather, the head-centered coordinate system associatedinferotemporal reci-ptive fields has yet been reported. I with area LIP suggests that it is probably not involved inalso noted (section 3.1) that the parvo system probably the voluntary scanning of complex scenes, which remainssends a balanced projection to areas V3 and VP, which largely intact following parietal damage (Karpov et al.runs counter to Crewther's suggestion. 1968).

Bruce & MacAvoy dispute the evidence that neuronal Many commentators, (Breitmeyer, Crewther, Kins-receptive fields in posterior parietal cortex are biased bourne & Duffy, and Siegel) continue to assign to thetoward the LVF. They claim that the data contained in dorsal system a role in processing optical flow informa-Figure 8 of the target article (reprinted from Robinson et tion. If one views "optical flow" in the generic sense (suchal. 1978).do not unequivocally demonstrate the existence as the optical flow fields resulting from a head movementof a LVF bias, but it should be emphasized that Robinson in space), then I agree with Siegel that many neurons inet al. themselves mentioned the parietal LVF bias in their the dorsal system probably engage in some sort of flowarticle. As regards Bruce & MacAvoy's interpretation of analysis. My specific objection is to the frequently es-Mountcastle et al.'s (1984) findings about the vertical poused view that the classical dorsal system (and es-distribution of PP neuronal fields, Mountcastle himself pecially area 7a) is critically involved in the processing ofwrites that "we have always had the impression that the optical flow information during locomotion, which wouldlower fields are more intensely represented, but we could obviously conflict with its emphasis on peripersonalnever prove it because our equipment . . . is all placed in space. I originally offered what I considered to be fivesuch a way as to make it difficult to examine the lower reasonably compelling arguments as to why the globalfields as extensively as we would like." (V. Mountcastle, motion analyses performed by area 7a neurons are proba-personal communication) Finally, a LVF bias in area PP bly not related to visually guided locomotion. Although Iwould clearly be expected by virtue of the fact that area may have overstated some of the arguments, the basicMT - whose map is dramatically skewed toward the LVF conclusion remains valid. As but one example, Siegel'sbased on an extensive sampling of almost its entire ter- observation that a more even distribution of centrifugalritory (Maunsell & Van Essen 1987, Figure 8) - provides and centripetal opponent-vector neurons exists than wasPP indirectly (via area MST) with its major source of found by Steinmetz et al. (1987) remains incompatiblevisual input (Maunsell & Newsome 1987). with the fact that radial flow patterns during locomotion

The putative lack of involvement of dorsal system are almost exclusively expanding (i.e., centrifugal). Byneurons in spatial localization and saccadic eye move- contrast, a balanced distribution of radial flow processingments was challenged by Bracewell, Marsolek, and Sic- is exactly what would be expected if such neurons weregel, who all argued that parietal neurons do exhibit, at primarily involved in monitoring the outward and in-least at the population level, the resolution required for ward movement of the arm in peripersonal space.saccadic localization in exti.,personal space. Several Some of Siegel's other concerns (e.g., the bias of dorsalpoints can be raised in a rebuttal of their position, neurons toward crossed disparities) are more legitimatelyhowever. First, the "one-degree" resolution limit cited based on the dearth of definitive findings on the subject.'by Siegel is based on a theoretically derived population His criticism of my statement that opponent-motion nev -

model that has yet to be empirically tested, as Zipser and er occurs during forward locomotion is also partly accu-Andersen (1988) concede. Second, even a one-degree rate, in the sense that opposite flow does occur for imagesresolution capability is hardly impressive, and more than that are in front of Nersub behind the fixation point. Untilanything else serves to confirm the tremendous loss of it can be shown to what extent observers actually main-high-frequency spatial vision that follows isolation of the tain a near ur intermediate optical focus during terrestrialmagno system (which projects to area 7a). Third, the locomotion, 1 ioweN er, I remain conv inced that primarilymajor issue should not be whether or not most saccade common image motion (with relative translation) is expe-movements in extrapersonal space are greater than the rienced as a consequen-ec of forward locomotion. As toone-degree limit but whether or not such resolution is whether the maximum extent (- 40 degrees) of area 7aadequate for processing the details of objects that require ncuronal fields is sufficiint to procss optical flow inforscanning in extrapersonal space. For example, if pri- mation, the fact that the functional ficld-of-%view for main-

BEHAVIORAL AND 89AIN SCIENCES (1990) 13'3 563


taining visual orientation extends well beyond 100 de- because the Borough Council won't allow it inside, andgrees (Dichgans & Brandt 1978) indicates that area 7a is at I should then find the bathroom and the bath would beleast insufficient for this purpose. Rather, I continue to in it.' Thus she clearly visualized the landmarks of thehold that dorsomedial posterior parietal cortex - which correct route (emphasis added). When asked to de-lies much closer anatomically to the lower limb projection scribe how she would find her way from the tuberegion in somatosensory cortex and whose extremely station to her flat she described this in detail correctlylarge receptive fields are ideal for analyzing motion flow and apparently visualizing the landmarks, but sheacross the entire visual field - will eventually be disclosed consistently said right instead of left for the turningsas the critical processing center for radially expanding except on one occasion. (Brain 1941, p. 259)flow information resulting from forward locomotion. Inthe meantime, I await the outcome of a good "ecological" 3.3. Neuropsychological Issuestest of the competing positions - namely, the stimulation There were relatively few specific comments that directlyof area 7a opponent-vector neurons with the image of the disputed the neuropsychological evidence and interpre-animal's hand and arm during reaching in front of the tation presented in the target article. Strong raised thefixation point versus an equivalently sized optical looming general issue of whether "phenomenological unity" needpattern presented at optical infinity, require a particular type of localization in the brain. If the

Finally, I wish to dispute the neurophysiological argu- brain operated strictly in a distributed fashion (i.e., as aments presented by Bruce & MacAvoy, Kinsbourne & "physical mind"), then Strong's argument would carryDuffy, and Marsolek in favor of other previous dorsal- more weight. In reality, the brain is a biological organventral dichotomies, especially that of Ungerleider and with an evolutionary history (that aligned both the lower-Mishkin (1982). Although I acknowledged in the target body somatosensory and LVF representations in thearticle that the dorsal and ventral systems do differ dorsal half of the posterior brain) and physiological con-somewhat along the central/peripheral dimension, such straints (transmission time, noise, etc.). I argue that thedifferences are overshadowed by the tremendous overlap basic partitioning of the primate brain, to the extent thatof their neuronal receptive field locations in two-dimen- it exists, should be related to some primordial attribute ofsional space. Area 7a neuronal fields, for example, rarely the visual environment that maximizes parallel process-extend beyond 40 degrees in diameter (Andersen et al. ing of information located in different sectors of space,1985) and, in most cases, overlap the foveal region. 4 This different spatiotemporal 'lomains, or different visu-accords with (a) the well-documented role of the dorsal omotor coordinate frames. it would be extremely ineffi-pathway in foveal pursuit, and (b) the sharp demarcation cient for the brain to engage in parallel processing ofof the parietal neglect phenomenon, such that even foveal within-object and between-object perception, which oc-vision in the contralateral visual field is typically affected cur in the same sector of space (extrapersonal), involve(see Fig. 4 of the target article). By comparison, the similar types of processing (local contour analysis), andaverage IT field extends - 25 degrees in diameter and is us the same visuomotor coordinate system (oculo-cen-variably centered anywhere from directly on the fovea to tered). It would be even less efficient to assign neuronalas far as 12 degrees off-center (Gross et al. 172, Fig. 3). "double-duty" for near and far visual stimuli that occur inThus, the assumption of Marsolek that IT fields are different depth regions (whose switching time is at least"almost always found on the fovea" is rather misleading, 100 msec), require antagonistic spatiotemporal process-which in turn casts doubt on the validity of the computa- ing capabilities (local/sustained vs. global/transient), andtionally derived support for the notion that the dorsal are largely framed in different coordinate systems (body-system is a "spatial" processor and the ventral one an vs. oculo-centric)."object" processor. A more specific disagreement with Strong relates to his

In fact, the parietal lobe is only a spatial processor to acceptance of the view that parietal processing is orientedthe extent that left-right confusions, inability to perceive more toward "allocentric" than "egocentric" space. Theobject rotation, inattention to certain portions of the predominant opinion of researchers today is unquestion-visual field, and other symptoms of an impaired ego- ably that parietal cortex is critical for the neural repre-centric/peripersonal attentional system seriously de- sentation ofegocentric space (see Stein 1989). As Mount-grade overall spatial perception (see Crowne et al. 1989). castle states:The ability of parietal patients to recall accurately the The parietal lobe, together with the distributedconfiguration of extrapersonal space often remains re- system of which it is a central node, generates anmarkably intact, as illustrated by one of Brain's (1941) internal neural construction of immediately surround-patients: ing space, of the location and movements of objects in it

...When she set out for the bathroom she arrived in relation to body position, and of the position andat the lavatory, which was a door on the right, and when movements of the body in relation to that immediatelyshe tried to go to the lavatory she made a similar surrounding space. (Butter et al. 1989, p. 145)mistake, took a turning to the right and got lost again. In contrast, evidence favoring the involvement of par-Yet when she was asked how she would find her way to ictal cortex in the representation of allocentric spacethe bathroom, the door of which was on the left at right remains highly equivocal. While some findings involvingangles outside her bedroom door, she replied. 'I should the "landmark" prmblem and similar tasks (Ungerleider &go first to the cupboard in which my husband keeps his Mishkin 1982) provide support for Strong's notion, thereclothes.' (This was near the bedroom door.) 'Then I have been several failures to replicate these findingsshould open the bedroom door and outside would be when lesions are limited to the inferior parietal lobe itselfwhere the coats are hung up. I should then look for the (see Lynch 1980). It is not clear, moreover, what effectelectric light switch which is outside the bathroom, left-right confusions and other egocentric factors have on



supposedly "allocentric" tasks, since such tasks have ing from the various commentaries, as well as the findingstraditionally required manual responses on the part of the of Pettigrew and Dreher (1987), the distinction betweenanimal. near and far v'sion may be useful in understanding tile

I also disagree with Strong's suggestion that the par- physiology of n, my, if not most, vertebiate visual sys-ictal lobe is crucial for perceiving feature-conjunctions. tems. It is also clear that certain specializations prev iouslyHis view conflicts with a large body of evidence that regarded as limited to primates (such as the presence ofpoints to a role of inferior temporal cortex in such activity "face-neurons" in the anterior temporal lobe) can be(see Harter & Aine 1984). It would be most surprising *f found in other species (see Bracewell).the temporal lobe were heavily invol'ed in most aspects Goodale & Graves provide intriguing evidence of aof object perception (which almost everyone, including division in the pigeon visual system that resembles that ofmyself, acknowl, dgcs) yet not responsible for performing the primate. While a parallel exists in that the near % isualthe very processing that is essential for tile integration of field is biased toward the LVF in both species, there alsofeatures into complex form percepts. appear to be certain differences between the partitioning

The commentaries yielded one new neuropsycholo- of the two visual systems. For one, the far visual field ingical finding - that of Heilman et al., who documented primates is clearly binocular, as neurons in the ventralthe existence of UVF/far visual neglect following bilateral system are precisely tuned to binocular disparity (Fel-inferior temporal lobe damage. Tile importance of this leman & Van Essen 1987). In many respects, therefore,finding for the present theory cannot be overstated, as the the monocular far field of pigeons may actually resemblepostulated relationship between the temporal lobe and more closely the visual background field of primates,far vision and the UVF represented perhaps its weakest which I have previously shown to be divorced from thelink, given that no direct evidence supported it and no far attentional field. Anlther difference relates to theprevious conceptualization had ever entertained it. more important role of monocular information in peripel -Heilman et al.'s finding, though preliminary, provides sonal visual functioning in primates. For example, thethe first direct empirical support for the role of tile visual registration of the initial position and motion of theinferior temporal lobes in far vision, although it has long arm during reaching begins in the monocular portion ofbeen recognized that abnormal temporal-lobc activity in the peripheralv isual field in primates, although binocularepilepsy and schizophrenia can lead to an aberrant em- vision (and especially crossed-disparity information) ispiasis on far "ps)ciologicalr space ke.g., hallucinations, obviously of great importance in primate peripersonalgrandiose themes and delusions, distorted body-images). space.Heilman ct al.'s finding also counters Kinsbourne & Williams claims that LVF-UVF asymmetries in theDuffy's claim that the relationship between higher-order early stages of mammalian visual processing may actuallyventral processing and the UVF need not require a be greater in other species than in primates. As notedspecific involvement of the temporal lobes in far vision, earlier, however, he unwarrantedly dismisses the clearHowever, Kinsbourne & Duffy's "distance principle" can evidence of ganglion cell density differences put .- rth bybe challenged on other grounds as well, in that anatom- Perry et al. (1984) and Van Buren (1963). A more accurateical proximity (i.e., the inferior temporal lobe is closer to assessment would probably be that LVF-UVF retinalthe inferior/UVF than the superior/LVF occipital cortex) asymmetries are highly' similar across most mammaliandoes not necessarily translate into neural proximity (i.e., species, and probably stem from a factor - possiblythe inferior temporal lobes apparently receive a balanced evolutionary - that is common to each (e.g., the lumi-projection from the inferior and superior occipital nance gradient from sky to ground). On the other hand, Iregions). certainly agree with Williams's suggestion (3) that one

Finally, I agree with Bryden & Underwood that a should expect to find the most striking evidence of pri-general theory of the neuropsychology of human visual mate-specific vurtical asymmetries in those areas beyondperception must ultimately be able to account for dif- area 17, since higher cortical regions are much moreferences in the way the two hemispheres process local likely to be influenced by the primate's behavioral experi-versus global information. Although I briefly noted in the ences in its three-dimensional visual world. This is not totarget article (section 4.1) that hemispheric differences in say that magno-parvo vertical asymmetries will notglobal and local processing probably exist, a more exten- eventually be found in area 17, or even at subcorticalsive treatment of this topic is found in a companion levels. After all, the evidence obtained to date can hardlytheoretical article (Previc, in preparation). The most be described as definitive.plausible scheme based on current evidence is that the In summary, the differentiation of near and far space inleft temporal lobe and right parietal lobe are most adept at the primate somewhat parallels that found in many otherprocessing local and global information, respectively, species. However, some species (e.g., avians) have excel-while tie left parietal and right temporal lobe occupy lent far vision but more iinted near vision, whereasmore intermediate positions along this processing dimen- others (e.g., the nocturnal mammals) may be much moresion. Partial neuropsychological support for such a notion adept in the processing of information found in periper-has recently been provided by Robertson et al. (1988)!, sonal than in far visual space. The remarkable achieve-

ment of primate evolution has been to assemble animpressive command of far visual space with an extremely

4. Comparative Issues sophisticated peripersonal visuomotor capability.

The final set of issues that evoked a large number ofcommentaries pertained to the differentiation of near-farand LVF-UVF processing in nonprimate species. Judg-


References/Previc: Visual field specialization

5. A prolegomenon to a future visual science number of MSTcells whose diameter exceeded 40 degrees, buttheir maximum visual screen diameter was only 80 X 66 de-

I conclude that my theory concerning the divergence of grees. Furthermore, the average neuronal receptive field di.near and far vision in the primate and its implications for ameter in area MT rarely exceeds 20 degrees (Komatsu & Wurtzthe partitioning of the highe:" brain pathways remains 1988, Fig. 3; Maunsell & Van Essen 1987, Fig. 2), and isquite tenable. This theory has built carefully upon pre- typically centered less than 20 degrees from the midline.vious neurophysiological and ecological conceptualiza- 5. See also Yund et al.'s (1990) data concerning target recog-

nition in the four frontal-plane quadrants.tions and has marshalled a substantial amount of diverse 6. Additional findings in support of a LVF neglect bias areevidence in its favor, some of which is by now virtually presented in a recent paper by lalligan and Marshall(1989) thatunassailable (e.g., the greater LVF impairment in par- appeared too late to be included in thc target article.ietal neglect). 6 At the very least, this theory deserves astatus beyond the "premature" one implicitly accorded itby Bryden & Underwood.

In emphasizing the importance of the individual's eco- Referenceslogical visuomnotor interactions in three-dimensionalspace, this theory has tried to clarify several emerging Letters "a" and "r" appearing before authors' initials refer to target article andlines of thought about how the human brain processes response respectively.

Abrams, R. A. & Landgraf, J. Z. (in press) Differential use of distance andvisual information. Many of these conceptual elements - location information for spatial localization. Perception &including multiple coordinate systems, three-dimension- Psychophysics. lRAAlal attention, altitudinal neglect, disparity pools, etc. - are Abrams, R. A., Meyer, D. E. & Koniblum, S. (1989) Speed and accuracy ofstill in their infancy, as only scattered references have saccadic eye movements. Characteristics of impulse variability in the

ben made to them until recently. Hence, this theory's oculomotor system. Journal of Experimental Psychology: Huanbeen mPerception and Performance 15:529-43. [RAAImain contribution is to point out some of the guideposts ins press) Eye.hand coordination. Oculomotor control in rapid aimed limbfor future researchers to follow in their efforts to unlock movements. Journal of Experimental Psychology. Hluman Perception andthe secrets of three-dimensional visual processing in the Perfomance. IRAAIhuman brain. No longer can visual cortical physiology Adler-Grinberg, D. & Stark, L. (1978) Eye movements, scanpaths, and

dyslexia American Journal of Optometry and Physiological Opticsremain in the province of dorsal VI, nor can spatial vision 5:557-70. laFHPIcontinue te be relegated to processing along the horizon- Aiple, F. & Kruger, J. (1989) The brain is an input driven neural network.tal meridian. Nor can disorders such as teleopsia and Connectivity analysis of monkey visual cortex by cross correlating spikepelopsia ("nearness") remain largely neglected by visual trains. In. Neural networks from models to applications, ed. L. Personr.az

& C. Dreyfus. IDSET. lCWSlneuropsychologist(, given that good objective techniques Albert, M. L., Soffer, D., Silverberg, R. & Recies, A. (1979) The anatomic(e.g., tactile bisection and "dark focus") now exist for basis of visual agnosia. Neurology 29.876-79. (aFiIP)measuring the sagittal distance of patients' attentional Albright. T. D. & Desimone, R. (1987) Local precision of visuotopieand optical foci. Nor should it be considered proper in the organization in the middle temporal area (Ml) of the macaque.future to refer to a neuron in such nonecological terms as Experimental Brain Research 65:582-92. IRMSI

whe ma AlIman, J. (1977) Evolution of the visual system in the early primates.opponent-vector" when it may respond just as well, if Progress sit Psychobiology and Physiological Psychology 7:1-53. [aFHP,not better, to the reaching movement of the arm in three- RMSIdimensional peripersonal space. Nor should researchers (1988) Variations in visual cortex organization in primates. In. Neurobtologycontinue to see the conservative hand of evolution direct- of ncocortex, ed. P. Rakie & W. Singer. Wiley & Sons. laFIIP]ing the development of every higher-order perceptual AlIman, J., Miezin, F. & McGuinness, E. (1985) Direction- and velocity-specific responses from beyond the classical receptive field in the middlemechanism, when the posterior parietal cortex and other temporal visual area (MT). Perception 14:105-26. (aFfIPIbrain areas that house such mechanisms are almost en- Andersen, R. A. J1987) Inferior parietal lobule function in spatial perceptiontirely dependent on visual experience for their specific and visuomotor integration. In landbook of physiology, vol. 5, part 2,development (Hyvarinen 1982, Chapter XIc). ed. V. B. Mountcastle, F. Plum & S. H. Geiger. American Physiological

Society. [aFIIPThe maturation of a three-dimensional visual science (1988) Visual and visual.motor functions of the posterior parietal cortex. In:

will inevitably lead to revision or abandonment of many of Neurobiology of neocortex, ed. P. Rakic & W. Singer. Wiley &this theory's specific tenets. As a general theory, how- Sons. IaFIIPIever, its legacy will, one hopes, endure much longer. Andersen, R. A., Bracewell, R. M., Barash, S., Fogassi, L. & Cinadt, J. W.(in press) m.>e position effects on visual, memory and saccade-related

activity in areas LIP and 7a of macaque. Journal of

NOTES Neuroscience. [RMBJ

1. I regret that I was not made aware of Pettigrew and Andersen, R. A., Essick, C. K. & Siegel, R. M. (1985) Encoding of spatialDreher's article until after the target article was submitted for location by posterior parictal neurons. Science 230.456-58. [arFHP,

RMB, CJM\, RMS]commentary. Anderson, S. J., Mullen K. T. & [less, B. F. (1989) Asymmetry in the retina

2. As Volezal points out, the perceived nearness (and associ- for motion and colour detection, Investigative Ophthalmology & Visualated micropsia) of objects in his experiments was probably also Science 30:53 (Supplement). [aFIlIP]due to the effect of the occluding tube's frame in drawing Anstis, S. M. (1978) Apparent movement. In. Handbook of scnsoryvergence closer, physiology, vol. 8, ed. R. field, H. W. Leibowitz & If.-L. Teuber.

3. Dr. G. Poggio recently informed me, however, that a Springer-Verlag. [aFrIPIcrossed-disparity bias exists among the disparity-sensitive neu- Badcock, D. & Lovegrove, W. (1981) The effects ofcontrast, stimulusrons located in the dorsal (LVF) portion of area V1, so a z,,. lar duration, arid spatial frequency on visible persistence in normal and

LVF-dominated higher magno specifically disabled readers Journal of Experimental Psychology: llumancrossed-disparity bias in the LVFPerception and Pcrfornnnce 7:495-505. [aFIIPIstructures remains extremely likely. Bahill, A. T., Adler, D. & Stark, L. (1975) Most naturally occurring human

4. The claim of Kinsbourne & Duffy that MST fields are saccades have magnitudes of 15 degrees or less. Investigative"often more than 100 degrees in diameter" is highly overstated. Ophthalnology 14:468-69. laFtiPiNot only did Komatsu and Wurtz (1988) record only a small Barbeito, B., Bedell, 1. E., Flom, M. C. & Simpson, T. L. (1987) Effects of


Rcferences/ Previc: Visual field specialization

luminance on the visnial acuity or strabismic and anisometropic amblyopes Operations Symposium at Harvard. Positive Development Duringand optically blurred normals. Vision Research 27:1543-49. [aFlIPI Adolescence and Adulthood, June Cambridge, MA. (JRB1

Barbour, C. C. 6c Coss, R,. G. (1988) Differential color brightness as a body Brannan, J. R. & Williams, MI. C. (1987) Allocation of visual attcntion in goodorientation ciue. Hluman Factors 30:713-17. [aFHP] and poor readers. Perception and Psychophysics 41:23-28. (IB

Barlow, H. B. (1986) Why have multiple cortical areas? Vision Research Bravo, M., Blake, R,. & Morrison, S. (1988) C~its sec subjective contours26:81-90. [RMBI Visioin Research 28-861-65. [aFIlPI

Barlow, 11. B. & Levick, W. R. (196) The iseural msechianisnm of direttionally llreitineycr, B. G. (1980) Unimaskiing visual masking. A look at the "why"selective units in the rabbit's retina. Journal of Physiology (London) behsind the veil of the "how",' Psychological Rewn 87.52-69. [aFIIP]178:477-508, (aMBI Breitineyer, B., Battaglia, F. & Bridge, J. (1977) Existeiice aiid implications of

Barlow, J. S. (1970) Vestibular aisd ison-doiniant parietal lobe disorders iii a tilted binocular disparity space. Perceptioni 6.161-64. [aFI-P, BGBIman. (Two aspects of spatial disorientation iii man.) Diseases of the Brcitineycr, B., Julesz, B. & Kropfl, WV. (1975) Dyinamsic randons-dotNervous System 31-.667-73. [aFhIP1 stereograms reveal up-down anisotropy aisd left-right isotropy betweeii

Bender, M. B. (1980) Brain control of conjugate horizontal and vertical eye cortical hemnifields. Science 187:269-70. [aFHPImovements. A survey of the structural aisd functional correlates. Brain Bndlgensan, B. 4.1989) The psychopsics of the Isursut oculoinotor system103:23-69. [aFHPI Perception & Psychophysics 46:220-26. (aFIIP]

Bender, NI. B. & Feldman, NI. (1972) T'he so-called "visual agniosias." Brain Bnidgemnsi, B., Lewis, S., Heit, C. &c Nagle, MI. (1979) Relation betwees95:173-86. (aFtHP] cognitive and nsotor-orieiited systems of visual psositioin perception.

Bender, M. B., Feldman, \M. & Sobin, A. J. (1968) Palinopsia. Brain 91.321- Jou; nal of Experimental Psychology. lumn Perception and Peirfornianicc38.t IaFHP1 5:692-700. [RAAI

Bender, M. B. & Furlowe, L. T. (19415) Phicnoinoin of visual extinction in Brown, B., Haegerstrom-I'ortnoy, C., Yingling, C. D., Hferron. J., Calmn, Dhomonymous fields and psychiologie principles involved. Archives of & Marcuq, \I. (1983) Tracking eye inovensents are normal in dyslexicNeurology and Psychiatry 53:29-33. IaFHIPI childrein. Amierican Journial of Optometry and Physiological Optics

Benton, A. L., Levin, H. S. & Van Allen, M. WV. (1974) Ceograpshic 60.376-83. IaFHPIorientatioin in patients with uinilateral cerebral disease. Neuropsychologia Bruce, C. J. (1988) Single neuron activity in the monkey's psrefrontal cortex12:183-91. [aFHPI In: Neurobiology of neocortcx, ed. P. Rakic, & WV. Singer. Wiley&

Berkley, W. L. & Bussey, F. R. (1950) Altitudinal hemnianopia: Reisort cf two Sonis. [arFHP1cases, Amnerican jouirnal of Opthalosology 22:5931-600. [aFHIP1 Bruce, C. J., Desinsone, R. & Cross, C. C. (1986) Both striate cortex aisd

bishop, A. (1962) Control of the hand in lower primates. Annals of the Neu, superior colliculus contribute to visual propserties of neuiroins in superiorYork Academy of Sciences 102:316-37. [aFHP] temporal polysensory area of macaque monkey. Journial of

Bisiach, E. & Luzzatti, C. (1978) Unilateral neglect of representational space. Neurophysiology 55:1057-75. [CJBICortex 14:129-33. [R'MB] Brice, C. J., Coldberg, NM. E., Bushnell, NI. C. 6c Staiston, C. B. (1985)

Bisiach, E., Luzzatti, C. & Perani, D. (1979) Unilateral neglect, Primate frontal eye fields. I1. Physiological and anatomical correlates ofrepresentatiunal schema and consciousness. Brain 102,609-18. IRNIB, electrically evoked eye movements. Journal of Neurophysiology 54.714-rFHP1 34. [arFIIP, CJBI

Blakemore, C. & Vital-Durand, F. (1986) Organization and tsost-natal Bmvden, NI. P. (1952) Literality, Academic Press. INIPBIdevelopment of the monkey's lateral geniculate nucleus. Journal of (1982) Laterality. Functional asymmetry ini the intact brain. AcademicPhysiology 380:45,t-91. [aFHP1 Press. JAWYI

Blasdel, C. C. & Fitzpatrick, D. (1984) Physiological organization of layer 4 in Bnirkhalter, A., Felleman, D J., Newsome, W. T. & Van Esseis, D. C. (1986)macaque striate cortex. Journal of Neisroscience 4.880-95. [aFIlP1 Anatomical and physiological asymmiietes related to ',isual ar, as V3 aiid

Blasdel, C. C. & Lund, J. S. (1983) Termination of afferent axons in macaquie VP in macaque extrastriate cortex. Vision Research 26 63-80 1aFHP[striate cortex. Journal of Neuroscience 3.1389-1113. [aFIIP] Burkhalter, A. & Vais Esscr, D. C. (1986) Processing of color, form and

Bogacz, J., Nlcndilalsarsu, C. & De Nleidilaharsu, S. A. (1974) Electro- disparity informsation in visua~l areas VP aiid V2 of ventral extrastriateoculograpliic abnormalities during pursuit movements in developimental cortex in the macaque monkey. Journal of Nemroscience 6:232.'-dyslexia. Elect roenscephalography ands Clinical Neurophyslology 36.651- 51. [aFHPJ56. [aFilP1 Bushnell, MI. C., Coldberg, M. E. & Robinison, D. L. (1981) Behavioral

Boinnet, C. (1987) Les temps de traitement dans la perception visuelle des enhancement of visual responses in monkey cerebral cortex. Iformes. In. Psychologie cognitive. Moddes et indthodes, ed. J. P. Caverni, Miodulation in posteriorparietal cortex related to selective visualC. Bastion, P. Mendelsohn & C. Tibcrghicn. Presses Universitaires de attention. journal of'Iemrophysiology 46:755-72. [aFIIP)Crenoble. [arFilP] Butter, C. NI. (1969) Impairments in selective attention to visual stimuli in

Bracewell, R. M., Barash, S., Fogassm, L., Cnadt, J. W. & Andersen, 11. A. nmonkeys witls inferotemisoral and lateral striate lesions. Brain Rlesearch(1989) Saccade-related activity in the lateral intraparietal cortex of rhesus 12:374-83. [aFHPImonkeys. Perception 18:514. [RNIBI Butter, C. NM., Evaiis, J., Kirsch, N. & Kcwman, D. (1089) Altitudisal

Bradshaw, J. L. & Nettleton, N. C. (1983) fhuman cerebral asymmietry. neglcct following traumsatic braiii injury. A case report Cortex 25.135-Prentice-.Hall. [AWYl 46. [aFI-P, CNIBI

Bradshiaw, J. L., Nettletoms, N. C., Piersoii, J. NI., Wilson, L E. & Nathan, Campbell, A.. Jr (1978) Deficits in visual learniisg produced by posteriorC. (1987) Coordies of extracorporeal space. In. Neurophysical and temsporal lesions in cats. Journal of Comparative and PhysiologicalmeuropsychologiMa aspects of spatial nieglect, ed, M. Jeannerod. North- Psychology 92:45-57. [aFHPIHolland. [aFIIPI Capspa, S., Sterzi, It., Valla,, C. & Bisiacli, E. (1987) Reinission of

Bradshaw, J. L., Nettleton, N. C. 6c Taiylor, M. J. (1981) Bight hemisphere herniiieglect aisd anosognosia duriing vestibular stiinulationi,language aisd cognitive deficit in sinistrals? Neuropsychologia 19.113- Neuropsychologma 25:775-82. [aFIIPI32. [AW Y] Carlow, T. J., Flynn, J. T. 6c Shipley, T. (1976) Color Iserimetry. Archives of

Bradshaw, J. L. & Sherlock, D. (1982) Bugs aisd faces in the two visual fields: Ophthalmology 9,1:1492-96. [sF1111T1e analytic/hiolistic processing dichotomy and task sequenicing Cortex Cannon, A., Nachishon, I. & Staruisky, It k.1976) Developmental aspects of18:211-26. [MPBJ visual lieinifield diffeorences in psercepstion of verbsal material. Brain andI

Bradshaw, NI. F., Frisby, J. P1. & Mayhiew, J. E. W. (1087) The recovery of Laniguage 3.463-69. [AWYJstructure from motion. No evideiie fur ai sise(ial link wilsthle vuivergeus Uavaiiagh, P., lboegli, J. & Liznier-1usreaiu, Q. k198 5

) I'eiceptiun of imiotiondisparity mechanismi. Perceptioni 16-351-57. 1011I1) in equiluinous kineinatograms. Perception 14.151-62 [aFIl P]

Brain, W. It. (1941) Visual disorieiitation isvth special reference to lesions of Chsikims, J. D., Corbin, If. If. & Volkunann, J. (1962) Mlapping a field ofthe right cerebral hemisphere. Brain 64:2441-72. [1KNII, rFIIP[ shiort-tis visual search. Science 138:1327-28. [aFIIP]

Braitinan, D. J. (1984) Activity of neuroins in monkey posterior teimporal Chiang, J. J. & Julesz, B. (1984) Coopserative pheuuouneuia in apparentcortex duriing mnultidiinensioiial visual discrimnation tasks. Brain niovineist perception of randuim-dut Linvisatugrains. Visioni hlesarsisResearch 307:17-28. [0FII11 11:1781-88. oJRB, rFIIP]

Brandt, H1. F. (1945) The psychology ofseeimig. Phiilosopivsl Libsrary. JNIPBJ Chiedris, F., LebIarunv, NI & Lherunitte, F k1973, Visual searching in iiorimalBraiidt, T., Wist, E. B. & Dicligaiis, j. k1975) Foreground aind baivgrounsl in aud bran-dainiagcd subject,% kcoutribution to the study of unilateral

dynaniic spatial orientation. Perceptiton & Psychophysics 17:4197- iniattentioni). Cortex 9.941-111. [arFIIPI503. larFIlPI Clarke, P1. C. II. & Whitteridge, D. (1976) Thie projection of the retina,

Braniian, J, It. & Camp, C. J. t.1987) \ isual processing correlates of adult ilduuig the "red area" onto the optiL tectuiis of the pigeon Quarterlycognitive developsment. P'aper presented at the TiursI Beyond Formal Journal of Experimental Physiology 61.351-58 [MACI


References/ Previc: Visual field specialization

Cocito, L., Favale, E. 6c Tartaglione, A. (1977) Asinsmetrie funzionali tra Duffy, C. J. (1981) The legacy of association cortex. Neurology 34.192-ernicampo visivo superioro ed inferiore isel soggetto nornislc. Bollettino 97. IMKIDella Societa lialiana di Biologia Spcrimcntale 53:629-33. IaFIIPI Duiffy, C. J. & Wiirtz, R, IL. (1989) Radial and axial dlirectionality in cortical

Cogan, A. 1. (1979) Thec relationship between the apparent vertical and the visual arca MST. Society for Neurosciece Abstracts 15. 119. [M K.vertical horoptcr. Vision Research 19:655-65. [aFIIPI Rms]

Coltheart. M. (1983) Ecological necessity of Iconic memiory. Behavioral and Eckliorn, It., Blauer, R.. Jordan, WV., Brosch. MI. Kruise, WX.. Muink. MI. &Brain Sciences 6:17-18, (0PIIP] Reitbsseck, 11. J. (1988) Cohsereint oscillatios;: A hclanisis of featuire

Conisolly, MI. & Van Essein, D. C. (1984) The representation of the visual lining iii the visual cortex. Biological Cyberuictics 60J121-30,field in parvicellsilar ausd isagnocelhuilar layers of the lateralgenlesilate JGXVSJinucleus fin the issacaqsse msonkey. Journsal of Comnparative Neurology Eiclsenblasnn, H-., Wiener, S. I., Shsapiro, M. L. & Cohen, N. .1. (1989) Thle226:544-64. (aFHPI organization of spatial coding fin the liippocamsps; A stud~y of icural

Cowey, A. (1981) Whly are there so nsa-iy visual areas? In., organization of die enseible activity. Journal of Ncuroscicncc 9.2761-75. [GXVSIcerebral cortex, ed. K. 0. Schmitt. MIIT Press. IRNIRI Eisssrton, J. (1983a) Functionsal msorphiology of the vissial system. III:

Critelsley. NI. (1953) The parictai lobes. Hafner. [aFIIP, RNIB, BNISI Physiology aiid behavior of the pigeon, el. NI. Abs. AcadsemsicCrowne, D. P., Dawson, K. A. 6c Richsardsons, C. NM. (1989) Uinilateral Press. [INIAG]

periarcuate and posterior parietal lesions impair coisditionial position (1983b) Visioni. In. Physiology ansd beliatior of tie pigeoni, ed. NI. Abs.discrimination learning in the mnkey. Ncuropsyclsologia 27:1119- Acadenmic Press. INIAG]27. (rFIIP) Engel, r. L. (1971) Visual conspicsiity, directed atcistion aid retiinal locus.

Cummiisgs, J. L., Landis, T. & Benson, D. r. (1983) Enviroinental Visioni Research 11:5W3-76. [aFIIPIdisorucitation: Clinical and radiological findiisgs. Neurology 33.103- Enrighit, J. T. (1989) Coinvergeince dsring humas vertical sacca des. Probable4. (aFHPJ causes arid perceptual consequeinces. Journal of Physiology .I10..15.

Cyander, M. & Begais, D. (1982) Neurons in cat visual cortex tuined to the 65. [rFlIPIdirectioin of motion fin depth: Effect of positional disparity. Visioni Eisrotls-Cugell, C. & Robson, J. G. (1966) Theo coatrast seissitivity of retiisalResearch 22:967-82. (0F1IP1 gansgl ion ells of the cat. Journal of Physiology 187.517-52. [aFUIP.

Damasio. A. R1. & Bcnton, A. L. (1979) lInpairmeint of hand movemeints RMIunsdcr visual guidance. Neurology 29:170-78. (sF111'] Ericlsei, J. T., Ilodos, WV. R., Eviriger, C., Bessette, B. B. & Phillips, S. J.

Damaaio, A. B. & Damasio, H-. (1983) Trhe aisatoinic basis of pure alexis. (1989) Hlead orientation iii Iigeelusz Postural, locomotor, and visualNeurology 33:1573-83. (aFI1PI determinants. Brain, Behavior andl Evolution 33:268-78. [IAG)

Damasio, A. R., Dansasio, If. & van Iloesen. G. WV. (1982) Prosopagnosia. Eviatar, L., Miraisda, S., Eviatar, A., Freeisani, K. & Borkowski, NI. (1979)Ainatomic basis ansI behavioral mechanism,. Neurology 32 331- Development of nystagusus iii respoinse to vestibular stimulation iii41. [aFIIP] infaists. Aiinials of Neurology 5:508-11. JaFIIP]

De Hanisler, K. S, (1978) Stereopsis and unilateral braiin disease. Feinberg, T. E., Pasik, T. & Pasik, P. (1978) Extragemiculostriate visioin InlInvestigative Opt halomnology & Visual Science 17:336-43. (aFHP] the moiskey VI. Visually guided accsurate reaching behavior. Brain

Delis, D. C., Robertusn, I. C. & Efron, R. (1986) Hemispheric specialization Research 152:422-28. (GJB)of memory for visual hierarchical stimuli. Neuropsychologia 24:205- Felleissan, D. J. & Van Essen, D. C. (1987) Receptive field properties of14. [aFIIP] neurons inl area V3 of issacaque monkey extrastriate cortex. Journal of

Deregowsk-i, J. B. (1989) Real space anid represented space: Croscultural Neurophysiology 37:8889-920. [arFIPjpserspectives. Behsavioral amid Braiii Sciences 12:51-119. [aFIIP] Felselon, B., Neill, R. A. & White, C. T. (1986) Evoked potentials to

Dcrrington, A NI , Krauisknipf, J &c Lensnie, 1P (1984) Chromatic mnechainisms dysamie random dot stereograms inl tipper, center aisd lower fields.in lateral geniculate nucleus of macaqsie. Journal of Physiology 357:241- Documenta Ophthalonniologica 63:151-56. [aFIIPJ65. [aFIIP] Findlay, J. NI. (1980) The visual stimulus for saccadic, eye imovemeints in

Derringion, A. NI. 6& Lennie, P. (1984) Spatial and temporal contrast hiumain observers. Perception 9:7-21. (JNIF, rFHIPJsensitivities of ismuroises in lateral geniculate nucleus of macaque. journal (1932) Global processing for sacche eye movements. Vision Researchof Physiology 357:219-40. (aFIIP] 22:1033-45 (JNIFJ

Desiiuone, R & Gross, C G (1979) Visual areas iii the temporal cortex of (1985) Eye movements and visual cognition. LAnnee Psychologique 85:101-thme mascaque. Brain Research 178:363-80. [CJB, NI 36. [JNIFI

Desimone, R. & Scheiin, S J (1987) Visual properties of iieurons in area V4 Findlay, J. NM. & Harris, L. R. (1984) Small saccades to double-steppedof the macaque Sensitivity to stimulus form. Journmal of Neutrophmysiology targets msoving in two dimensions. Iii. Theoretical and applied aspects of

57:35-8. aFIP, KIeye iso' entent research, ed. A. G. Gale & F. Johiisons. Northliollaisd.Desiisone, II., Scheins, S. J., Moran, J. & Ungerleider. L. G. (1985) Contour, (NIPB, rFlIP]

color and shape analysis beyond the stiate cortex. Vision Recsearch Fiiske, R. A. & Kosslyn, S. NI. (1980) NMental imagery acuity in the peripshiral25:441-52. [arFlIPI visual field. Josurnal of Experimewntal Psychology:, llunman Perception arid

DeYoe, E. A. & Van Essen, D. C. (1988) Conscurrent processinug streams in Performance 6:126-39. (aFIIPImonkey visual cortex. Trends in Nesirosciences 11.219-26 [aFhIPI Fischer, B. & Boch, R. (1981) Enhanced activation of neurons us preluiiate

Diamonsd, I T & flail, W. C (1969) Evoluition of neocortex. Science cortex before visually guided saccades of trained rhesus monkeys.164:251-62. [aFHP, BWWJ kExperimnenital Brain Research 414:129-37. [aFIIP]

Dichgans, J & Bransdt, T (1978) Visuisl-vestibular interactions. Effects on self- Fischsr, B. & Baoch, R. (1985) Peripheral atteintion versus ceistral fixation.motion perception and psostural control In. Hlanmdbook of Sensory Modulation of the visualactivity of prelumate cortical cells of the rliesusPhysiology V111, Perception, ed. R. field, 11, Leibowitz & I-I.-L. Teuber. monkey. Brains Research 345:111-23. [sF111']Spsriisger.Verlag (BOB, rFIIPJ Fischer, B. & Breitmeyer, B. (1987) Mechaissis of visual attention revealed

DiZio, P 6c Lackner, J R. (1989) Perceived self-motion elicited by postrotary by saccadic eye imovemeists. Neur , ychmologia 9,5.73-83. [rFIIPJhead tilts is a varyiing gravitoinertisl force backgrouisd. Percepsoss & Floms, NI. C. & Bedell, H. E. (1985) Identifying amblyopia usiing associatedPsychophysics 46:114-18, [aFlHP] conuditions, acuity, and nonscuity features. American journial of

Dolezl, H. (1982) Living in a world transfornned: Perceptual and Optomnetryj and Physiological Opstics 62:153-60. [aFHIPJperforsnatory adaptation to visual distortion. Academic Press. [NO. Fox. R. (1982) The effect of depths position ii stimulus interactin. In. Thiree-rFHP) dhimensional displays: Perceptual research and applications to military

Donovan, W J. (1978) Structure and function of thie pigeon visual system, systems, ed. D. J. Getty. National Academy of Sciences. [aFhlIllPhysiological Psychology 6:403-37. (MAC) Fox, R. & Pattersoii, 11. (1981) Depth separation ausd lateral interference.

Drager. U. C. & Hulsel, D. I1. (1976) Topography of visual aisd Perception & Psychophysics; 30:513-20 (aFIlI']sosnatosensery projections to mouse superior colliculus Journal of Friberg, L., Olsen, T. S., Roland, 1' E., Paulsoni, 0. B. & Lassen, N. A.Neurophysiology 39:91-101. (aFHP] (1985) Focal increase of bslood flow ini the cerebral cortex of inaii dsiriiig

Drehser, B., Fukada, Y. 6c Rodieck, R. %V. (1976) Identification, claszi."icatmon vestibular stimulation. Brain 108:609-23 (sF111']and anatomical segregation of cells with X-hike amid Y-like psrop~erties in Friedlander, NM. J., Stanford, L. II. & Sherman, S. MI. (1982) Effects oftlie lateral genicislate nucleus of old-world primates. Journal of msonociular deprivation ois the atructsire-functmon relations~hip of iiidividualPhysiology 258:433-52. (aFIIP ineuirons in the( cat's lateral geniculate inucleus. Journal of Neuroscience

Dubner, Bt. & Zeki, S NI. (1971) Response properties aisd receptive elds of 2:321-30. [aFHiP)cells in an anatomically definied region of the superior tempora. sulcus inl Frisby, J. P. (1980) Seeiiig, illussiosn, brain and mnindu, Oxford Unuversitythme inikey. Brain hRsearch 35:528-32. (NI Press. (BXW]J


llcferencesl1'revic: Visuial field specialization

Fuksida, T. (1959) Vertical writing with eyes covered, A new test of sestilsulo. smooth pursuit cyc movements in the rhesus monkey frontal cye fieldsspinal reaction. Acta Oto-larnygologica 50.26-36, [aFIIPI Society for Neuroscience Abstracts 15:1203. [CJBJ

Fuster. J. MN. & Jervcy, J. P. (1982) Neuronal firing in the infcrotcmoporal Graboivaka, A. (1983) Lateral differences in thc detection of stereoscopiccortex of thc nmonkey in a % isual mnemory task. Journial of Neuroscrcncc dcpth. Neuropsychologia 21:2,19-57. (aFIIPI2:361-75. [aFIIPI Craves, J1. A. & Coodale, MI. A. (1979) Do training conditions affect

Calaburda, A. NM., Sherman, C. F., Rosens, C. D.. Aboitiz, F. & GCschsriuid, interocular transfer in thie pigeon? lin. Structure and function of theN. (1985) Developmental dyslexia. Four consecutive cases with cortical cerebral commuissurcs, edl. L. Steele Russell, MI. %an I of & C.anomsalies. Annals of Neurology 18:222-33, [aFIIPI Berluccehi. Macmillan. [MAGI

Califr-et, Y. (1968) Les diverse aires foanctionelles de la retine dii pigeon. Gray, CM\., Koening. P., Engel, A. K. & Singer, WV. (1989) OscillatoryZeiischirifl furzellforschnntg 86-535-415. [MAGI responsses in cat visual cortex exhibit inter-eolumnar synchironizatioi

Garraghty, P. E, (1985) Mixed cells in the cat lateral gesticulate nucleus. which reflects global stimulius properties. Nature 338.33.-37. ICGVS1Functional convergence or error in development? Brain, Bchastior and Gray, C, M. & Singer, W. k1l989) Stimtulus-specific nesironial oscillptions usEvolution 26 58-61. [aFIIP1 orientation columns of cat visual cortex. Proceedings of thec National

Cattass, R. & Cross, C. C. (1981) Visual topsographsy of striate projection zoine Acadeny of Sciences. USA 86:1698-1702. [CWSI(NIT) lin posterior superior temporal sulcus of the inacaque. Jourtial of Gregory, 1B. L. k1977) % ision wvithi isolsiisiiaist colour contrast. 1. A psrojectionNeurophysiology 46.621--38. [aFlIPJ technique and obsservations. Perception & Psychophysics 6:113-

Gawryszewski. L. C., Riggio, L., Ilizzolatti, C. & Uniiita. C. (1987) 19. [aFlIPIMoveinits of atteiitioii ii the three spsatial diimensioins aiis tile meaning Cross, C. C., Bender. D. B3. & Gersteiin, C. L. (1979) Activity of iinferiorof "isetitrail" ces. Ncuropsyelologia 25.19-29 1aFIIPI temporal neuronss in behuavinig mnikeys. Neuropsychologia 17.215-

Cazzaisiga, NI. S. & Lsulavas, E. (1987) Disturssiuces in spsatial attent'iomn 29. [aFIIP]foallowing lesion or disconinection of the right isarietal lobe. lit. Gross, C. C., Bender. D. B. & Rloclus-Miranda, C. E. (19741) lInfe'r'.enporalNeurophysiological anid neiiropsycliological aspects of spatial neglect, ed. cortex. A siiigle-uiit aiialysis. lIn. The neuroscienices. Third studyNI. Jeaisierodl. North-llollasd. (aFlI 5I programs, ed. F. 0. Schmitt & F. C. WVorden. MITl Press. [aFHPI

Claim. 11-f. & Einsas, P. If. (1988) Global aius local pirocesig by 3- aisd 4- Cross. C. C.. Cosvey. A. & Maniniig, F. J. (1971) Further analysis of visualnioiith-old iinfants. Perception & Psychophysics 13.165-71. IaFHP] discriiniation deficits followinig foveal Isrestriate and hiferotemporal

Gibsoin. J. J. (1958) Visually cointrolled locomotiomi aiid visual orieistatioin in lesions is, rhesus monkeys. Journal of Comparative and Physiologicalanimals. British Journal of Psychology -19:182-91. [BCB1 Psychology 76:1-7. [aFIIPI

(1966) The senses considered as perceptual systems. Hloughtomn. Cross, C. C.. Desimnoise, B.. Albrigoht, T. D. & Schwartz, E. L. (1985)Niflin. [aFIIPI Iniferior temporal cortex and patiete recognition. lit: Patterni recognition

(1979) The ecological approach to Uisual perception. Lswreince Erlbasiin iechaniisns, ed. C. Chiagas, It. Cattass & C. Cross. Spsringer-Associates. [IKI Verlag. [aFll1PI

Ginsburg, A. P. (1986) Spatial filtering and visual form perception. lit. Cross, U. C. & Mishkiin, M. (1977) Ilse neural basis of stimulus equivalencelHand!book of perceptioni and performnace, vol. 2, ed. K. R. llssf, L.. across retinal translatiomi. lit. [ateralizatioi in the nervous system. cad. &.K'aufinan & J. P. Thomas. Wiley & Sonss. [aFIP) Ilarnad, R Doty, J. Jaynes, L. Goldstein & C. Krauthsamer. Academic

Girotti, F., Milansese, C., Casazza, NM., Allegrsiiza, A., Corridori, F. & Press. [CINIIAvaiszini, C. (1982) Oculoinotor disturbances in Balint's syndrome. Cross, C. C., Roclia.Niranda, C. E. & Bender, D, B. (1972) VisualAiiatomoclinical findiiigs aiid electroocilographic analysis in a case. properties of ineurons in inferoteinporal cortex of the macaque. Journal ofCortex 18:603-14. [aFHPlJ Neurophysiology 35:96-111. [CJNI. rFHIPI

Cloiuing, K. & Quatember, It, k966) Nlethodiseluer hseitrag ziur unterauchung G~russer, O..J. (1983) Niultimodal structure of the extraperaonal space. In.der prosopagncesie. Neuropsychologma -1:133-14I1. [aFIIPI Spatially oriented behavior, ed. A. Hein & NM. Jeamsierod. Springer.

Cmsadt, j, W. & Ainderseii, 11. A. (1988) Memiory related planiniimg activity iii Verlag. [rFhlPjthe posterior parietal cortex of mnacaque. Expersuental Brain Research Cuillery, B. W. (1979) A speculative essay ois geniculate lamination and its70:216-20. [RBB development. In; Progress in braini research: Deelopmenmt amid chemmical

Giiadt, J. W. & NMays, L. E. (1989) Posterior psarietal cortex. The oculomotor specificity of neurons, vol. 51, ed. MI. Cuciod, C. WV. Kreutzberg 6& F.isear response and spatial coding in 3-D space. Society for Neurosciencee E. Bloom. ElsevmeriNorthi.Hollaiid Biomedical Press. [aFHIPIAbstracts 15:786. [ISI Guitton, D., Douglas. It. \I. & Volle, NM. (1984) Eye-liead coordination in

Codoy. J., Luiders, H., Diisier, D. S., NMorris, If. H. & WVyllie, E. (1990) cats. Journal of Neurophmysiology 52.1030-50. (aFHPIVersive eye movemeiit elicited by cortical stimulation of the hsunman l-aaxma, R. &c Kuiypers, 1H. C. J. NI. (1975) listralucinispheric corticalbrain. Neurology 410.296-99. [rFIIP] connexions and visual guidaiice oflhand and finger movemeists in the

Goldberg, NM. E. & Robiissons, D. L. (1978) Visual system: Superior collicilsis. rhesus inikey. Braini 98:239-60. [aFHIPJ1mm. Haond!book of behavioral iieiirobiology, vol. 1, edl. R. B. Ntastertois. Hiaber, R. N. 6c Herslsenson, NI. (1973) The psychology of visual perception.Plenum. [aFHPI Holt, Riinehart aisd Wimnston. [aFIIP]

Goldenberg, C., Podreka, I., Steiiier, NI., WVmllmes, K., Suess, E. & Deecke, Haclkian, B. B. (1910) Au experimental study of variability in ocular latency.L. (1989) Regional cerebral blood flow patternts fi visual imagery. Joiirinal of Experinital Psychology 27:546-48.Neuropsychologia 27:611-%61. [arFIIPI Haenany, P. E. & Schiiller, P. H. (1988) State dependeint activity in monkey

ColdmnsasRakic, P. S. (1987) Circuitry of primate prefrontal cortex asid visual cortex 1. Siingle cell activity in VI and V4 oil visual tasks.regulation of behavior by representational memory. lin: Handbook of Experimental Braiii Research 69:225-441. [aFIIPIphysiology, ,ol. 5, ed. V. Moutcastle, F. Plium & S. It. Geiger. Hall, L. C. (1985) Searching high aisd low. The developmnt of efficient visualAmserican Physiological Society. [aFIIPI search iii hearing, deaf and learniisg disabled children. In: Eye

(1988) Topography of cogiiitioii. Parallel distributed networks iii primate inacerwisets and humnan informmatomi processinug, edl. B. Groner, C. WV.association cortex. Annual Rleview of Neuroscience 11:137-56. [CWVS1 MeConkie & C. Nienz. Northi-lolland. [NIPB, rFHP1

Coldstein, L. HI., Canavan, A. C. NI. & Polkey. C. E. (1989) Cognitive Hlalligans, P. W. & NMarshiall, J. C. (1989) Is neglect (only) lateral? A quadrantmapping after unilateral temporal lobectomy. Neiiropsychologia 27.167- aisalysis of line caiicellatioin, Journal of Cliiiical aiid Expetmnital77. [aFEIP] Neiiropsychology 11:793-98. [rFIIPJ

Coodale, NI. A. (1983a) Visuomotor organization of pecking in the pigeon in. Hamasaki, D. 1. & Flynn, J. T. (1981) Amblyopic eyes have longer reactionAdvances ii vertebrate neuroethology, edl. J,.-P. Ewert, It. B. Capranica times. Investigative Ophthlomology & Visual Science 21:846-&c D, J. Ingle. Plenum. [MAC] 53. [aFIIP]

(1983b) Visually guided peckiiig in the pigeon (Columnba livto). Braiii, Hlansoii, B. MI. & Skavenski, A. A. (1977) Accuracy of eye positionBehavior, and Evolution 22:22-41. [MACI information for motor control. Vision Rlesearch 17.919-26. [BAA]

(193c) Vision as a sensorimotor function. In, Behavioral approaches to Harris, C. S. (1965) Perceptual adaptation to inverted, reversed, andbrain research, ed. T. E. Robison. Oxford University Press. [NIACI displaced vision. Psychological Rleviewo 72.419-441. [B'sWW

(1988) NModularity in visuomotor coistrol. Front iiiput to output. lIn. Hlarter, Ni. B. & Aine, C. J. (1984) Brain mechiaiisms of visual selectiveComputational processes in human vision. Aut interdisciplinary atteintioin. In. Varieties of attentioui, ed. It. Parasuraisan &c D. R. Davies.perspective, edl. Z. Pyhyshmyn. Ablex. [NMAGI Academic Press. [rFIIPI

Coodale, NI A. & Craves, J. A. (1982) Interocular trainsfer in the pigeon. Ilarwerth, B. S. & Boltz, 11. L. (1979) Stereousis ii inikeys u.Sing random-Retinal locus as a factor. In. Anialysis of visual behiis~or, cad. D. J. Ingle, dot stereograms. The effect of viewiing duratioin. Vision Research 19:985-NI. A. Goodale & It. J. WV. Maiisfield, NIT P'ress. (MACI 91. (aFIIPI

Gottlieb, J. P., NfacAvoy, NI. C. & Bruce, C. J (1989) Unit ,activity related to Hassler, It. (1966) Compjarative aniatomy of the central visual system of day-


ReferenceslPrevic: Visual field specialization

and night-active primates. In. Evolution of the forebrain, ed. R. lassler Jeannerod, M. & Biguer. B. (1987) The directional coding of reaching& H. Stephan. ThIicme. [RWWI movements: A visuoinotor conception ofspatial neglect. In:

Ilebb, D. O. (19419) The organization of behat.ior. Wile & Sons. 1BGB] Neurophyssological and neuropsychological aspects of spatial neglect, ed.Ilecaen, II. & Albert, NI. L. (1978) Ihnani ncuropsychology. Wiley & M. Jeannerod. North-Holland. [aFIIPI

Sons. [aFIIPI Jeannerod, M., Gerin, P1. & Penier, J. (1968) Deplacements et fixations duleilman, K. M. & Aalenstein, E. (1979) Mechinisms underlying hemispatial regard dans I'exploration hbre d'une scene visuelle. Vision Research

neglect. Annals of Neurology 5:166-70. [IMIII 8:81-97. [aFIIPIHilnan, K. M. & van Den Abell, T. (1980) Right hemisphere dominance for Jeeves, ,IL A.. Milner, A D.. Perrett, D. 1. & Smith. P. A. J. (1983) Visual

attention. The mechanism underlying hemispheric asymmetries of cells responsive to direction of movement and stimulus form in theInattention (neglect), Neurology 30.327-30. [aFHPI anterior superior temporal sulcus of the macaqIne monky. Journal of

Heilman, K. M.. Watson, R. T, & Valenstein, E. (198) Neglect and related Physiology 341:80P. (aFIIPIdisorders. In Clinical neuropsychology. 2d d,., ed. K. M. Heilnan & E. Jensen, i. & Secdorff, Ii. Il. (1976) Temporal lobe epilepsy and neuro.Valenstein. Oxford University Press. (KMII ophthalmology: Ophthalmological findings in 74 temporal lobe resected

Hendrickson, A. E. (1985) Dots, stripes and columns in monkey visual cortex, patients. Acta Oplithalnologica 5,:827-11. [aFHPJTrends in Neurosciences 8:406-10. [aFHIPI Jones, R. & Kerr, K. E. (1972) Vergence eye movements to pairs of disparity

Hess. R. F. & Bradley, A. (1980) Comtrast perception above threshold is only stimuli with shape selection cues. Vision Research 12.1425-30. (aFIIPI

minimally impaired in human amblyopia. Nature 287;463-61. [aFHP] Jones-Gotman, NI. (1986) Memory for designs. The hippocampal contribution.

llcuer, H., Dunkel-Abels, C., Bruwer, M., Kroger, II., Romer, T. & Neuropsychologia 21:193-203. (aFIIPIWischmeier, E (1988) The effects of sustained vertical gaze deviation on Julesz, B. (1978) Giobal stercopsis. Cooperative phenomena in stereoscopic

the resting state of the vergence system, Vision Research 28:1337- depth perception. Iu: Handbook of sensory physiology, vol. 8, ed. R.

44. [arFIlPI leld, II. W. Leibowitz & H..L. Teuber. Springer.Vrlag. (aFIIPI

leuer, H. & Owens, D. A. (1989) Vertical gaze direction and the resting Julesz, B., Breitmeyer. B. & Kropfl, W. (1976) Binoular.disparitydependent

posture of the eyes. Perception 18:363-77. [rFHPI upper-lower hermifield anisotropy and left-right hemifield isotropy as

leywood, S. & Churcher, J. (1980) Structure of the visual array and saccadic revealed by dynamic random-dot stereograms. Perception 5:129-

latency: Implications for oculomotor control. Quarterly Journal of 41. [aFHPIExperimental Psychology 32:335-41, (arFlIP, MPB, JMFI Kaas, J. It. (1989) Changing concepts of visual cortex organization in primates.

Hickey, T, L. (1977) Postnatal development of the human lateral geniculate In. Neuropsychology of visual perception, ed. J. W. Brown. Lawrencenucleus: Relationship to a critical period for the visual system. Science Erlbaum Associates. [IMKI198:836-38. [aFtlPI Kaplan, E. & Shapley. R. NI. (1982) X and Y cells in the lateral geniculate

Hilz, t., Rentschler, I. & Brettel, H. (1977) Myopic and strabismic nucleus of macaque monkeys, Journal of Physiology 330:125-amblyopia: Substantial differences in human visual development. 43. [aFllP]Experimental Brain Research 30.445-46. [aFHPI Karpov, B. A., Luria, A. R. & Yarbuss, A. L. (19r-8) Disturbances of the

Hine. T. & Thorn, F. (1987) Compensatory eye movements during active structure of active perception in lesions of the posterior and anterior

head rotation for near targets: Effects of imagination, rapid head regions of the brain. Neuropsychologia 6:157-66. (arFHPIoscillation and vergence. Vision Research 27:1639-57. [aFHP] Kaufman, L. & Richards, W. (1969) Spontaneous fixation tendencies for visual

Hines, D., Jordan-Brown, L. & Juzwin, K. R. (1987) Hemispheric visual forms. Perception & Psychophysics 5:85-88. [aFHPIprocessing in face recognition. Brain and Cognition 6.91-100. [aFHPI Keating, E. C., Cooley, S. G. & Kenny, D. V. (1985) Impaired tracking and

Hinton, G. E., McClelland, J L. & Rumelliart, D. E. (1986) Distributed loss of predictive eye movements after removal of the frontal eye fields.representations. In: Parallel distributed processing: Explorations in the Society for Neuroscience Abstracts 11:472. [CJBImicrostructure of cognition, ed. D. E. Rumelhart & J. L. McClelland, Kelly, D. H. (1977) Visual contrast sensitivity. Optica Acta 24:107-MIT Press, [RMBI 29. [arFtPI

Hedos, W. R. (1976) Vision and the visual system; A bird's-eye view, In: Kendrick, K. NI. & Baldwin, B. A. (1987) Cells in the temporal cortex ofProgress in psychobiology and physiological psychology, ed J. Sprague & conscious sheep respond preferentially to the sight of faces. ScienceA. N. Epstein. Academic Press. [MAG 236:448-50. (RMB]

Holmes, C (1938) The cerebral integration of the ocular movements British Kesner, R. P., Farnsworth, G. & DiMattia, B. V. (1989) Double dissociationMedical Jourual 2:107-12. (aFlPI of egocentric and allocentric space following medial prefrontal and

Honda, H. (1985) Spatial localization in saccade and pursuit-eye-movement parictal cortex lesions in the rat. Behavioral Neuroscience 103:956-conditions: A comparison of perceptual and motor measures. Perception 61, [GWS]and Psychophysics 38:41-46, (JMFI King-Smith, P. E. & Ca-den, D. (1976) Luminance and opponent-colo,

Hong, X. & Regan, D. (1989) Visual field defects for unidirectional and contributions to visual detection and adaptation and to temporal andoscillatory motion in depth. Vision Research 29.809-16. (aFHPI spatial integration. Journal of the Optical Society of Anerica 66:709-

Horak, F. B., Shumway-Cook, A., Crowe, T. K. & Black, F. 0. (1988) 17. [aFIP]Vestibular function and motor proficiency ofchildren with impaired Kimsbouroe, M. (1972) Eye and head turning indicates cerebral lateralization.hearing, or with learning disability and motor impairments. Science 176:539-41. (aFHPIDevelopmental Medicine aid Child Neurology 30.64-79. taFtPI Kinsbourne, NI, & Hicks, R. E. (1978) Functional cerebral space. A model for

Horel, J, A. (1978) The neuroanatomy of amnesia, A critique of the overflow, transfer and interference effects in human performane. In:hippocampal memory hypothesis. Brain 101:403-45. (aFHPI Attention and performance VII, ed. J. Requin. Lawrence Erlbaum

Hoyt, C. S. (1982) Abnormalities of the vestibulo.ocular response in Associates. INIKIcongenital esotropia. American Journal of Ophthalomology 93.703- Kinsbourne, M. & Warrmngton, E. K. (1962) A disorder of simultaneous form8. (aFHPI perception. Brain 85:461-86, [aFIIPI

Hughes, A (1977) The topography of vision in mammals of contrasting Komatsu, H. & Wurtz, R. H. (1988) Relation of cortical areas MT and MST tolifestyles. Comparative optics and retinal organisation. In Handbook of pursuit eye movements. i. Localization and visual properties of neurons.sensory physiology, vol. VII/5, ed. F. Crescitelli. Springer. Journal of Neurophysiology 60.580-603. [MK, rFHPIVerlag. ["WW] Komatsu, H., Roy, J. P. & Wurtz, R. H. (1988) Binocular disparity sensitivity

Hughes, Ht C & Zimba, L D (1987) Natural boundaries for the spatial of cells in area MST of the monkey. Society for Neuroscience Abstractsspread of directed visual attention. Neuropsychologia 25:5-18. [arFHPi 14:202. [aFHP, MKJ

Hurvich, L. M. (1981) Color vision. Sinauer Associates. (aFItPI Kooistra, C. A. & Heilman, K. NI. (1989) ilemispatial visual inattentionHylkema, B. S. (1942) Examination of the visual field by determining the masquerading as hemianopua. Neurology 39l1125-27. [arFHP, CMB]

fusion frequency. Acta Ophthalmologica 20.181-93. [aFHPI Kriss, A. & Halliday, A. M. (1980) A comparison of oceipital potentials evokedHyvarinen, J (1982) The parietal cortex of monkey and man Springer- by pattern onset, offset and reversal by movement. In. Evoked potentials,

Verlag. [arFHPJ ed, C. Barber. MTP Press. [aFHPIIkeda, M. & Takeuchi, T (1975) Influence of foveal load on the functional Krose, B. J. A. & Jslesz, B (1988) Evidence against rapid voluntary shifts of

visual field, Perception & Psychophysics 18:255-60. [aFHP] attention to non-cued target locations in a visual search task. InvestigativeJampel, R. S. (1959) Representation of the near-response on the cerebral Ophthalmology and Visual Science 29.400, jJRBI

cortex of the macaque. American Journal of Ophthalmology 48:573- (1989) The control and speed of shifts of attention. Vision Research82. (aFIPI 29:1607-19. (JAB, rFHPI

Jeannerod, M , ed (1987) Neurophysiological and neuropsychological aspects Lanman, J., Bizzi, E. & Allum, J. (1978) The coordination of eye and headof spatial neglect. Elsevier, (RMB] movement during smooth pursuit. Brain Research 153:39-53. [aFIP]



Larrabee, 0. J., Leviir, H. S., Huff, F. J., Kay, M. C & Cuirsto, F. C., Jr. and to samcdcs in humans: A clinical study. Acta Otolaryngologica(1985 VtrrL( agisisia contrasted wvith visoal.vcrbal disconnection. 101-361-43. [aFHP)Neuropsychologla 23:1-12. (aFIIP] Malpell, J. P. & Baler, 1. IL. (1975) The representation of the visual field in

Laley, D. J., Kivirn, J., Rich, L. & Flynn, J. T. (1984) Stereo-discrimrnation the lateral gensculate nucleus of Mfacaca rsulat ta. Jon raol of Comsparatiebetween diplopic images in clinically normal observers. Investigative Neurelogy 161:569-94. (aFHIPJOphrthalmsology & Visssa! Science 25:1316-20. [sF1111 Manning, F. J. (1971) 'Me selective attention "deficit" of monkeys with

Latta, R. & Cowey. A. (1971) Fixation changes after frontal eye-field lesions abslations of folveal prestriate cortex. 's yclsonornic Science 25:291-in nmonkeys. Brain Research 30.25-36, IaFIIPI 92. [aFIHPI

Legge, C. E., Pelli, D. C., Rubin, C. S. & Schleske, M, M. (1985) Manning, M. L., Finlay. D. C., Neill, R. A. & Frost, B. C. (1987) DetectionPsychophysics of reading 1. Noi mal vision. Vision Research 25.239- threshold differences to crossed and uncrossed disparities. Vision52. IJRRI Research 27:16&3-86, [aFHP, MPBI

Legge. C. E., Rubin, C. S. & Luebker, A. (1978) Psychophysics of reading Niarr, D. (1982) Vision. WV, 11. Freeman, ICJM, RWW1V. T7he role of contrast in normal vision. Vision Research 27.1165- Marrocco, R. T.. McClurkin, J. WV. & Young, R. A. (1982) Spatial sunmmation77. iJRB) and conduction latency classification of cells of the lateral genictilate

Leisoncn., L- Hyvarinen, J.. Nyman, C. & Linisankoski, 1. (1979) Functinal nutcleuisof nsacaqtres. Joumnafof Ncurosciessce 2.1275-91. [aFHPJproperties of neurons in lateral part of association area 7 in awake Martens, W., Blake, R., Sloanse. M. & Corrack, B. Hf. (1981) What masksnmonkeys, Part 1. Expenniwntal Brain Research 34:299-320. IKMIIJ utroesilar discrimination. Perception & Psychophysics 30:521-

Leinorren, L. & Nynman, C. (1979) Fuictional properties of cells iii 32. (aFIIPIanterolateral part of area 7 associative face area of awake monkeys, P'art 2. Martin, F. & Lovegrove, W,. (1984) Thei effects of field size and luiminaisce oilExperimntal Brol'zs sarch 3.1321-33. [KI~III contrast sensitivity differences between specifically reading disabled and

Levine. D. N., Warach, J. & Faral. NM. (1985) Two visual systems in mental normal children. Ncuropsychological 2207-77. (aFIIP]imagery Dissociation of ~wlat" .nd "where" in imagery disorders due ts' Martin, K. A. C. (1988) From enzymes to visual pserception. A bridge too far?bilateral posterior cerebral lesioins, Neurology 35:1010-18. (aFIIPJ Tren&L ili Neuroscinces 11:380-87. [aFIIPI

Levinson, H. (1988) Tuec cerebellar.vestibular basis of learning disabilities us Miartin, M. (1979) Hemispheric spsecialization for local and global processingchirldrein, adolesceints aisd adults. Hlypothiesis aisd study. Perceptual andc Quarterly Journral of Experiniesital Psychology 17,33-40, [MPB)Motor SIills 67:983-1006. (aFIIPI Niartinoya, C., Rey, J. & Bloch, S. (1981) Limits of the pigeon's biinocular

Levy.Sclioei, A. (1969) Deternuatioin et latence de la respsoisse oculointrice a field and direction for best binocular viewing Vision Research 21,1197-deux stimulus. L'Ann~c Psychologiquc 69:373-92. JiF, rFIIPI 200. [MAC)

Liederman, J, & Meehiai, P. (1986) When is betweei.heriiispliere division of Niatelli, M.. Olivieri, Ni. F., Saccani, A. & Rizzolatti, C. (1983) Upperlabor advantageous? Neuropsychologia 24:863-74. (NIPBI visual spsace neglect arid motor deficits after section oftire midbrain

Liedern, J., Niemla, J. \i. & Martinez, S. (1985) Interhemsispsheric coinmissures iii the cat. Belrarioral Braisn Research 10-263-85collaboration irs response to bilateral iinput. Neirropsycliologia 23:673- [sFl-IP]83. [NiPB1 Nattir, L., Niatiis, E. & Pearce, D. C. (1969) Visual percepstion ofidirection

Lisberger, S. G. & Pavelko, T. A. (1989) Topographic and directional when voluntary saccades occur 1. Relation of visual direction of a fixedorganization of visual motion inputs for the initiatioii of hiorizonstal and target extinguished before a saccade to a flash presented during avertical sinootlr-pursuit eye movements in monkeys. Journal oif saccade. Perception and Psychophysics 5 65-80. [BAA]Neurophytiology 61:173-85. [CJB, rFHPJ Nisunsell, J. H. R. (1987) Physiological evideince for two visual subsystems.

Livingstone, Ni. & Hulsel, D. (1988a) Segregation of form, color, imoveinit, In. Matters of lsrtellrgesce Coniceptsual -structures in cogiltiveand depth: Anatomy, physiology, aisd perception. Sciensce 240:740- necurosciesnce, ed. L. Ni. Vaina. Reidel Pusblislhing. [aFHP]49. [aFHIP, LNICJ Niunsell, J. H. B. & Newsome, WV. TF. (1987) Visual processing inl monkey

Livinsgstone, Ni. S. & Hlubel, D. H. (1988b) Do thre relative mapspinig extrastriate cortex, Asnnual Review of Neuroscience 10:363-401. larFlIPidensities of thre magno. aisd parvocellirlar systems vary with ecceintricity? Niatinsell, J. 11. R. &c Van Essen, D. C. (19S3a) Functional properties ofJoursial of Nesrroscience 8:4334-39. [aFH-PI neurosis in middle temporal visual area of thre isiacaque moinkey 1.

Logothetis, N. K. & Schiall, J. D. (1989) Neuronal correlates of subjective Selectivity for stimulus direction, speed, and orientation. Journal ofvisual perception. Science 245:761-63. [aFHP1 Neurophysiology 49:1127-47. [aFIIP, MK, BMS]

Longuet-lHiggins, H. C. & Prazdniy, K. kI980) The interpretations of a minrg (1983b) Fuisctional properties of neurons in middle temporal visual area ofretmr'al image. Proceedisigs of the Royal Society of Losndoni, series B the macaque msonkey 11. Binocular interactions and tire sensitivity to208:3b,1-97. filMS] binocular disparity. Journral of Neu rophyslology 49:1148-67. [aFHP,

Low, F. N, (1943) Ire peripheral visual acuity of 100 subjects. Aoier'cai NMK, RM~S]Joursnal of Physiology 140.83-88. [aFHP) (1983c) Tire connections of thre middle temporal visual area (NiT) and their

Lu, C. & Fender, D. 1-1. (1972) Ire interactiosn of color and luminance in relationship to a cortical hierarchy in the macaque monkey. Joursnal ofstereoscopic vision. Isnvestigative Ophthalmology 11:482-90. [aFHIPJ Neuroscienrce 312563-86. [RNIS]

Lurrdh, B. L., Lennerstrand, C. & Derefcldt, G. (1983) Ceintral aird (1987) The topographic organization of thre middle temporal visual area inperipheral normal contrast sensitivity for static ai5d dysianiri. sinusoidal the macaqure monkey. Representational biases and the relationship togratings. Adta Ophthasswlogica 61:171-82. [aFHIP) callosal connections arid mycloarchitectonic bsoundaries, Journal of

Lupp, U., flauske, C. & Wolf, WV. (1976) Perceptual latencies to sinusoidal Comrparative Neurology 266:55-55. [arFH-P]gratngs Vison eseach 6:96-72 fa~lPNiays, L. E. & Spsarks, D. L. (1980) Saccades are spatially, not retinotopically

Lura, A. R., Pravdisra-Vinarskaya, E. N. & Yarbiras, A. L (1963) Disorders coded. Scitnsie 208:1163-6. 1 RAA]of ocular movement in a case of srmultagnosia. llrvis 86.210-28. MeColgin, F. H. (1960) Niovemeint thresholds in peripheral vision. Journal of[aFHPJ the Optical Society of America 50.774-79. [aFHPI

Lynrch, J. G (980) The fulctiuil urgarn ization of postieriur ,,.rietad .issoc~lrriin MLDonnrcll, P'. M. k1975) The development of visually guided reaching

(1987) Frointal eye field lesions in mnikeys disrupt visual pursuit. McFic, J. & Zangwill, 0. L. (1960) Visual-coiistructive disabilities associatedExperimrenital Braiss Research 68.437-41. [CJB, rFIIP) with lesions ofthre left cerebral hemisphere. Brain 83:243-60. [aFlI P)

swnh,_I C . ionnrAutleo V B , Talut, WV 11 & Yin T, C, T. 119771 Meadows. J, C. 11974a) Disturbed [perceptioin of colours associated withParietal lobe mechanisms for directed visual attention. Joursnal of localised cerebral lesions. Braisn 97:615-32. [aFIIP]Neurophysiology 40.369-89. [aFIHP] (1974b) Thec anatomical basis of prosopagnosia. Joursnal of Neurology,

NiacAvoy, Ni. C. & Bruce, C. J. (1989) Ocsrlosssotor deficits associated with Neurosurgery, and Psychiatry 37.489-SO). [aFlHP]lessons of thre frontal eye field area of iii marcaque monkeys. Society for Mienzel, E W. (1973) Chrimpanuzee spatial memory organization. ScicsnceNeurrosciesnce Abstracts 15:1203. [CJB] 182:9,43-45. [aFHPJ

MacAvoy, Ni. C,, Brtuce, G. J. & Guttlieb, J. 11988) Smooth eye moveints Nierigars, W. 11. k1989) Chromatic arid achiromatic vision of mnacaques. Role ofelicited by inicrostiinulativss ii thre frontal eye fields region of alert tIe Pp~athiway.Jorirsial of Neuroscienrce 9:776-83. [aFIIPImacaque monkeys, Society Neurroscicsice Abstracts 14.956. [CJBI Nierigari, WV. If & Eskin, T. A. (1986) Spatio-temporal vision of imacaques

Mack, A. (1978) Thiree modes; of visual perceptron. is. Modes of Percesin~sg with severe loss of PPI retinal ganiglion cells. Vision Research 26.1751-anid Processing Isiforsnrotion, ed. H1. L. Pick, Jr. & E. Saltzmn Wiley &61. [aFIIP]Sons. [CWS] Meyer. D. E., Abrams, It. A., Komnblum, S., Wright, C. E. & Smsithi, J. E.

Magnussoni, Ni., Pyykko, 1. & Norrvrrrg, B. 1,936) lire relartioniship of K. (1988) Optimsality in hruman motor performance. idcal control of rapidoptokinetre irystaginus to pursuit eye movements, vestibulair iiysagr,.,s 111ie0 mnovemnits Psychological Reiecn 95.340-70. [BAA]



Meyer, D. E , Smith, J. E K. & Wright, C. E. (1982) Models for the speed Navon, D. (1977) Forest before trees. Thc precedence of global features inand accuracy of aimed movemeints. Psychological Review 89:449- visual perception. Cognitive Psychology 9:353-83. [arFlIP, JRBI82. (RAA1 Newman, R. P., Kiinkel, WV. R. & Jacobs, L. (1984) Altitudinal hcmianopia

Michael, WV. F 6c Halliday, A. M (1971) Differences between thc occipital caused by occipital isfaretions. Clinical and computerized tomographicdistribution of upper and lower field pattern-evoked responses in man, correlationts. Arcliies of Neur-ology 41:413-18. (aFIIPIBrain Research 32:311-24. [aFIIP1 Newsome, WV. T. & %Vurtz, R. 11. (1988) Probing visual cortical function with

Miles, F A. & Kawano, K. (1987) Visual stabilization of the eyes. Trends s discrete chemsical lesions. Trends its Neurosciesicca 11.3941-400. [arFHPl'Neiirosciessces 10:153-58. [aFHP] Niederlasdova, Z. & Litvisieisova, V. (1973) Stabilograpsic finsdiisgs in

Miles, P. WV. (1954) Ass aiialysis oftdejsth factors ii sssisojsia and strabismus. Aggrcssologie 1413:33-36. [aFHPJassisodoininaisce. Americani Journal of Ophithalmology 37.98- Norton, T. T. & Casagrande, V. A. (1982) Laminar organizations of recepstive-106. [aFHP] field psroperties in lateral geniculate nucleus oflbush baby (Calago,

Miles, WV. R1 (1936) The reaction time of the eye. Psychological Mounographs crassicaudatss). Journal of Netsrophsysiology 47.715-4 1. 1aFIIP]47.268-93. (rFIIPI Nye, P. %V. (1973) Ois the functioinal differences betwceis frontal and lasteral

Miller, j. M., Oiso, 11 & Steinbach, NI. J. (1980) Additivity of fusioisal visual fields of t' e pigeon. Visioni Research 13 559-74. [MIACIvergence and pursuit eye movements. Vision Research 20.43- O'Reilly, RI. C., Kosslyn, S. M., Marsolek, C. J. & Cliabris, C. F. (is press)47. [aFIIP] Receptive field characteristics that allow parietal lobe ineurons to eiscodc

Miller, L, K. (1969) Eye-snovensent latency as a function of ago, stimulus spsatial properties of v-sual input. A compjutational analysis Journal ofuncertainty, amid position fi the visual field. Perceptual & Motor Skills Cognitive Nettroscicssce. [CJMJ28:631-36. [JNIF, rFHP] Orssitz, E. \I. (1970) Vestibular dysfunction is schizophrenia and childhood

Millodot, M. 6c Lsmont, A. (1974) Peripheral visual acuity its the vertical autism. Comsprehiensive Psychiatry 11.159-73. [aFI-PIplane. Visioni Research 14:1497-98. [0FIIP] Osaka, N. (1976) Reaction time as a funsction of peripheral retinsal locus arounid

Mishkin, NM. (1972) Cortical visual areas and their iisteractions. The brain fovea. Effect of stimulus size. Perceptual asid Motor Skills 43.603-and human behavior, ed. A. Karczmar & J. C. Eccles. Springer- 6. [NO)Verlag. [aFHPI (1978) Naso-teriporal difference, in huisan reaction tinme in tlso peripheral

Mishkin, M. & Forgays, D. C, (1952) Word recognition as functioms of retinal visual field. Neuropsychologia 16:229-303. [NO)locus, Journal of Experinuesital Psychology 43.43-48. [NIPB, rFIIP] Osmusan Hill, %V. C. (1972) Evolutionary biology of the primnates. Academic

Mitchell, D. E. &c Rlakemore, C. (1970) Binocular depth psercepstions nd the Press. [sFH1P1corpus callosum. Vision Research 10:49-51. [sFH1P1 Osterberg, C. (1935) Topography of the layer of rods and cones in the humais

Miyashita, Y. (1988) Neuronal correlate of visual associative loisg-terms retiusa. Adta Oplsthasnologica 6:1-102. (Supplement) [arFHlP1nmemuory in the psrimsate temporal cortex. Nature 335 817-20. Ottes, F. P., vais Cisbergen, J. A. MI. & Eggermost, J. J. (1985) Latency[aFIIP] depenidensce of colousr-lased target v!. nosntarget information by the

Mohler, C. WV. & Wnrtz, R. I-I (1977) Role of striate cortex aiid supierior saccadic system. Vision Research 25:849-62. [JMFJcolliculus in visual guidaisce of saccadic eye inovemeists in monkeys. Paliard, J. (1982) Tuie coistribution of periphecral and central visios to visuallyJournal of Neurophysiology 40:74-94. [CIII] guided reaciig. it: Analysis of Ususal behavior, ed. D. j. Ingle, NM. A.

Moran, J. &c Desimoie, R. (1985) Selective attention gates visual processing Coodale & R. J. WV. Nasssficld. MIT Press. [aFHP, RNIBJin the extrastriate cortex. Sciensce 229:782-84. [aFHPJ Pandya, D. N. & Seltzer, B. (1980) Cortical consnectioiss amid the functioiial

NMorris, R., Mickel, S , Brooks, \I., Swavely, S & Hleilman, K. (1986) organization of posterior porietal cortex. Behavioral and Brains ScienaesRecovery from ineglect. Journal of Clinitcal assd Experimental 3(4):511-13. [aFH-P]Nessropsychology 7:609. (aFHlP] Parker, D. NI. & Salzen, E. A. (1977) Latency changes in the human visual

Nlotter, B. C. & Ntouistcastle, V B. (1981) The functioiial properties of the evoked respoinse to sinusidal gratings. Visions Research 17.1201-light-sensitive neurosis of the posterior parietal cortex studied iii waking 4. IaFHPJmonkeys. Foveal sparing aiud opponent vector organization. Jouirnal of Pasik, P., Pasik, T. & Renider, NI. B. (1969) The pretectal syndrome inNeuroscience 1:3-26. [NMK, CJNI] monkeys 1. Disturbanices of gaze and body posture. Braini 92-521-

NMotter, B. C., Steinmetz, NI. A., Duiffy, C. J. & NlousuteastlL, V. B. (1987) 34. [aFHP]Functional psropserties of parietal visual neurosis: NMechianisms of Pavlidis, C. T. (1981) Do eye movements hold the key to dyslexia?directionality along a single axis. Journal of Neuroscience 7:151- Neuropsychologia 19:57-61. [aFH-P)76. [aFIIP, NIK, CJNI] Payine, %V. HI. (1967) Visual reaction times on a circle about the fovea. Science

Ntouistcastle, V. B. (1976) The world arouisd us: Neural command fuinctioims 155:481-82. [aFIIP)for selective attention Nemirosctesices Research Program Bulletin 14.1- Penfield, WV. (1957) Vestibular sensatios and the cerebral cortex .Annals of47. [aFHP] Otology, Rhinology and Laryngology 66.691-93. [aFIIP1

(1979) Aut organizing principle for cerebral function. The utnit module aiud Penfield, %V. & Rasmussen, T. (1950) The cerebral cortex of inan.the distributed system. lIn: The nessrosciencez fourth study program, ed. Mac~tillan. [aFH-PIF. 0. Schmitt & F. C XWordets. NMIT Press. [CWVS] Pensual, B. E. (1977) Hluman erebral asymmetry in color discrimination.

Nlounteastle, V. B., Nlottcr, r 7., Steinsmetz, NI. A. & Duffy, C. J. (1984) Neuropsychologia 15:563-68. [aFHP]Looking and seeiisg. The visual functions of the parietal lobe. fIt. Perenin, NI. T. & Vighetto, A. (1983) Optic atxia. A spsecific disorder inDynamic aspcts of sseocortical fitnctiomi, ed C NI. Edeltman, %V. E. Call visuomnotor coordination. In. Spatially oriented behiavior, ed. A. Hemn &c& WV. NI. Cowan. Wiley 6& Sonss. [CJB, rFIIPJ NI. Jesisierod. Spriiuger.Verlag. [aFHPJ

Nlovshoms, j. A , Adelsoin, E 11 ,Cizzi, NI. S. & Newsotne, WV T. (1085) Thle Perrett, D. I., Smith, P. A. J., Potter, D. D., Nlistlin, A. J., Head, A. S.,analysis of moving visual patterns. Iii. Patterni recognsition inechsanismns, Miliner, A. D. & Jecve., NI. A. (19841) Neurotics responisive to faces ined C. Chagas, R Cattass & C Cross Springer-Verlag. [sFPr] the tempioral cortex. Studies of functionsal organizations, senssitivity to

Mullen, K. T. (1985) The contrast setnsitivity of human colour vision tW red- idetitity atid relation to percepstion. Hlumani Neurobiology 3.197-greens and blue-yellow chromatic gratings. Journal of Physiology 359 381- 208. [arFI-P1400, [aPI-IP1 (1985) Visual cells in the temporal cortex senssitive to face view asid gaze

NIutrasugi, C. NI & Hloward, 1. P. (1989) Human hiorizonital oistokinetic direction. Proceedings of the Rloyal Society of Londoni B223.293-itnpmuq s'liiteh by ibme uppser v.rsuc thes lnwm'r vitiusl IO&ml Vitsual 317 IRMNIB, MK]

Neuroscience 2:73-79. (aPUP, LMC1 Perry, V. If., Oclsler, R. & Cowey, A. (198&1) Retinal cells that project to theMurray, I., MacCans, F. & Kulikowski, J J. (1983) Constribution of two dorsal lateral goimculate nuicleus in the macaquc monkey. Nesuroseence

movement detecting mechanisms to censtral and peripheral visiots. Visioni 12:1101-23. (ar~hIP]Research 23:151-59. [aFH-P] h'ctcrhiats, E. & you Der Heydt, R. (1989a) Elemntts of form perception its

Nlustillo, P. (1985) Binocular mnechasmis mnediaiting crossed atud uncrossed msonkey V'2. A correlation wills the cytohlrosnc oxidase pattern. Societystercopsis. Psychological Bulletin 97:187-201. (aFIIP] for Neuroscience Abstracts 15:161. [aFPP

Nakayaina, K (1983) NMotions parallax sensitivity atid sipace percepstion. lii. (1989b NMechiaisms of contour percepstion ini monukey visual cortex 11.Spatially oriented behavior, ed. A. Hein & M. Jeassnerod. Spriniger- Contours bsridginig gaps. Josurnial of Neurosciemice 9.17419-63. [rF11l')Verlag. [apUiP] Pettigrew, J. D. & Drehier, B. (1987) Parallel processinig of bsinocular disparity

Nakayama, K. & Silvertmans, C. If. (1986) Serial and parallel pirocessinig of us thle cat's retimsogeiiiculocortn.,il paithways. Proceedings of the Rloyalvisual feature conjunctions. Nature 32:26.-65. [aPIIP] Society, series B 232.297-321. [eFIIP]

Nathan, P XV (19416) Ots simnultancous bilateral stinmulationi of the body in a Phitllipss, C. (1933) l'ereptios of flicker its lesions of the visual pathway,%.lesions of the isarictal lobe. Brain 69:325-34. (aPIIP] Brain .56.46.-78. (011111


References/ Previc: Visual field specialization

Piaget, J. (1969) The mechanisms of perception. Rutledge & Kegan. jaFl1P1 piroperties of periarcuate neurons in macaquc mnonkeys 1. SoniatosenlsoryPierrot-Descilligny, C., Cray, F. & Birunet, P. (1986) lnfarcts of both inferior responses. Behavioral Brain Research 2.125-46 [KNIIII

parietal lobLIes with impairment of visually guided eyc movements, (1981b) Affercent properties of periarcuate neurons in inacaque mnikeys IIperipheral visual inattention and optic ataxia. Brain 109.81-97. [aFI-Il Visual responses. Behiavioral Brain Recsearch 2.147-63. [KNIIII

Plant, C. T., Zimmern, R. L, & Durdcn, K. (1983) Transient visually evoked Robertson, L. C., Lamb, NI. R. & Knight, R. T. (1948) Effects of lesions ofpotentials to the pattern reversal and onset of sinusoidal gratings. tcmporal-parietal junctioni oii perceptual and attentional processing inElect rocncephalography and Clinical Neurophysiology 56.147- hunians. jotsrnal of Neuroscience 8:3757-69 [arFHlPI58. faFHPj Robinson. D. A. (1975) Oculomotor control signals. lIn: Basic usechanisms Of

Poggio, C. R1. &c Poggio, T. (1984) The analysis of stereopsis. Annual leiiu. ocular mnotility and thieir clinical implications, edl. G. Lennerstrand & P~of Neuroscience 7:379-412. (aFIIPI Ilach-y-Rita. [RNIIII

Polit, A. & Bizzi, E. (1979) Characteristics of niotor programs underlyinig arm Robinsoin, D. L., Goldberg, NI. E. & Staiiton, G. B3, (1978) Parietalmovements in monkeys. Journal of ,Vuropliysiology 12.183-94. [RAAM association cortex iii(the primsate. Seiisory iiechanisims aiid behavioral

Polyak. S. (1957) The vcrtebrate visual system. University of Chicago modificationis. birial of Neuroliysiology4 1.910-32. [arFHlPIPress. [aFIIP, RWVWj Robson, J. 6c Craiauii, N. (1978) Probability sumination and regional variation

Post, R. B3. & Leibowitz, HI. W. (1986) Two modes of processing visual in sensitivity across the visual field. iiitestigative ophthalmnology andinformation: Implications for assessing visual imipairmnent. American Visual Science 17:221. [JRIIj

Journal of optomnetry amid Physiological Optics 63.91-96. [rFHP] Rodnian, 11. Kt, Gross, C. G. & Albirighit, T. A. k1985) Remo al of striatePrevic, F. H. (1988) The nesiroplsysiological significance of the N I aiid P1 cortex does iiot abolish responsiveness of iieurons in visual area NIT of

components of the visual evoked potential. Clinical Visioni Sciences the macaque. Society for N~euroscienice Abstracts 11.1246. [Cull3:195-202. [aFHPI Rlodman, 11. R., Gross, C. G. & Albrighat, T. D. (1986) Respoiises of ineuros

(in preparation) A general theory concerining the preiiatal origins of 4.erebral in visual area NIT after renioval of the superior colliculnv. Society forlateralization in humans. Psychological Review. [arFHlP] Neurosciensce Abstracts 12:1369. [CmlJ

Previc, F. H. & Harter, NI. R. (1982) Electroplsysiological and behavioral Rogers, BI. &t Graham, N1. (1982) Siimilarities betvseni nmotioii parallax aisdinisdcansts of selctive attention to multifeature gratings. Pcrceptioiin stercopass in humiin deipthi perception. Visioii Research 22.261-Psychophysics 32:465-72. (aFEIPI 70. [aFHPI

Ramschiaisdran, V. S. & Cavanagh, P. (1987) Motions capstuire aisisotrosly. Rolls, E. T. & Baylis, G. C. (1986) Size and cointrast have only small effectsVision Research 27:97-106. [aFIIPI on the responses to faces of neurons iii the cortex of the supserior

Rapcsak, S. Z., Cimino, C. R. & Heilman, K. NI. (1988) Altitudinal neglect tempsoral sulcos of the monikey. Experimntal Braini Research 65 38-Neurology 38:277-81. [aFIIP. CI, KNuHl 438. INIKI

Rausclsecker, J. P., von Grunau, NI. W. & Poulin, C. (1987) Centnifugal Rolls, h,. T., Blaylis, G. C., Ilasselino, NM. E. & Nalwa, V. (1989) The effect oforganization of dirctioni preferences in the cat's lateral suisrasylv an learnisigosi the face selective responses of neuroiis in the cortex iii thevisual cortex and its relation to flow field psrocessing. Joiirnial of superior temporal siilcus of the imoinkey. Experinwmital Brain RlescarchNeuroscienice 7:913-58. IDGIII 76:153-64. [GWSJ

Rsyner, K. & NicConkie, G. WV. (1976) WVhat guides a reader's eye Rothistein, T. 1). & Sacks, J. G. (1972) Defective stereopsis aili lesions of thlemovements? Vision Research 16-829-37. [JRIII parietal lobe. AnsericaiiJournial of Ophmthaliiology 73:281-81. [aFIIPI

Rayner, K. & Pollatsek, A. (1981) Eye movement control duinig reading Rovaino, J. & Virsu, k. (10979) Aui estimationi .d application of the corticalEvidence for direct control. Quarterly Journial of Expcrinemital maginificatioii factor. Experimental Brai ReIlsearch 37.495-510 jJMFIPsychology 33A:351-73. [JRBj Roy, J. P. 6c Walltz, 11. HI. (1989) Effect of the pseripheral visual field oii

Regan, D. & Beverley, K. 1. (1979) Visually guided locomotin. dispsarity sensitive cells is cortical visual area NIST. Society forPsychophaysical evidence 'or a neural niechanism sensitive to flow Neuroscience Abstracts 15:119. INIK, RNIS)patterns. Science 205-3.1-13. [IIGBJ Rouberts, A. BI. (1985) Caloric stiinulatioii and unilateral visual ineglect,

(1983) Visual fields for frontal plane motion aiid for chmaiiginsg size. Vision Neurology 35:1019-2., [aFIIP]Research 23:673-76. [aFHPI Rueckh, J. C., Cave, K. R. & Kosslyn, S. NI. (1989) Wh~y are "what"~ aiid

Regan, D., Eckelens, C. J. & Collewipn, H. (1986) Visual field defects for "where" processed by sepsarate cortical visual systems? A computationialvergenco eye movements and for stereomotion psercepstion, lnsuestmgatiie investigatioin. Journal of Cugtitmste Neuroscience 1.171-86. ICJNIIOphthalnmology & Visual Science 27.806-19. [aFHIPJ Sagar, S. NI. (1987) Soinstostatiii-like iinmuiioreactive material iii the rabbhit

Richards, WV. (1972) Disparity masking. Vision Research 12.1113-241. jaiFHl I retina. inniosthinmslstauining using inioclonal antibodiesRichards, W. & Liebermaii, H. R. (1985) Correlation between stereo ability Joiial of Comiparative Neurology 266.291-9)9. ILNIC)

and the recovery of structure-from-motion. American Journal of Sagar, S. NI. & NMarshall, P'. E. (1988) Soiiatostatiis-like iminiimoreactiveOptometriy and Physiological Optics 62.111-18. [aFH-PI material in associational gaiiglion cells of huni retina. Neuroscience

Richards, W. &c Regan. D. (1973) A stereo field iiap with iinplicatioiis for 27.507-16. [LNIC, RWW, rFIIPIdisparity processing, Investigative Ophhialmwlogy 12.9041-9. [1i0II11 Saligal, A. &c Iversen, S. D. j1978) Categorization ainu retrieval after selective

Richmond, B. J., Wurtz, R. H. & Sato, T. (1983) Visual respomises of inferior iiiferotenioral lesions in imonkeys. Brain Rlesearch 1416 341-50 jaFIIP[temporal neurons in awake rhesus *.oikey. Journal of Neurophystology Saito, H-., Ynkio, NI.. Tanaka, K., Ilikosaka, K.. Fukada, Y & Iwai, E (1986)50:1415-32. (NIKI lIitegration of direction sigiials of imiage motion in the supserior teiiporal

Rickman, D. W. & llrecha, N. C. (1989) NMorphiologies of soinitostatiii- sulcus of the iiiacautue monkey. Journal of Ncuroscicnce 6.1,15-immunoreactive neurons in the rabbit retina. In: Neurobiology of the 57. (01111, RNISIInner Retina, ed. R. Weiler & N. N. Osbsoriie. Springer-Verlag. [1,ICI Sakata, 41., Sliibutaiis, 11. & Kawaiio, K. (1980) Spsatial psroperties of visual

Rijsdijk, J. P., Kroon, J. N. & van der Wildt, G. J. (1980) Contrast seiisitivity fixation iieuroiis iii posterior parietal assoc iation cortex of the inikeyas a function of position oii the retina. Vision Research 20,235- Journal of Neurophysiology 43:1651-72 [MIK, RNIS]41. [aFHP] Sakata, If., Sliibutani, IL., Kawano, K. & Hlarringtoii, T'. L. (1985) Neural

Riopelle, A. J. & Bevan, WN., Jr. (1953) The distribution of seotopir sensitivity imechinismsis of spaice vision iii the isarietal associastion cusetex of thein human vision. American Journial of Psychology 66:73-80. [aFPP inikey. Visioni Research 25:453-63. [aFHlPI

Ristanovic, D. & Haidukovic, 11, (1981) Effects of spsatially structured stimulus Salman, S. D. & von Noordeii, G. K. k1970) iiduced vestibul1ar nystagiius infields ois pattern reversal visual evoked poteiitials sirabisinic patmientss. Annals of Oiulugy, Iihiinolugy andI Lot yiigulogyElect roencephsalography amid Cliniical Neur:'plystology 51:599- 79.352-57. [aFIIP)610. [aFHPI Scarisbrick, D. J., Tweedy, J. It. & Kuislansky, C. (1987) Hland preference

Rmzzolatti, C. & Camarda, R. (1987) Neural circuits for spatial attention aid and jserforniaiic effects on huie bisection "ieuiropsychologia 25 695-unilateral neglect. In: Neurophystological and ncuropsychologtcal aspects 99. [aFHlPIof spatial neglect, cad. M. jeannerod. Northm.Holland. [aFIIPI Schein, S. J., Marrocco, It. T. & De iomiaterio, F. NI. (1982) Is there a high

Rizzolatti, C., Gentilucci, M. & Niatelli, Ni. (1985) Selective spatial atteiition. concentration ofclrslcieclsin area NA of imionkey visual cortex?One center, one circuit, or many circuits? In. Attention and pserformnancc Jouirnal of Ncurohysiology 47.193-213. [aFIIP]XI, ell. NI. 1. Posner & 0. S. M. Niarin. Lawrence Erlbaum Schiiller, P. HI. (1986) The ceiitral visual system. Visioni Research 26 1351-Associates. [arFlIP, KMIII 86. [aFIIPJ

Rizzolatti, G., Riggio, L., Dascola, 1. & Uinita, C. (1987) Reorieinting Schiller, P. If. & Colby, C. L. (1983) Ilie respoiises of single cells iii theattention across the horizontal and vertical ineridiaiis, Evideiie iii favor lateral geiiicilate incleus of the rhesus mnuikey to color Asud hilnimmiaiceof a premotor theory of attention. Neuropsychologia 25:31-40. [aFIIPI coiitrast. Visions Research 23:1631-41. [aFIIPj

Ilizzolatti, C., Scandolara, C., NMatelli, Ni. & Gentilucei. NM. (1981a) Affereiit Schiiller, P'. 1I., Logothetis, N. K. & Charles, E It, (1989) rhic fuiictioiis of



the color-opponent (C-0) and broad-band (B-B) chanriels in perception at strabismus. The Occupational Therapy Journal of Research 2:220-isoluminance. Investigative Ophthalmology & Visual Science 30.323. 33. [aFHP](Supplement) [aFHPJ Sloan, L. L. (1947) Rate of dark adaptation and regional threshold gradient of

Schiller, P. H. & Malpeli, J. G. (1978) Functional specificity of lateral the dark-adapted eye: Physiologic and clinical studies. American journalgeniculate nucleus laminae of the rhesus monkey. Journal of of Ophthalnology 30:705-20. [aFHP]Neurophysiology 41:788-97. [aFHP] Smith, A. T. & Hammond, P. (1986) Hemifield differences in perceived

Schiller, P. H., Malpeli, J. G. & Schein, S. J. (1979) Composition of velocity. Perception 15:111-17 [arFHP]geniculostriate input to superior colliculus of the rhesus monkey Journal Snodderly, D. NI. (1979) Visual discriminations encountered in food foragingof Neurophysiology 42:1124-33. [aFHP by a neotropical primate: Implications for the evolution of color vision.

Schiller, P. H., Stryker, M., Cynader, M. & Berman, N. (1974) Response In. The behavioral significance of color, ed. E. Burtt. Garland. [aFlIPIcharacteristics of single cells in the monkey superior colliculus following Soper, H. V., Diamond, I. T. & Wilson, H. (1975) Visual attention andablation or cooling of visual cortex. Journal of Neurophysiology 37:181- inferotemporal cortex in rhesus monkeys. Neuropsychologia 13:409-94. [aFHPI 19. [aFHIPI

Schmidt, R. A., Zelaznik, 11., Hawkins, B., Frank, J. S. & Quinn, J. T., Jr. Soso, NI. J., Lettich, E. & Belgum, J. H. (1980) Pattern-sensitive epilepsy 1l:(1979) Motor-output vanability. A theory for the accuracy of rapid mote, Effects of pattern onentation and hemifield stimulation. Epilepsia 21.313-acts. Psychological Review 86:415-51. [RAA] 23. [aFItP]

Schneider, G. E. (1969) Two visual systems. Science 163:895-902. [CJB, Sprague, J M , Berluceli, G. & Rizzolatti, G. (1973) The role of the supenorLMC] colliculus and pretectum in vision and vis;'ally guided behavior. In:

Schor, C. & Narayan, V. (1981) The influence of field size upon the spatial Handbook of sensory physiology, vol. 7, part 3, ed. R. Jung. Spnnger-frequency response of optokinetic nystagmus. Vision Research 21:985- Verlag. [aFHP]94. [aFHP Stein, J. F. (1989) Representation of egocentric space in the posterior parietal

Schor, C. M. & Westall, C. (1984) Visual and vestibular sources of fixation cortex Quarterly Journal of Experimental Physiology 74:583-instability in amblyopia. Investigative Ophthalmology & Visual Science 606. [rFHP]25:729-38. [aFHP] Steinbach, M. J. (1976) Pursuing the perceptual rather than the retinal

Schwartz, S. & Kirsner, K. (1982) Laterality effects in visual information stimulus. Vision Research 16:1371-76. [aFHPjprocessing: Hemispheric specialisation or the orienting of attention? Steinmetz, M. A., Motter, B. C., Duffy,, C. J. & Mounteastle, V. B. (1987)Quarterly Journal of Experental Psychology 34A:61-77. (aFHP] Functional properties of parietal visual neurons: Radial organization of

Schwartz, T. L., Dobson, V., Sandstrom, D. J. & van lof-van Dun, J. (1987) directionalities within the visual field. Journal of Neuroscience 7:177-Kinetic perimetry assessment of binocular visual field shape and size in 91. [arFHP, RMS]young infants. Vision Research 27:2163-75. [aFHP] Stone, J. (1983) Parallel processing in the visual system. Plenum. [LMC,

Sedgwick, H. A. (1986) Space perception. In: landbook of perception and RWWJperformance, vol. 1, ed. K. R. Boff, I. Kaufman & J. P. Thomas. Wiley Strong, C. W. & Whitehead, B. A. (1989) A solution to the tag.assignment& Sons. [aFHP] problem for neural networks. Behavioral and Brain Sciences 12 381-

Sejnowski, T. J. (1988) Neural populations revealed. Nature 332:308. [RMB] 433. [GWS]Semmes, J. (1968) Hemispheric specialization. A possible clue to mechanism. Strucl, M., Prevee, T. S. & Zidar, 1. (1982) Dependence of visual evoked

Neuropsychologio 6:11-26. [MPB] potentials on change of stimulated retinal area associated with differentSergent, J. (1983a) "'he role of the input in visual hemispheric processing. pattern displacements. Electroencephalography and Clinical

Psychological Bulletin 93:481-512. [MPB] Neurophysiology 53:634-42. [aFHP](1983b) The effects of sensory limitations on hemispheric processing. Tanaka, K., Iikosaka, K., Saito, H., Yukie, M., Fukada, Y. & Iwai, E. (1986)

Canadian Journal of Psychology 37:345-66. [MPB] Analysis of local and wide-field movemeits in the superior temporalShapley, R. & Perry, V. H. (1986) Cat and monkey retinal ganglion cells and visual areas of the macaque monkey. Journal of Neuroscence 6:134-

their visual functional roles. Trends in Neurosciences 9.229-35. [aFHPI 44. (RMS]Shatz, C. J. & Stretavan, D. W. (1986) Interactions between retinal ganglion Tanaka, K & Saito, H.-A. (1989) Analysis of motion of the visual field by

cells during the development of the mammalian visual system. Annual direction, expansionicontraction, and rotation cells clustered in the dorsalReview of Neuroscience 9:171-207. (aFHPI part of the medial superior temporal area of the macaque monkey.

Shelton, P. A., Bowers, D. & Heilman, K. M. (1990) Peripersonal and Journal of Neurophysiology 62:626-41. [MKIvertical neglect. Brain 113:191-205. [KMH] Tootell, R. B. H., Silverman, M. S., Hamilton, S. L., De Valois, R. L. &

Shulman, G. L., Sullivan, M A,, Cish, IX. & Sakoda, W. J. (1986) The role of Switkes, E. (1988) Functional anatomy of macaque striate cortex III.spatial-frequency channels in the perceptiou of local and global structure. Color. Journal of Neuroscience 8:1569-93. [aFHP]Perception 15:259-73. [aFHP] Tootell, R. B. H., Switkes, E., Silverman, M. S. & Hamilton, S. L. (1988)

Siegel, R. M. (1987) Parallel distributed processing that obtains three- Functional anatomy of macaque striate cortex II. Retinotopicdimensional structure from motion. Society for Neurosctence Abstracts organization. Journal of Neuroscience 8:1531-68. [aFHP]13:630. [RMSI Treisman, A. M. & Schmidt, H. (1982) Illusory conjunctions in the perception

(1988) Discovery of structure from motion in monkey, man and machine. of objects. Cognitive Psychology 14:107-41. [aFtIP]In: Neural information processing systems, cd. D. Z. Anderson. Trevarthen, C. B. (1968) Two mechanisms of vision in primates.American Institute of Physics. [RMS] Psychologische Forschung 31:299-337. ICJBI

(1989) Motion analysis in the inferior panetal area 7a of the rhesus monkey. Tusa, R. J., Demer, J. L. & Herdman, S. J. (1989) Cortical areas involved inSociety for Neuroscience Abstracts 15:162. [RMSJ OKN and VOR in cats: Cortical lesions. Journal of Neuroscience 9:1163-

Siegel, R. M. & Andersen, R. A. (1987) Representation of head-centered 78. [aFHP]visual space in the inferior parietal lobule of macaque monkey. Tusa, R J. & Palmer, L. A. (1980) Retmotopic organization of areas 20 and 21Neuroscience 22:1293P (Supplement). [RMS] in the cat. Journal of Comparative Neurology 193:147-64. [aFHPJ

(1938) Perception of three-dimensional structure from two-dimensional Tychsen, L. (1989) Primary maldevelopmeit of visual motion pathway inmotion in monkey and man. Nature 331:259-61. [1(maJ humans. Investigative Ophthalology & Visual Science 30:302

Singer, W. (1977) Control of thalamic transmission by corticofugal and (Supplement). (aFHP]ascending reticular pathways in the visual system. Physiological Reviews lychisen, L. & Lisberger, S. G. (1986a) Visual motion processing for the57:386-420. [aFIP] iiitiation of smooth-pursuit eye movemeits ii humans. Journal of

Singer, W. (1985) Central control of developmental plasticity in the Neurophysiology 56:953-68. [aFHP, CJBImammalian visual cortex, Vision Research 25.389-96. [aFHP] (1986b) Maldevelopmcnt of visual motion processing in humans who had

(1985a) Human contrast sensitivity. Regional retinal differences. Hlumn strabismnus with onset ii infancy. Journal of Neuroscience 6.2495-Neurobiology 4:97-99, [aFHP 2508. (aFHlPI

(19&5b) Critical flicker fusion aid double flash discrmination in different Tyler, C. W. (1987) Aialysis of visual modulation sensitivity Ill. Mendonalparts of the visual field InternationalJournal of Neuroscience 25.225- variations ii peripheral flicker sensitivity. Journal of the Optical Society31. [aFHP] of America A:1612-19. [aFIIP]

Skrandies, W. (1987) The upper aid lower visial field of man. Tynan, P. & Sckuler, It. (1975) Moving visual phantoms. A new contourElectrophysiological and functional differences. In: Progress in sensory completion effect. Science 188:951-52. [aF1P]physiology (vol. 8), eds. 11. Autrum ct al Springer-Verlag. Ullnaa, S. (1979) The interpretation of visual rnotior,. MIT Press. [RMSJ

Slavik, B. A. (1982) Vestibular fuietion in children with nonparalytic Underwood, G. (1976) Attention and memory. l'ergamon. [MPB]



Ungcrlcider, L. C. 6c Mishkin, M. (1982) Two cortical visual systems. In. Investigative Ophthalmsology and Visual Science SupplementsAnalysis of visual behavior, ed. D. J. Ingle, M. A. Coodale & R. J. 29:196. (LMCJMansfield. MIT Press. [arFlIP, MK, CJM, CWS] Wlhiteside, T. C. D. (1952) Accommodation of tlie human eye in a bright and

Vaina, L. M. (1989) Selective impsairment of visual motion interpsretations empty visual field. Journal of Physiology 118:65P. [rFHPJfollowing lesions of the right occipito-parietall area us humans Biiological WVickler, K. C., WVilliams, R. W. & Rakic, P. (us press) The photorceptorCybernetics 61L347-59. LaFIIPI mosaic: Number and distribution of rods and cones in the rhesus monkey

Van Buren, J. NM. (1963) Thle retinal ganglion cell layer. Thiomas. [arFlIPI retina. Journsal of Cornparative Neurology. IRW'bWiVan Burens, J. NM. & Baldwin, M. (1958) The archsitecture of tile opstic Williams, L. C. (1966) The effect of target specification onl objects fixated

radiation in the tensporal lobe of mil. Brain 81.15-40. [aFIIPI durinsg visual scarcs. Perception & Psychophysics 1.315-18. JaFHPIVan Essen, D. C. (1985) Funsctsonal organsization of primate % isual cortex Iii. Williams, MI. C. & Brannan, J. R,. (1986) Global precedensce in good and poor

Cerebral cortex V'ol. 3 Visual cortex, ed A. P'etera & E. C. Jones. readers. Paper presented at the ansnual meeting of the PsychononiePlenum Press. (BNI13I Society, November, New Orleans, LA. IJRBI

Van Essen, D. C. & Maunsell, J. H. R. (1983) Hierarchsical organization and Williams, R. W., Cavada, C. & lteinoso-Suarcz, F. (1989) Evolutionaryfunctional streanms in the visual cortex. Trendss in Neurosciences 6.370- plasticity in the feline retino-geniculate system. Society for Neuroscience75. (aFHP) Abstracts 15:457. [RWWI

Van Essen, D. C., Maunsell, J. H. R. & Bixby, J. L. (1981) The middle Wilson, NM. E. & Toyse. NM. J. (1970) Retino-tectal and cortico-tectalteinposal visual area in the macaque. Myeloarelsitecture, connections, projections iii Macaca mnulatta. Brain Research 24:395-406. [aFHPIfunctional properties and topsographic organ,zation. Journaol of Wilson, H. R. & Ciese, S. C. (1977) Threshold visibility of frequency gradientComparative Neurology 199:293-326. [aFHPJ patterns. Vision Research 17:1177-90. [JRBI

Van Essen, D. C., Newsome, W. T. & Maujisell, J. If. R. (1984) Thec visual Wilson, NM., Wilson, WV. A., Jr. 6& Remez, R. (1977) Effects of prestriate,field representation in striate cortex of thle macaque monkey. inferotemporal and superior temporal sulcus lesions on attention and gazeAsymmetries, anisotropies, and individual variability. Vision lResearchi shifts iii rhesus monkeys. Journal of Comparative and Physsological24:429-48. (aFHPI Psychology 9W:261-71. [aFHPI

Vassilev, A. & Strashiimirov, D. (1979) On tile latency of humsan visually Wolfe, J. NM. & Franzel, S. L. (1988) Biniocularity and visual search.evoked response to sinusoidal gratings. Vision Research 19:843- Perception & Psychophysics 44:81-93. [aFHPI45. (aFHPI Woodworth, B. S, (1938) Experimental psychology. Holt 6c Co. [aFHPI

Ventre, J. & Faugier-Crnaud, S. (1986) Effects of posterior parietal lesions Yasuma, T., Miyakawa, N. & Yansazaki, J. (1986) Clinical application of time-(area 7) on VOR in monkeys. Expep mnental Brains Research 62.65&1- depensdent perlmetiy 1. Results in normal subjects. Japanese Journal of58. [aFHPI Ophthalmnology 30:330-37. [aFHPI

Voii Der Nlarlsburg, C. & Singer, WV. (1988) Principles of cortical network Yazulla, S. (1974) Intraretinall differenstiation in the synaptic organization of theorganization. In: Neurobiology of neocortex, ed. P. Rakic & WV. Singer. iisner plexiform layer of the pigeon retina. Journal of ComparativeWiley & Sons. [aFHPJ Neurology 153:309-24. [MAGI

Von Noorden, C. K., Crawford, NM. L. J. & Levacy, R. A. (1983) The lateral Yin, R. 1 (1969) Looking at upside-down faces. Journal of Expersvnentalgeniculate nucleus in human anisometropic amblyopia. Investigative Psycholog ( 81:141-45. [rFHPIOphthalmology & Visual Science 24:788-90. (sF11PI Young, L. R. & Oman, C. NI. (1974) Influence of head position and field on

Wagman, 1. H. (1964) Eye movements induced by electnic stimulation of visually induced motion effects in three axes of rotation. In. Proceedingscerebrum in monkeys and their relationship to bodily movements, In. of the Tenth Annual Conference on Manual Control, Wright-PattersonThe oculomotor systemi, ed. NI. B. Bender. Harper & Row. [sF111P1 Air Force Base, OH. [aFHPI

Walls, C. 1. (1942) The vertebrate eye and its adaptive radiation. Craishmok Yund, E. W., Efron, B. & Nichols, D. B. (1990) Detectability gradients as aPress, [RWWJ function of target location. Brain and Cognition 12:1-16. [rFHPI

Warrington, E. K. & James, NI. (1967) Disorders of visual perception in Zalutsky, R. A. & Miller, R. F. (1987) The physiology of somatostatin in thepatients with localised cerebral lesiomis. Neuropsychologia 5:2-53- rabbit retinia. Society for Neuroscience Abstracts 13:380. ILNICI66. [aFHPI (1988) Neuropeptide mechanisms in the rabbit retina. Investigative

Warrington, E. K. & Taylor, A. NI. (1973) The contribution of the right Ophthalmnology and Visual Science Supplements 29:271. [LMCJparietal lobe to object recognsition. Cortex 9.152-64. [aFIIPJ Zee, D. S., Yamazaki, A., Butler, P. 1H. & Cucer, C. (1981) Effects of

Wassersteins, J., Zappulla, R., Rosen, J., Cerstman, L. & Rock, D. (1987) In ablation of flocculus and parallocculus on eye movements in primates.search of closure: Subjective contour illusions, gestalt completion tests, Journal of Neuropliysiology 46.878-99. [aFHPJand implications. Brain and Clsgnition 6.1-14. [aFHPI Zeki, S. (1980) The representation of colours in the cerebral cortex. Natulre

Weinstein, E. A. (1980) Clinical features of lsemi-inattesstion. Behavioral auid 284:412-18. [aFHPJBrain Sciences 3:518-20. [aFHIIP Zeki, S. M. (1973) Colour coding in rhesus monkey prestriate cortex. Brain

Weymouth, F. W., Wiles, D. C., Acres, L. H., Raaf, J. E. & Wheeler, M. Research 53:422-27. [IKIC. (1928) Visual acuity within the area centralis and its relation to eye (1977) Colour coding in the superior temporal aulcus of rhesus monkeymovements and fixation. Amserican Journal of Ophthalnology 11.947- visual cortex. Proceedings of the Royal Society, series B 197; 195-60. [aFHP] 223. [MK)

White, C. A., Chalupa, L. NI., Johnson, D. & hlrecha, N. S. (in press) Zihil, J., von Cramoss, D. & Mtai, N. (1983) Selective disturbance ofSomatostatin immunoreactive cells in the adult cat retina. Journal of movement vision after bilateral brain damage. Brain 106.313-Comparative Neurology. [LNI 40. [aFHP)

White, C. A., Johnson, D., Clialupa, L. NI. & Breclia, N. C. (1988a) Zipser, D. & Andersen, R. A. (1988) A back-propagation programmedCharacterization of aomatostatin-like immunoreactive cells in the adult cat network that simulates response properties of a subset of posteriorretina. Society of Neuroscience Abstracts 14:1121. (LMI parietal neurons. Nature 331:679-84. [rFHP, RMSI

(1988b) Somatostatin immunoreactivity (SRIF-h) in the adult cat retina.


Documents

CLASSIFICATION OF THIS REPORT DOCUMENTATIO AD ...ASC 05 08 Space perception; Visual association €ortex; Visual attention 19. ABSTRACT (Continue on reverse If necessary and identify