Representation of the visual field in the lateral intraparietal area of macaque monkeys: a quantitative receptive field analysis

Abstract The representation of the visual field in theprimate lateral intraparietal area (LIP) was examined, us-ing a rapid, computer-driven receptive field (RF) map-ping procedure. RF characteristics of single LIP neuronscould thus be measured repeatedly under different be-havioral conditions. Here we report data obtained using astandard ocular fixation task during which the animalswere required to monitor small changes in color of thefixated target. In a first step, statistical analyses wereconducted in order to establish the experimental limits ofthe mapping procedure on 171 LIP neurons recordedfrom three hemispheres of two macaque monkeys. Thecharacteristics of the receptive fields of LIP neuronswere analyzed at the single cell and at the populationlevel. Although for many neurons the assumption of asimple two-dimensional gaussian profile with a centralarea of maximal excitability at the center and progres-sively decreasing response strength at the periphery canrepresent relatively accurately the spatial structure of theRF, about 19% of the cells had a markedly asymmetricalshape. At the population level, we observed, in agree-ment with prior studies, a systematic relation betweenRF size and eccentricity. However, we also found a moreaccentuated overrepresentation of the central visual fieldthan had been previously reported and no marked differ-ences between the upper and lower visual representationof space. This observation correlates with an extensionof the definition of LIP from the posterior third of thelateral intraparietal sulcus to most of the middle and pos-terior thirds. Detailed histological analyses of the record-

ed hemispheres suggest that there exists, in this newlydefined unitary functional cortical area, a coarse but sys-tematic topographical organization in area LIP that sup-ports the distinction between its dorsal and ventral re-gions, LIPd and LIPv, respectively. Paralleling the physi-ological data, the central visual field is mostly represent-ed in the middle dorsal region and the visual peripherymore ventral and posterior. An anteroposterior gradientfrom the lower to the upper visual field representationscan also be identified. In conclusion, this study providesthe basis for a reliable mapping method in awake mon-keys and a reference for the organization of the proper-ties of the visual space representation in an area LIP ex-tended with respect to the previously described LIP andshowing a relative emphasis of central visual field.

Keywords Parietal cortex · Monkey · Electrophysiology ·Receptive field · Visual representation

Introduction

Higher-order areas in the occipitoparietal cortical path-way carry multiple classes of signals related to sensory,motor and cognitive parameters. One such example isthe lateral intraparietal area (LIP), which, by its connec-tivity and by the response properties of single neurons indifferent behavioral conditions, is thought to be involvedin predictive visual processing, visuospatial attention andsaccadic eye movement programming (Lynch et al.1977; Gnadt and Andersen 1988; Blatt et al. 1990;Duhamel et al. 1992; Colby et al. 1996).

In macaques, area LIP has been described to occupythe most posterior third of the lateral bank of the intra-parietal sulcus. Its major sources of input originate in theextrastriate visual cortex. It has an extensive set of con-nections with both dorsal and ventral stream areas, in-cluding V2, V3, V3A, MT, MST, PO, V4, MDP, DP,TEO, and TE (Seltzer and Pandya 1986; Colby et al.1988; Blatt et al. 1990; Andersen et al. 1990a; Fellemanand van Essen 1991; Bullier et al. 1996). Area LIP is re-

S. Ben Hamed · J.-R. Duhamel (✉ ) · F. Bremmer · W. GrafCNRS Collège de France, 11 place Marcelin Berthelot,75005 Paris, Francee-mail: [email protected].: +33-04-37911235, Fax: +33-04-37911210

F. BremmerDepartment of Zoology and Neurobiology, Ruhr University,44780 Bochum, Germany

S. Ben Hamed · J.-R. DuhamelInstitut de Sciences Cognitives, CNRS UPR 9075,67 boulevard Pinel, 69675 Bron, France

Exp Brain Res (2001) 140:127–144DOI 10.1007/s002210100785

R E S E A R C H A RT I C L E

S. Ben Hamed · J.-R. Duhamel · F. BremmerW. Graf

Representation of the visual field in the lateral intraparietal areaof macaque monkeys: a quantitative receptive field analysis

Received: 30 October 2000 / Accepted: 27 April 2001 / Published online: 24 July 2001© Springer-Verlag 2001

ciprocally connected with other parietal areas (VIP, 7a)(Seltzer and Pandya 1986; Blatt et al. 1990; Andersen etal. 1990a) and has outgoing connections directed to thefrontal eye fields (Cavada and Goldman-Rakic 1989;Andersen et al. 1990a; Schall et al. 1995; Bullier et al.1996), to different subdivisions of the premotor cortex(Cavada and Goldman-Rakic 1989), and to the interme-diate layers of the superior colliculus (Asanuma et al.1985). The strong visual input received by LIP predictsan equivalently strong visual responsivity, and indeedmost LIP neurons display reliable and robust responsesto visual stimulation (Andersen et al. 1985; Colby et al.1996). However, other types of stimuli can activate LIPneurons or modulate their visual responsiveness. LIP vi-sual responses have been shown to be enhanced throughthe manipulation of stimulus behavioral relevance or sa-liency (Lynch et al. 1977; Colby et al. 1996; Ben Hamedet al. 1997a; Platt and Glimcher 1997). Learned respons-es to auditory stimuli have also been described(Stricanne et al. 1996; Grunewald et al. 1999; Linden etal. 1999), as well as by eye and head position signals(Andersen et al. 1990b; Brotchie et al. 1995; Bremmer etal. 1997, 1998), and to fire before purposive saccadiceye movements in the absence of visual stimulation(Barash et al. 1991a, 1991b; Colby et al. 1996). Such di-verse response properties have led to postulating a cen-tral role for area LIP in multimodal spatial analysis, andin attentional and oculomotor functions.

Considerable effort has been made in recent years todocument the effects of behavioral variables in area LIPthrough the use of elaborate experimental paradigms.However, with the exception of a single study conductedin anesthetized animals (Blatt et al. 1990), comparativelylittle attention has been given until recently to basic re-ceptive field properties and to visual field representationin LIP (Platt and Glimcher 1998). Such information isnecessary in order to better understand how visual spaceis represented in a single area, how visual representationvaries across the different subdivisions of the cortical vi-sual system, and how these variations relate to theachievement of specific functional goals. Furthermore,sampling visual responses over a large portion of spacemay help highlight the mechanisms through which a giv-en behavioral context influences information processingacross the whole visual field representation in a givencortical area and not only at the locus where the attendedstimulus or the programmed saccade endpoint is posi-tioned.

Classical studies that aimed to characterize the recep-tive fields (RFs) of subcortical and cortical visual areasused hand-mapping procedures for which the decisionabout the significance of the visual response is depen-dent upon the experimenter. The determination of RFs isshown to be constant across two successive mapping ses-sions for a given experimenter but not necessarily fordifferent experimenters. As a result, estimates of RF sizevary from one study to another. However, the relation-ship between size and eccentricity remains fairly consis-tent for a given visual area. The main limit of this proce-

dure is that it is qualitative and does not allow a refinedanalysis of both temporal and spatial properties of RFs.

The introduction of automatic mapping proceduresusing white noise stimulation and reverse correlationtechniques has yielded a new understanding of the pro-cessing of visual information at the level of the RFs.This approach has been used to study the RFs of lowervisual areas in anesthetized animals (in LGN: Cai et al.1997; in V1: DeAngelis et al. 1993a, 1993b). It has alsobeen extended to higher visual areas such as MT andMST (Raiguel et al. 1995, 1997). The precise spatiotem-poral characteristics of RFs were analyzed in relation tothe process of feature extraction from an ongoing visualscene. Independently, other studies have aimed to ana-lyze the changes in visual responsiveness in relation toattention to stimulus attributes such as color or spatial lo-cation (Bushnell et al. 1981; Goldberg et al. 1990;Motter 1993, 1994; Steinmetz et al. 1994; Duhamel et al.1995; Connor et al. 1996; Ben Hamed et al. 1997b), or tothe preparation of an eye-orientation movement (Duhamelet al. 1992). These latter approaches, carried out in high-er visual areas in awake animals, have not examined indetail the neurons’ RF characteristics or substructure.

In the present study, we tested an experimenter-freemethod that uses a computerized stimulus presentationprocedure and allows the generation of a quantitativerepresentation of the RF structure of single neurons. TheRF data presented here have been obtained during theperformance of a foveal fixation task. In the first part ofthe present study, the reliability and limits of this com-puterized mapping method were investigated, and the ba-sis for further analysis and comparisons of RF structurewere laid out. In the second part, the representation ofthe visual field in LIP was investigated by two comple-mentary approaches: (1) spatial properties of a single RFwere studied and a critical perspective was laid on theinformative parameters of the fine structure of RFs. (2)Information about all the recorded RFs was used to re-present the allocation of spatial resources (in terms ofcell number) as a function of visual field extent.

We thus describe an overrepresentation of the centralfield at the cortical level with respect to other areas. Wealso show that the visual, saccadic and delay activitieswhich characterize LIP neurons are not restricted to theposterior third of the lateral intraparietal bank but alsoextend into most of the middle third of it. This newly de-fined anterior part of LIP contains both an emphasis onthe lower visual field in the ventral part of the bank andon the central space in the dorsal part. Thus it emergesthat LIP does not hold a representation of space with anup-down asymmetry, but rather with a center-to-periph-ery asymmetry. The coarse but systematic topography ofLIP is also confirmed and expended: the central visualfield is found to be mostly represented in the dorsal re-gion, while the visual periphery is represented more ven-trally, with an anteroposterior gradient from the lower tothe upper visual field representations.

128

Materials and methods

Animal preparation

Two female macaque monkeys (Macaca mulatta and M. fascicul-aris) were used for this study (3.8 kg and 4.6 kg, respectively).Before surgery, they were trained to sit in a primate chair and tofixate a spot of light to receive a liquid reward (Wurtz 1969). Theywere surgically prepared under general anesthesia (induced with10 mg/kg propofol and maintained at 15 mg/h/kg) for chronic neu-rophysiological recording by implantation of scleral search coils(Judge et al. 1980), headholding devices, and recording chambersthrough which electrodes could subsequently be introduced intothe cerebral cortex. Recording chambers (1.8 cm diameter) wereanchored over the intraparietal sulcus at stereotaxic coordinatesAP –11 and ML 16 mm for a first chamber and AP 5 and ML 11for the two other chambers. The chambers were placed flat againstthe skull. Given the orientation of the intraparietal sulcus, thisplacement was approximately orthogonal to the skull, and allowedlong, tangential electrode penetrations through either banks of thesulcus. Animals were watched closely following surgery and giv-en analgesics as needed. During the recording period, animalweight and health status were carefully monitored. Fluid supple-ments were given as needed. Recording chambers were flushedwith saline before and after each recording session and antibioticswere applied as needed. When necessary, granulation tissue accu-mulated over the exposed dura in the recording chamber was re-moved with the animal under ketamine anesthesia. All experimen-tal procedures were in compliance with local and European regu-lations (European Communities Council Directive 86/609/EEC).

Physiological methods

Recordings were made with flexible tungsten commercial micro-electrodes (Frederick Haer electrodes) introduced through stain-less steel guide tubes. A nylon grid held rigidly in the recordingcylinder was used to maintain the guide tubes in place and permit-ted highly reproducible electrode penetrations with a resolution of0.5 mm (Crist et al. 1988). The electrode was lowered into thebrain with a step-motor driven hydraulic microdrive (Narishige).Penetrations performed several months apart at the same grid loca-tion were found to yield neurons with similar response types atsimilar depths.

During a recording session, the head-fixed monkey sat in a pri-mate chair in a totally dark room facing a tangent screen 57 cmaway. The screen was 1×1.4 m in size, and the effective projectionsurface subtended slightly more than 90° in width and 70° inheight. Visual stimuli of adjustable size, shape, color and motionpattern were produced by computer software developed in the lab-oratory. The stimulations were back-projected onto the screen viaa liquid-crystal projection system. Horizontal and vertical eye po-sition signals were measured using the magnetic search coil meth-od. Behavioral control, eye position monitoring, stimulus positionand timing and unit recording were performed using a PC-basedreal time experimental system (REX, Hays et al. 1982). Eye posi-tion was sampled and stored at 250 Hz and discriminated unitswere stored at 1000 Hz. The computer program was able to dis-play rasters online, synchronized to one of several events such asachievement of fixation, stimulus appearance or extinction, eyemovement onset, or reward. Unit discharges, eye position traces,task parameters and behavioral indicators were saved on disk foroffline analysis.

The monkeys were trained on a series of tasks designed to dif-ferentiate sensory, attentional and motor correlates of neural activ-ity. In particular, a memory guided saccade task was used to testfor specific saccade-related discharges. In this task, the monkeyhad to maintain fixation on a central target, while a peripheralstimulus was flashed for 150 ms at one of several possible prede-fined locations. After a delay of 1600 ms, the fixation spot was ex-tinguished, providing the cue for the monkey to make a saccade tothe remembered location of the peripheral flash. In each electrode

penetration, and while the monkey was performing this task, weactively searched for neurons with visual, delay-period and sac-cade-related activity which are characteristic of LIP neurons.However, because we were interested in the modulation of visualresponse properties of LIP neurons by both external and internalcues, and thus very aware of this phenomenon, we also exploredthe responses of the isolated neurons in a wide range of othertasks, aiming to characterize the specificity of its visual response(orientation selectivity, color selectivity, habituation properties, at-tentional modulations, etc.).

When a neuron was isolated at the electrode tip and the signalwas stabilized, its visual and saccadic properties were thoroughlyexplored across the visual field encompassed by the tangentialscreen and beyond in all directions, using static and dynamichand-held objects and projected stimuli on the side walls and ceil-ing. This was done for two main reasons: (1) towards the fundusof the sulcus, LIP is adjacent to area VIP. The neurons of this areahave unmistakable properties (sensitivity to stimulus direction ofmovement, optic flow responses, bimodal responses, etc.), whichenable us to distinguish them from LIP neurons and thus to identi-fy the border between the two areas (Colby et al. 1993). The RF ofVIP neurons can, however, be very eccentric and we thus system-atically explored as large an extent of the visual field as possible.(2) The focus of the present study and of the study being conduct-ed in parallel being the visual representation in LIP, it was crucialnot to miss any type of visual response in the explored cortical re-gion.

We were specifically interested in recording from neurons withdefinite visual responses. Neurons with purely visual responseswere intermingled with visual and saccade related neurons. Typi-cally, complete data sets of neurons with mostly presaccadic activ-ity were not recorded.

Histological methods

Histology was carried out on the Macaca mulatta, the other mon-key still participating in ongoing experiments. After recording ses-sions were terminated in the rhesus monkey, microlesions (50 µAfor 15 s) were made at specific locations in the lateral banks of theintraparietal sulci of both hemispheres. The animal was initiallyfixated with a buffered solution (pH 7.4) of 4% paraformaldehyde.The head was severed and introduced into the stereotaxic appara-tus. Marking pins were inserted through the periphery of the re-cording grid with a microdrive to document the extent of thechamber as well as the orientation of the recording grid. Subse-quently, the head was postfixed in paraformaldehyde solution forseveral days. At the end of this period, the pins were removed, andthe brain was extracted. A block of tissue containing the intrapari-etal sulci and neighboring regions was cut from both hemispheres.These tissue blocks were immersed in 0.4 M phosphate buff-er/10% sucrose solution for 2 days. The tissue was then cut on afreezing microtome (50-µm sections). In order to capture completeelectrode penetrations within single brain sections, the tissue wascut along planes parallel to the marking pins. Sections were count-erstained with thionine. One in ten sections were myelin stainedusing the Schmued method (Schmued 1990). Sections were digi-tized using a microscope coupled to a Neurolucida system. Two-dimensional flattened reconstructions of the intraparietal sulcuswere subsequently obtained using software developed in our labo-ratory.

Receptive field mapping procedure

In the present study, and after an exhaustive characterization of theproperties of the neural response of the isolated unit, receptivefields were mapped while the monkey was performing a standardcentral fixation task in an otherwise totally dark room. The mon-key gazed at a central green fixation point and was rewarded forholding its eye position within a 2° wide electronically definedwindow for a 2300- to 3000-ms interval and for releasing a lever

129

within 700 ms of the dimming or change in color of the fixationpoint (Fig. 1a). While the monkey was performing this task, thereceptive field of the recorded neuron was mapped in two steps:

Optimization of the mapping parameters to each recorded neuron

(a) The receptive field of the isolated unit was first approximatedmanually with a hand held projector. (b) The cell was tested withseveral computer-controlled visual stimuli varying in color, size,shape and orientation. The mapping was carried out using a simpleachromatic 1° wide spot, which was an optimum stimulus for thevast majority of neurons. (c) The neuron’s response to repeatedpresentations of a brief flash at a fixed location inside the RF wasanalyzed in order to estimate its latency and the time course of itsdecay. This information was used to optimize the duration of themapping stimulus and the interstimulus interval. (d) A matrix ofstimulus locations was defined in order to cover an area largerthan the RF. The matrix contained 7×7-square subregions in abouttwo-thirds of the cells. The remaining third of the cells was testedat a higher resolution (9×9). For the majority of the neurons, eachsubregion had a width of 6°, but it could be as small as 1° or as

large as 10°, depending on the manual estimate of the RF size(Fig. 1b).

Online receptive field mapping

Once the spatial and temporal mapping parameters were set to theneuron’s response characteristics, stimulation cycles were repeat-edly run on the screen during the pre-discrimination period of thefixation task. A single cycle stimulated the center of each subre-gion in random order. Typically, stimulus durations and interstim-ulus intervals were 100 ms and 200 ms, respectively. About six toeight stimuli were presented during a single fixation trial. Thus, acomplete stimulation cycle was obtained every seven to nine tri-als. Depending on the neural responsiveness, 7–12 stimulation cy-cles were run for a given unit. Stimulus order was reshuffled everytime a new cycle began. Thus, each point of the stimulation gridwas stimulated at least 7 times, each time embedded in a differentspatial and temporal context.

Data analysis

Offline reverse correlation analysis

Peristimulus histograms (PSTHs) were constructed for each ma-trix pixel using the information collected over the successive map-ping cycles. The PSTHs of the ten stimuli giving the highest neu-ronal activity were displayed using an interactive interface. A timewindow was adjusted interactively around the responses on the ba-sis of the latency and the time course of the response. The sametime window was applied to all the locations of the matrix in orderto determine the average frequency of discharge induced by thecorresponding stimulus. LIP neurons responded remarkably wellto these briefly flashed stimuli presented in rapid sequence. Thereverse correlation method implies that the averaging window isadjusted to the latency and peak response, which yielded relativelyhigh mean firing rates for cells with a sharp phasic burst of activi-ty. The obtained rates are somewhat higher than the typical meanfiring rates reported in other studies using wider averaging win-dows set on stimulus onset rather than burst onset.

Construction of receptive field maps

Visual receptive fields are often approximated by gaussian or Ga-bor functions. These approximation procedures make essential hy-potheses on the spiking properties of neurons, and more particu-larly on their spatial profile. When this profile is unknown orwhen, as in the present study, there is a specific interest in a quan-titative description of the spatial distribution of activity within re-ceptive fields, these methods are not appropriate. However, in or-der to increase the size of each single recording set beyond the ex-perimental limitations of the spatial and temporal resolutions, atwo-dimensional cubic interpolation method was used. It consistedof new interleaving data points that were generated between theexperimentally obtained ones to the best fitting polynomial func-tion of the third degree. The input to this operation was a squarematrix containing the average spike responses for each singlestimulus. The output was a larger square matrix that necessarilycontained the experimental data points. In essence, three newpoints were generated between each two experimental data points.Thus if the input was a 7×7 matrix of experimental data points, theoutput was a 25×25 matrix.

Description of receptive fields

Several descriptive parameters were associated with the represen-tation of RFs for each data set. (1) The maximum discharge ratewas referred to as the peak. (2) The x- and y-locations of the peakwere expressed in degrees of visual angle with respect to the cen-

130

Fig. 1A, B RF mapping procedure during a standard central fixa-tion task. A Visual stimuli presented during fixation (fixationpoint and mapping stimuli); bar status and eye position are shownrelative to the same time scale. The monkey had to grasp a bar inorder to initiate a trial (straight line in bar status). This triggeredthe appearance of the fixation point (black in fixation point status),which the monkey was required to fixate. After 2300–3000 msfollowing the beginning of fixation, the central stimulus changedcolor (gray in fixation point status), which signaled to the monkeyto release the bar within a maximum delay of 700 ms (dashed linein bar status) in order to obtain a liquid reward. During a singlefixation trial, six to eight mapping stimuli were flashed (numberedstimuli). B Stimuli were flashed at randomly selected locationswithin a predefined grid, optimally centered on the RF of the cell(represented by the idealized ellipse), and not on the fixation point(solid dot)

ter of the screen. (3) The x- and y-locations of the center of gravi-ty, i.e., of the center of mass of activity for the portion of the gridfor which responses exceed half the maximum response of theneuron. (4) A width (the average extent of the RF) was defined indegrees as the square root of the surface of the RF whose activitywas above a critical value. This critical value was set to half thepeak discharge rate.

Results

A convenient way of representing neural activity over amapped region of the visual field is to use gray-shades-coded two-dimensional maps. The procedure of howsuch a map is derived from the peristimulus histogramsobtained for each stimulated subregion is illustrated forone representative neuron (Fig. 2). The first step in theanalysis of LIP receptive field data was to assess the reli-ability of our mapping procedure, since to our knowl-edge this is the first time such a method is applied to an-alyze the visual properties of parietal cortex neurons.

Validation of RF mapping procedure

The descriptive parameters were obtained from interpo-lated data matrices corresponding to averaged dischargefrequencies. However, standard errors around these meanvalues were available only for those frequencies corre-sponding to experimental data points. In a first step, weaddressed the question of the number of matrix cyclesnecessary to produce consistent estimates of RF parame-ters, with respect to the response variability of LIP neu-rons. This was necessary because the number of matrixcycles is the major limiting factor of the present experi-mental design in terms of time cost of a single cycle.

An analysis was carried out on the total set of recon-structed RFs from the three studied hemispheres (171cells) to establish 95% confidence intervals for differentRF parameters and for each stimulation cycle.

For the entire cell sample, normalized confidence in-tervals (the parameter r, expressed as a percentage) were

131

Fig. 2A–C Offline RF reconstruction. A For each set of flashespresented at a given location of the mapping grid, trials are cumu-lated. The three panels represent 200 ms of spike trains and thecorresponding poststimulus histogram (PSTH) relative to stimulusonset for three different locations of the mapping grid (calibrationbar on the left of the histogram = 250 spikes/s, 4-ms bins). B Tocompute the RF map, the highest PSTHs are displayed offline, anda time window (inverted triangles above histograms) is definedmanually. Average discharge rate throughout this window is calcu-lated for each stimulus location. This average discharge rate (andstandard error) is displayed as a function of stimulus position in across-section of the grid running 4° below the horizontal meridian.Solid symbols correspond to the values obtained from the left pan-el histograms. C The discharge rate of the neuron is also displayedas a function of stimulus position represented in color code. x- andy-coordinates are indicated in degrees. White cross indicates thelocation of the fixation point, i.e., the intersection of the retinalmeridians. White line indicates the level at which the section ofthe middle panel is taken. White dots indicate the mapping posi-tions for which discharge rate was significantly above the sponta-neous discharge rate of the neuron

calculated for maximum discharge rate and for RF width.The r-value for maximum discharge rate has a mean of16.6% and an SD of 9.64%. Estimates of width aresomewhat more reliable (mean = 9.57% and SD =5.85%). These values set the limit of our quantificationmethod of the neuronal response tuning at the single celland population level, as well as for qualitative compari-sons between two successive behavioral mapping condi-tions.

Confidence intervals for the horizontal and verticalpeaks and center of mass did not differ significantly;thus in subsequent analyses, only radial eccentricity ofthe peak and center of mass were considered. Eccentrici-ty of peak location could be determined with a 95% con-fidence interval of 2.5° for more than half of the cells(mean = 3.15°, SD = 2.79°). Even better estimates wereobtained when center of mass location was smaller(mean = 1.66°, SD = 1.08°).

Confidence intervals for discharge rate, size and ec-centricity were calculated on the total number of stimu-lation cycles obtained for each RF mapping, but an anal-ysis carried out on all cells indicated that these parame-ters show almost no variation after six stimulation cycles(see examples of nine representative cells in Fig. 3).

Characteristics of the population of neurons studied

Visual and oculomotor-related activity

During electrode penetrations visual and eye-movementrelated responses of the neurons were used to estimatethe recording location in the intraparietal sulcus. LIPneurons are known to carry multiple signals (Colby et al.1996). In order to establish the comparability of the pop-ulation of LIP neurons studied here with previously re-ported data, we characterized each neuron in terms of itsactivity during the memory-guided saccade paradigm.Four or eight standard saccade directions were used (7.5°of eccentricity, uniformly and radially arranged aroundthe fixation point at 90° or 45° intervals, respectively),and the direction eliciting the best responses was select-ed for analysis. For all directions, the presence of visualactivity, delay period and saccade related activity wasdefined statistically by comparing activity during these

epochs with baseline activity with a t-test using a signifi-cance criterion of P<0.05. The baseline of each cell wasdefined as the level of discharge over a period of 100 ms,following achievement of stable fixation and before thepresentation of the saccade target. Visual activity wasdefined as that activity which occurred during the 100-ms window following the appearance and the extinctionof the saccade target (that is both ON and OFF visual re-sponses were tested), and adjusted to the cell’s responselatency (visual latency distribution on a subpopulation of103 LIP neurons had a mean equal to 94 ms and an SDequal to 22 ms). Presaccadic activity was defined as ac-tivity taking place during the 100 ms preceding the be-ginning of the eye movement, and postsaccadic activitywas defined as activity during the 100 ms following theend of the eye movement. Because of our initial selec-tion criteria, the overwhelming majority of the cells ofour sample (296/313, 94.5%) had a visual response. Ofthese visual cells, 249/296 responded also before or afteran eye movement (79.5%), independently of any visualstimulation. The eye-movement related activity was pre-saccadic in 188/249 cells (75.5%) and postsaccadic in61/249 cells (24.5%) (Table 1). A small percentage ofthe cells (4.8%) had a purely oculomotor activity(15/313). In addition to visual and saccade related activi-ty, we estimated the presence of tonic activity during a500-ms time interval starting about 400 ms after the ex-tinction of the target of the saccade, and finishing about400 ms before the extinction of the fixation point. Aboutone-third of the cells had significant delay-period activi-ty (111/313). This activity was usually associated withsaccade related activity (107/111) and visual activity(108/111) (Table 2).

132

Fig. 3A–C RF parameter esti-mates as a function of the num-ber of stimulation cycles for arepresentative subset of nineneurons. Asymptotic values arewell established after six stimu-lation cycles. A Maximumspiking frequency in spikes persecond. B Width in degrees.C X-position of the spikingpeak in degrees

Table 1 Visual and saccadic characteristics of the recorded LIPpopulation (N=313) during a memory-guided saccade task

Presaccadic Postsaccadic Not saccadic

Visual 188 61 47No visual 8 7 2

Characteristics of the visual responses

Visual responses differed across the population of LIPrecorded neurons. In response to the flashed target in thememory-guided saccade task, transient increases in theinstantaneous discharge rate following the appearance(on response) or disappearance (off response) of a visualstimulus were observed. In 169 of the 296 visual neurons(57%), a strictly on-visual response was observed and 7(2.7%) had a pure off-visual response. The rest, 120(40.3%) neurons, had a biphasic on and off response.

In this study, mainly neurons with well-defined visualresponses were analyzed, regardless of the presence ofdelay-period or eye-movement related activity. Quantita-tive RF maps were obtained for 242 of the visual neu-rons, 70.6% of which (171/242) were selected for furtheranalysis. This selection was based on two criteria. First,the neuron had to remain well isolated throughout themapping procedure. Second, the obtained RF had to belargely encompassed by the mapping grid. The latter cri-terion explains why RF were not mapped in all of the296 visual neurons. Indeed, in some cases, a large regionof the RF appeared to be situated outside the mappinggrid limits (32.3% of that excluded). This occurred forcells with peripheral RF as well as with more central RFswhen the mapping grid was improperly centered over theregion of interest. The remaining cases were cells whichshowed visual responses to manual probes but could notbe driven by the computer-driven mapping stimulus(67.7% of that excluded).

Visual field representation in LIP

RF location

The location of the RF can be related either to the locationof the center of mass (cm) of the activity inside the RF, or tothe location of the peak (pk). Description of the popula-tion’s spatial distribution in terms of RF peak or center ofmass showed no statistically significant difference. InFig. 4, the absolute value of the horizontal eccentricity ofthe RFs of neurons recorded from both left and right hemi-spheres are projected onto the x-axis, so that positive ab-scissas are assigned to peaks and negative abscissas to cen-ters of mass; 11.7% of the neurons had the center of mass oftheir RF ipsilateral to the recording hemisphere and 5.8% ofthe neurons had the peak of their RF ipsilateral to the re-cording hemisphere. Thus, Fig. 4 can be assumed to pro-vide an accurate image of how RFs distribute over space.

The spatial distribution of the RFs of the populationof recorded LIP neurons in our sample shows an over-representation of the foveal space with respect to the pe-ripheral space. About 30% (51/171) of the cells havetheir peak in the central 3°, 46% (79/171) in the central5° and 76% (129/171) in the central 10°. Proportions ob-tained using the center of mass as parameter of RF ec-centricity are roughly similar: 22% (38/171) of the cellshave their center of mass in the central 3°, 40% (68/171)in the central 5° and 71% (121/171) in the central 10°(Fig. 4). No significant bias of upper visual field with re-spect to lower visual field could be noted in the presentdata collection.

RF size

The RF surface was taken as that portion of the RF forwhich neuronal discharge exceeds half of the maximumof the RF structure (i.e., the mean of the discharge of theneuron for a stimulus situated at its optimal response po-sition inside the RF). The width is estimated by thesquare root of this surface. On average, 9% (13/171) ofthe neurons have a width smaller than 5°, 42% (70/171)have a width smaller than 10°, and 67% (107/171) havea width smaller than 15° (mean = 12.8°, SD = 6.7°, min.= 1.8°, max. = 38.2°).

133

Table 2 Characteristics of the recorded LIP population (N=313)during a memory-guided saccade task: characteristics of neuronsexpressing a significant delay-period activity in the task (N=111)

Visual and Visual Saccadic saccadic only only

Delay 104 4 3

Fig. 4 Spatial distribution of peaks (open diamonds positive ab-scissas) and centers of mass (filled circles negative abscissas) forall mapped RFs (N=171). Limits of central 3°, 5° and 10° are indi-cated by lines. Peaks are rectified so that all the abscissas are posi-tive. Centers of mass are rectified so that all abscissas are negative

Relation between RF size and eccentricity

RFs can be best characterized by the combination of cen-ter of mass, peak and width measures. This latter param-eter is known for many visual areas to vary as a functionof eccentricity of RFs, increasing from fovea to periph-ery. A linear regression graph of width as a function ofeccentricity of the center of mass shows that such a trendexists in area LIP (Fig. 5). The line through the graphrepresents the best linear fit obtained with a least squaresregression method. The 0.69 correlation value betweenthe two variables indicates that 47% of the variability ofthe width of the RFs is a function of the center of mass.This linear fit accounted better for the relation betweensize and location than a logarithmic one, which ex-plained only 40% of the variance.

A similar analysis carried out on the eccentricity ofthe peak yields a regression line whose slope is 0.58, andwhose y-intercept is 8.7, but with a smaller, 0.54 correla-tion value. Thus, in this case, 29% of the variability ofthe width of the RFs is a function of the peak (data notshown). As for the center of mass, a linear fit accountedbetter for the relation between size and location of peakthan a logarithmic one (23% of the variance). For com-parison, this is quite close to the parameters obtained forthe same regression analysis carried out on area MT RFs,where slope = 0.52, y-intercept = 5.88 and correlationcoefficient = 0.48 (values calculated from data kindlyprovided by S. Raiguel and G. Orban).

RF asymmetry

A commonly made assumption is that RFs are roughlyradially symmetric and can be approximated by a contin-uous function such as a two-dimensional gaussian. In or-der to determine whether this is a valid assumption forarea LIP RFs, the degree of asymmetry of the RF as re-flected in the distance between peak and center of masscan be estimated as a first approximation. This analysiswas prompted by the observation that although most LIPRFs had smooth, continuous shapes, they were not nec-essarily monotonic. We thus calculated the distance Dbetween the peak and the center of mass for the completeset of data. For each cell we set as the limit of experi-mental significance of D the step used for projecting thestimuli on the screen during the RF mapping procedure.This implies that only cells whose RF surface (the regionwhere firing was about 50% of the peak discharge) wascovered by an array larger than 3×3 mapping pixels wereincluded. The majority of cells (80%) showed little or noasymmetry by this criterion. Thirty-three cells of the 171(19%) had a D value equal to or larger than one step ofthe mapping grid. An example of asymmetrical RF isshown in Fig. 6A. This subset of 33 units had RF centersthat were distributed evenly over the visual field repre-sentation. Since the size of D showed a significant posi-tive correlation with RF width (R=0.77), an asymmetryindex (AI) was defined in order to normalize the distancebetween peak and center of mass with respect to RF size.AI was calculated as follows:

AI=[abs (distance between pk and cm in °)/=(width in °)]×100

A null AI indicates that peak and center of mass fall onthe same location. An AI equal to 50% indicates that thedistance between peak and center of mass is half the ex-tent of the RF. AI ranged between 26% and 77% (mean= 43.8%, SD = 15.3%). No correlation was found be-tween the position of center of mass and AI values (R =0.003, Fig. 6B). Therefore a markedly asymmetric RFshape is not mainly a property of large, eccentric RFs butcan also be observed for smaller RFs of the central re-gion of the visual field.

The asymmetry index is based on the absolute valueof distance between peak and center of mass, but doesnot take into account their relative position. The RF il-lustrated in Fig. 6A is that of a cell recorded from a righthemisphere. Its main peak is located on the horizontalmeridian and extends into the upper visual periphery andinto the ipsilateral field. The RF is not mapped com-pletely but it is clear that the inferior border is muchmore sharply defined than the superior border. Conse-quently, it has an asymmetric shape in which the centerof mass is displaced eccentrically relative to the peak(Fig. 6A, right panel). This shape is characteristic of“open ended” RFs which seem to lack well-defined pe-ripheral borders. About one-third of the cells (12/33)with an asymmetric RF profile fell into this category.Another group of cells (15/33) had clearly defined con-

134

Fig. 5 RF width in degrees plotted as a function of the eccentrici-ty of the center of mass in degrees (continuous line linear fit isy=0.9x+5.482; within dashed lines confidence interval around thefit estimate)

tours in all directions but showed a similar type of asym-metry, with the center of mass displaced eccentricallyrelative to the peak. The remaining cells (6/33) fell intoneither of the two previous categories and could in partbe accounted for by an artifact of slightly incomplete da-ta sets.

Because asymmetries below the mapping resolutionare not taken into account, the previous analysis underes-timates the frequency of this phenomenon. However, itsimportance shows at the level of the population in thedifference in the slopes of the regression lines describingon the one hand the relation between size and eccentrici-ty of the center of mass (slope = 0.9) and on the otherhand the relation between size and the eccentricity ofpeak (slope = 0.58) (see previous section).

Overrepresentation of the fovea in LIP

In order to establish how the visual field was representedat the cortical level, we calculated the percentage of cellswhose RF enclosed each squared degree of the visualfield. This approach can be conceptualized as an alterna-tive to a direct measure of the cortical magnification fac-tor, which, for areas that have clearly defined bordersand a strong topographical organization, is obtained byestimating how much cortical tissue is dedicated to eachportion of the visual field. It is important to note that thecomputed representation not only reflects the distribu-tion of RF centers, but also takes into consideration in-formation about RF size, which we previously definedstatistically as the region of the RF map for which thedischarge rate is above half of the peak discharge of theneuron. A prerequisite of this approach applied to a

135

AFig. 6 A A cell with an asym-metrical RF. Left panel showsthe two-dimensional RF map incolor code. Right panel showsa vertical section through theRF at the level indicated by thevertical line on the RF map.Location of peak (star) and ofcenter of mass (circle) are indi-cated above. B Distribution ofthe AI index as a function ofRF eccentricity for the subsetof 33 cells with a significant Dvalue

strictly topographically organized area is that RFs areuniformly sampled. In a less well organized area, theuniformity criteria become less crucial (and harder tocontrol) as close RFs tend to have different RF proper-ties. Figure 7A shows a composite representation ob-tained from left and right hemisphere data, the latter be-ing rectified (reflected) on the former. Thus, for the hori-zontal axis, negative values are ipsilateral and positivevalues contralateral. The obtained distribution is bellshaped and is centered near the zero of the visual space.This can also be observed in the cross-section of Fig. 7Bwhere, for comparison purposes, is also shown the celldensity distribution for an area immediately adjacent toLIP, the ventral intraparietal area (VIP, computed fromdata from Duhamel et al. 1997). Several points can benoted: (1) in LIP, more cells are involved in the analysisof the central space than with that of the peripheralspace. Half of the cells represent the central 6°. By con-trast, in VIP, half of the cells represent a region extend-ing up to 20° around the fovea. (2) In LIP, the portion ofvisual field extending from 2° to 10° of eccentricity canbe approximated by a line of slope of –4.5%–1. Thismeans that for each visual angle displacement to the pe-riphery, the proportion of cortical visual resources dedi-

136

Fig. 7 A Three-dimensional representation of the percentage ofcell density in LIP coding for a given eccentricity of the visualscene. The representation is constructed from the RFs of the inves-tigated population (N=171). Cells from the right hemisphere arereflected on the left hemisphere. Reflected data from right hemi-sphere and data from left hemisphere are cumulated. Thick blacklines indicate the location of vertical and horizontal meridians.B Representation of visual space in terms of cell percentage cod-ing for a given degree of visual angle in area LIP (dark gray) andin area VIP (light gray). Calculations are based on rectified valuesso that all neurons are taken as neurons recorded from the lefthemisphere

Fig. 8A–E Histological reconstruction of recording sites and cor-relation of physiological data. A, B, C and E show the data ob-tained from a right hemisphere and D the data obtained from a lefthemisphere. A Upper left panel is a macaque brain in which theintraparietal and lunate sulci have been unfolded. Upper rightpanel is an enlarged view of the unfolded parietal sulcus. On thisview have been placed the limits of LIP (dark gray) and VIP (lightgray) as determined from a correlation between histology andphysiology. The two peripheral vertical lines on each side of LIPrepresent the limits of our recording chamber. The three middlevertical lines labeled sl1, sl2 and sl3 represent the level at whichthe three coronal sections sl1, sl2 and sl3 presented on the lowerpart of the figure are taken. On each coronal section, LIP is shadedin gray. On sl2 and sl3 the limits of VIP are plotted. On sl2, twoelectrode penetrations have been reconstructed. B The recordingsites of cells for which the RF has been mapped are plotted on aflattened intraparietal sulcus as reconstructed from the histologicalmaterial. Posterior is to the left, anterior is to the right, medial isup and lateral is down. The straight horizontal line is the funduswith respect to which the limit of the convexity is calculated. Theborder between sulcus and convexity is indicated by the lowergray curve. The dense myelin zone as determined from Schmued-stained sections is shown in gray. It corresponds to subdivisionLIPv. The light gray area shows the uncertain boundary betweenLIPv and LIPd. The locations of the slices sl1, sl2 and sl3 shownin A are positioned on the flat map. C Schema summarizing thecoarse topography of LIP. The flat reconstruction of the intraparie-tal sulcus is identical to that of B. Concentric arcs represent in-creasing eccentricities from the fovea (tails of the three arrows)towards the periphery (head of each arrow). D Flattened intrapari-etal sulcus as reconstructed from the histology. Anterior is to theleft, posterior to the right, medial is up and lateral is left. Thestraight horizontal line is the fundus and the gray line the convex-ity. Recording sites are plotted. Symbols are as in B. Coronal sec-tions at levels sl1, sl2 and sl3 are plotted below. LIP as identifiedfrom our recordings is plotted in gray on sl1. Three electrodetracks are also reconstructed. E The recording sites of cells whichhave a delay activity (o) or no delay activity (+) on a memoryguided saccade task are reported on the same flattened intraparie-tal sulcus representation as in B

▲

137

cated to this space is decreased by 4.5%. For compari-son, the linear portion of the cortical magnification fac-tor of primary visual cortex V1 corresponds to a de-crease in the visual resources from fovea to periphery of28% per degree (from Tootell et al. 1988). (3) Area LIPmainly codes the contralateral space, and the ipsilateralrepresentation drops sharply beyond 5° across the verti-cal meridian. Twelve percent of the cells had a center ofmass, i.e., most of their surface, located on the ipsilateralside of the vertical meridian (mean eccentricity of RFcenters in this ipsilateral subpopulation is 3°). By con-trast, VIP has a rather extended ipsilateral representation,with 20% of RFs which are mainly confined within theipsilateral field (mean = –14.42, SD = 6.38, range from–5 to –22.5).

Topographical organization of LIP

Histological verification has been carried out on bothhemispheres of one monkey, while the other monkey isstill used in ongoing experiments. Reconstructions weremade from digitized thionine-stained coronal sections.Figure 8A shows a schematic representation of the loca-tion of LIP relative to the intraparietal sulcus and the ad-jacent identified area VIP as reconstructed from the dataof a second electrophysiological experiment conductedin parallel on the same animal (Duhamel et al. 1997).The reconstructed surface for VIP is consistent with pre-vious estimates of the physiologically defined area VIPby Colby et al. (1993). Three sections are shown and ineach of them, the extent of LIP is shaded in dark gray,with two representative electrode penetrations. A flat-tened reconstruction of the lateral bank of the right intra-parietal sulcus was obtained after aligning the individualsections with respect to the fundus of the sulcus(Fig. 8B). The anterior part of the sulcus is on the rightand the posterior part is on the left, where the intraparie-tal and lunate sulci join. The estimated sites of recordingof the neurons described in the present study and record-ed from the right hemisphere are plotted on this flat map.A distinction is made between cells with RF within andbeyond the central 7° and within the upper and lower vi-sual fields. The progression of RF properties along pene-trations indicated by ‘a’ and ‘b’ are detailed in Fig. 9.Figure 9A shows a progression from relatively smallfoveal RFs to larger RFs mainly encompassing the uppervisual field, to RFs mainly encompassing the lower visu-al field. Figure 9B shows a progression from a smallcentral RF to larger RFs mainly encompassing the lowervisual field or centered on the horizontal meridian. Nostrict topography can be seen, but some rough systematictrends are observed: central RFs are located more dorsal-ly and peripheral RF more ventrally in the sulcus. In theanterior part of LIP, the RFs are mainly situated in theupper hemifield, the closest to the horizontal meridianbeing the deepest. More posteriorly, the extent of the up-per hemifield representation diminishes while a repre-sentation of the lower hemifield appears in the deepest

region of LIP and increases in size as one moves towardsthe posterior border of LIP (Fig. 8B). The upper andlower representations appear to be separated by cellswith RFs close to the horizontal meridian, but with ec-centricities running from the fovea to 15°. Although thisis suggestive of a continuous representation of verticaleccentricities in LIP, it should be considered with cautionas the meridial category is represented by a relativelysmall sample of neurons. In the left hemisphere most ofthe recordings were concentrated in the anterior LIP. It isinteresting to note that the region of the visual fieldwhich was mapped in that hemisphere was obtainedfrom a recording location which matched that of the ana-logue region in the right hemisphere, relative to land-marks such as sulcus length and anteroposterior location(Fig. 8D; level of sl1 of inverted Fig. 8D matches thelevel of sl2 of Fig. 8B). In the third hemisphere, forwhich no histology is yet available, the explored corticalportion yielded RFs with average RF size of 11.5 (SD =5.26, 81.25% of the cells having an eccentricity exceed-ing 7). Thus, the overrepresentation of the central visualfield is less pronounced in this hemisphere than in thetwo previous ones. This could be explained by a non-ho-mogeneous sampling in this chamber. Alternatively, wecould have missed the most anterior portion of LIP be-cause of the chamber location. We favor this latter inter-pretation as LIP recording sites were confined in the an-terior two-thirds of the chamber. Presaccadic delay activ-ity which is characteristic of LIP saccadic neurons isevenly distributed along all the studied extent of LIP, in-termingled with non-delay cells (Fig. 8E).

The upper and lower limits of the dense myelin zone(DMZ) were obtained from Schmued-stained coronalsections and plotted on the flat cortical representation ofthe intraparietal sulcus of Fig. 8B. The ventral myelinboundary on the lateral bank of the intraparietal sulcushas been identified as the site of LIP/VIP border on ana-tomical (Blatt et al. 1990) and physiological (Colby et al.1993) grounds. The upper boundary is the limit betweenventral LIP (LIPv, adjacent to VIP) and dorsal LIP(LIPd, adjacent to the convexity and to 7a). The DMZdefines LIPv. The putative progressive transition zonebetween LIPv and LIPd is represented in lighter shades.A segregation of cells is observed between these twosubdivisions of LIP, on the basis of the eccentricity oftheir RF. Cells with most foveal RFs are mainly found inLIPd whereas cells with more eccentric RFs are situateddeeper in LIPd.

Discussion

The results that are reported in the present study providenew information about the representation of the visualfield in area LIP using an automated mapping procedureof the RF properties. Its accuracy and repeatability sug-gest that this procedure can be extended to behavioralcontexts other than a standard central color discriminationtask. The full information carried by the RF of neurons is

138

available, and post hoc descriptive parameters characteriz-ing the RF can be calculated in order to quantitatively de-scribe the visual properties of the studied population.

Reliability of the mapping procedure

The limiting factor of this experimental design is timeand stability of the neural response. Thus, the initial partof the study consists of determining the optimal condi-

tions which enabled us to minimize both the mappingtime and the error on RF characteristics determination.The post hoc calculated variability sets the limits ofthis quantitative approach, and provides landmarks forinter- and intracell comparisons. Because of the ran-dom sequential projection of the stimuli on the screenfrom one trial to another, during the mapping sessions,both variation due to noise and baseline variations dueto ongoing behavior are evenly distributed over all thestimulations at different spatial positions.

139

Fig. 9A, B Reconstruction ofRF progression along two par-allel electrode penetrations.Numbers identify visual RFs ofcells which were successivelyencountered during a singlepenetration at the depth indicat-ed on the left diagram. A Pene-tration labeled a in Fig. 8B. Forclarity, RFs recorded at loca-tions 2, 4, 6 and 8 are notshown. The RF recorded in po-sition 2 is very close to that re-corded in 1, the one recorded in4 is close to that recorded in 5,and those recorded in 6 and 8are close to that recorded in 9.B Penetration indicated by b inFig. 8B

RF size, eccentricity and asymmetry in LIP,a cross-study perspective

In the present study, RFs were found to range from 0° to28° in eccentricity, and from 1.8° to 38.5° in size. Thedifferent eccentricities were not equally represented atthe level of the population, and the mean of the eccen-tricities of the RFs we collected (7.26°) indicates amarked overrepresentation of the central visual field.This leads us to compare our results with a previous de-scription of the visual field representation in area LIP.

Blatt and colleagues (1990) studied the RF character-istics of LIP neurons in anesthetized monkeys. In theirstudy, the eccentricity of LIP RFs ranged from about 7°to 50° with a homogeneous pattern of distribution, andRF size varying between 8° and 35°. The linear regres-sion of size and eccentricity was best fitted by a line ofequation of the type Size = 0.3 × Eccentricity + 11.2°,with RF eccentricity accounting for 21% of the variabili-ty of RF size. In the present study, a stronger relationshipwas found between these two parameters, which is ex-pressed both by a steeper slope (0.9, Fig. 5) and by alarger amount of variance in size being accounted for byeccentricity (47%). When eccentricity is no longer takenas being the position of the center of mass, but the posi-tion of the peak, a somewhat weaker dependency of sizeupon eccentricity is found (Size = 0.58 × Eccentricity +8.7°, explaining 29% of the variance). These two fits aresubstantially different from each other, although the sec-ond is closer to that reported in the above-cited study.These differences could be interpreted as follows: (1)The intercept is a measure related to both the spatial ex-tent of the most central RFs and to the precision in themeasure of the eccentricity of these RFs. We found thatthe center of mass is a more robust estimate of spatialposition of RFs than peak position. This parameter can-not be determined by hand-mapping, during which weexpect the mapping to be rather centered around thepeak. (2) The present mapping procedure provides uswith a high mapping resolution of small foveal RFs,which tend to have sharp borders, and with an objectivemeasure of RF size across all eccentricities. The steeperregression slope and the lower intercept in our studymean that we found a relatively greater proportion ofsmall central RF and of large peripheral RF than Blatt etal. This raises the possibility that the necessarily subjec-tive criteria used during hand mapping could lead both toan overestimation of the size of central RFs and to an un-derestimation of that of eccentric ones. (3) Receptivefield characteristics might be substantially different inanesthetized and awake behaving animals. Attentive fix-ation on a central target biases the visual analysis re-sources towards the foveal space (Ben Hamed et al.1996), a hypothesis that is explored in a separate study.The above analysis shows that data on RFs presented indifferent reports are highly conditional on the mappingcontext, and particular attention should be paid to the in-terpretation of such data while making cross-study com-parisons.

Visual response fields have also been studied by Plattand Glimcher in two independent studies (Platt andGlimcher 1997, 1998). In the first study, the authors com-pared the visual response of LIP neurons with saccadictargets or with visual distracters. The small cell sampleand the task conditions render the comparison with thepresent study difficult. In the study of 1998, they addressthe representation in LIP of the response of the neurons tothe onset of visual saccadic targets. Three factors renderthe comparability with the present work difficult. First,the visual response is assessed in the context of the prepa-ration of an eye movement, and both attentional and mo-tor preparation factors are likely to modify the profile ofthe recorded RFs. Second, visual responses are averagedover a window of 200 ms set on the onset of the saccadictarget, whereas in the present study the window of analy-sis of the visual response is adapted to the latency andphasic component of LIP neurons. Finally, spatial resolu-tion is favored by the authors and maps are constructedalmost on single trial values. We are not sure that such amethod yields a more reliable quantification of RF spatialproperties than hand mapping methods.

Our quantitative mapping approach allows examina-tion of the substructure of an RF in terms of the relationbetween its peak, i.e., the location of the area of maxi-mum activity or “hot spot,” and the center of mass of theactivity over its whole surface. Mapping studies in anes-thetized animals provide precise enough data to supportmathematical modeling of RF structure in early visualprocessing (Jones and Palmer 1987a, 1987b; Wörgötterand Holt 1991) based on gaussian or Gabor functions. Asimilar attempt has been carried out in area MT, which isa higher visual area projecting onto LIP (Raiguel et al.1995), as well as in area LIP itself (Platt and Glimcher1998). In area MT, RFs are often elongated and havebeen modeled by fitting oriented elliptical functions. ForLIP RFs, which are typically not elongated, we assessedthe validity of the assumption of a symmetrical structureusing a simple index based on the distance between peakand center of mass. For RFs that are radially symmetri-cal, these two parameters coincide. In our study the ma-jority of cells display a degree of asymmetry that did notexceed the nominal step of the mapping grid. For in-stance, if the cell’s RF was mapped with an array ofstimuli that were separated by 4°, we considered the RFas symmetrical if the distance between peak and centerof mass did not exceed this value. In about one-sixth ofthe studied cell population, RFs depart markedly from ageneral gaussian shape, with distances between peak andcenter of mass of up to 77% of the RF width. Because ofthe criterion used, this might constitute a lower estimateof the frequency of asymmetric RF shapes. Indeed, theubiquity of this phenomenon is revealed by the large dif-ference between the regression slopes of RF size andcenter of mass on the one hand, and of RF size and peakeccentricity on the other hand. This asymmetry suggeststhat the interpretation of the shape of the RF in terms ofpurely visual horizontal and vertical connectivity maynot be sufficient. One question raised by these findings is

140

whether the RF shape is a fixed characteristic of neuronsor whether the spatial distribution of activity within anRF can be modulated dynamically by behavioral vari-ables. Previous studies have interpreted the effects of se-lective attention in extrastriate visual cortex as a modula-tion of the sensitivity of different subregions of a neu-ron’s RF, with visual stimuli being filtered in or out de-pending on their relevance to the ongoing cognitive task(Moran and Desimone 1985). It would be important todetermine whether similar mechanisms are also presentin LIP neurons since their activity is also strongly modu-lated by the behavioral significance of visual stimuli(Colby et al. 1996; Gottlieb et al. 1998).

Representation of the visual field in LIP

Representation of the ipsilateral space

Most cortical representations of the visual field are hemi-field representations that include a large portion of thespace contralateral to the hemisphere of the consideredarea and a more or less restricted region of the verticalmeridian and of the near ipsilateral visual field. In areaV1, this is explained by the existence of cells located atthe V1/V2 border (Tootell et al. 1988) with RFs centeredon the vertical meridian and whose surface therefore ex-tends partly into the ipsilateral field. The representationof the visual space of LIP extends up to about 5° in theipsilateral hemifield. This larger ipsilateral extension can-not only be explained by the larger size of LIP RFs withrespect to V1, since many LIP neurons are not centeredon the vertical meridian but have a center of mass whichcan extend as far as 15° into the ipsilateral field (mean =–5.7°, SD = 4.39°). By comparison, the ipsilateral repre-sentation in MT extends to 10° and in MST to 30–40°(Raiguel et al. 1997). In area VIP, the ipsilateral represen-tation also extends quite eccentrically, to about 40° (com-puted from the data of Duhamel et al. 1997). Thus, therepresentation of the ipsilateral visual field is much morerestricted in area LIP than in the other extrastriate dorsalstream visual areas which have been investigated so far.

Representation of the visual space

Previous studies have described a mostly peripheral vi-sual representation in LIP (Blatt et al. 1990) as well as arelative overrepresentation of upper contralateral spacein LIP with respect to lower contralateral space (Li andAndersen 1994). It is not clear why this asymmetryshould exist in a major cortical saccadic area. Platt et al.describe a more extensive visual representation extend-ing into the ispilateral visual field (Platt and Glimcher1998). Here, we describe a symmetrical across the hori-zontal meridian, contralateral visual field representationand a small representation of the ipsilateral visual field.The discrepancy between the present study and the earli-est observations can be accounted for by the following.

Blatt et al. defined LIP as the most posterior third ofthe lateral bank of the intraparietal sulcus, with a highproportion of neurons presenting a delayed response ac-tivity in the memory guided saccade task, and projectingto the frontal eye field (FEF). Our recordings cover theposterior half of the intraparietal sulcus and thus both en-compasses the area Blatt et al. recorded from and ex-tends beyond it. Visual and saccadic activities are pres-ent along this entire portion. Moreover, the delay activityon the memory guided saccade task, which is thought tobe a robust marker of LIP, is also evenly distributed inthis region (Fig. 8E). This extension of LIP up to the halfof the lateral bank of the intraparietal sulcus completesthe visual representation of this area from an area em-phasizing the upper contralateral visual field to an areaequally representing both the upper and lower visualfield (note on Fig. 8C that the lower field representationis situated more anteriorly than the upper field represen-tation). This leads us to suggest extending the definitionof LIP to all this portion of the intraparietal sulcus.About a third of this newly defined LIP area is dedicatedto the analysis of the central visual field (Fig. 8C).

Overrepresentation of the central space

In area LIP, about 50% of the neurons were included inthe analysis of zero eccentricity (Fig. 7). As we move by1° to the periphery in any direction of the visual field,the amount of resources allocated to the analysis of agiven point of the visual field is decreased by 4.5%. Adiscontinuity is observed around 15° eccentricity, atwhich point the percentage of decrease in the resourcesallocated to the analysis of a given degree of spacechanges to 1.8. An important asymmetry thus exists be-tween the representation of the center of the visual fieldin LIP and its periphery. It could be argued that this un-expected observation is in part due to a selection bias to-ward cells with RFs that could be mapped during a fixa-tion task. Although we in fact only report neurons withwell-defined RFs, we did not select them a priori on thisbasis, as neurons were characterized on isolation with abattery of tests that included both fixation and saccadetasks. Two interpretations remain: One is that this is asampling bias that would have missed a cluster of neu-rons with peripheral RFs. The second is that LIP repre-sentation of space itself has a particular emphasis oncentral visual space. We favor the second hypothesis forthe following reasons. Although in one of the two histo-logically reconstructed hemispheres, we did not explorethe entire extent of the intraparietal region correspondingto area LIP and may have oversampled from the centralvisual field representation, analysis of the other hemi-sphere confirms that neurons were sampled in a ratherhomogeneous way along the sulcus (the penetrationssystematically run from the surface to the depth of thesulcus, Fig. 8A, D, and if no visual responses are report-ed in some cortical location, it is mainly due to the factthat none could be elicited).

141

Interestingly, when the cell density representation inLIP is compared with that of VIP (Fig. 7B) or with thatof the MT and MST (data not shown, analysis carriedout on the data published by Raiguel et al. 1995 andRaiguel et al. 1997) as well as with the reported magnifi-cation factor in V1 (Tootell et al. 1988), LIP appears tobe intermediate between strongly magnified (V1) andstrongly demagnified (MST, VIP) areas, but similar toarea MT, even though the cell density representation ofLIP appears to be somewhat displaced towards the centerof the visual field. A full comparative study of the celldensity representation in dorsal and ventral visual areasmay be needed to address the issue of how the visualfield is represented in each of these areas.

Topography: a physiological basis for distinguishingLIPd and LIPv?

As in many higher-order extrastriate visual areas, LIPdoes not appear to contain a continuous and orderly reti-notopic organization. Across nearby and successive pen-etrations, neurons seemed to be organized in clusters ofcells with similar visual and oculomotor properties, butsharp transitions from one cluster to another could be ob-served. However, a coarse topography can be seen whichconsists of a broad representation of the horizontal me-ridian running from the upper anterior border of LIP tothe deeper posterior border, the lower field representa-tion lying above this region, and the upper field repre-sentation lying below it, RFs moving away from the hor-izontal meridian as they move away from the limit of theupper and lower representation. This is the same trend asthat described by others (Blatt et al. 1990). Midwaythrough this representation of the horizontal meridian, afocus of more strictly foveal cells can be identified. Thiscentral visual field focus has been identified in both leftand right hemispheres in corresponding portions of theintraparietal sulcus.

LIP has been subdivided on the basis of myeloarchi-tecture (Blatt et al. 1990; Schall et al. 1995; Bullier et al.1996) into: (1) a ventral portion, LIPv, running parallelto area VIP in the fundus of the intraparietal sulcus, de-fined by the dense myelin zone (DMZ), which is foundon the lateral bank of the intraparietal sulcus, and (2) adorsal portion LIPd running between LIPv and area 7a.Interestingly, it has also been shown that these two sub-divisions of LIP share connection pathways with visualareas (V3, V3A, V4, MST, 7a, VIP, PO; Blatt et al. 1990;Schall et al. 1995) as well as with prefrontal areas (46,45 and 8a; Cavada and Goldman-Rakic 1989; Schall etal. 1995; Bullier et al. 1996) and subcortical areas (later-al pulvinar, anterior pretectal nuclei and deeper layers ofthe superior colliculus (Asanuma et al. 1985; Lynch andGraybiel 1985). However, each subdivision also hasunique or more specific connections with certain corticalareas. For example, LIPv is connected to the dorsalstream area MT (Ungerleider and Desimone 1986; Blattet al. 1990), whereas LIPd has more specific connections

with the ventral stream areas TEa, IPa, Id, TEm and witharea 24a (Blatt et al. 1990; Schall et al. 1995; Bullier etal. 1996). Regarding the connections of LIP to the fron-tal lobes, LIPv projects densely towards 8a (medial FEF)and areas 45 (lateral FEF) and 46, with an anteroposteri-or gradient (anteromedial LIPv is connected to lateral 8aand to medial 46, whereas posterior LIPv is connected tomedial 8a) (Blatt et al. 1990; Schall et al. 1995). LIPdhas denser connections with areas 45 and 46 and onlysparse connections with area 8a, but, overall, the connec-tions of LIPd with each of these three areas are weakerthan those of LIPv (Blatt et al. 1990; Schall et al. 1995;Bullier et al. 1996).

Visual analysis and visual exploration are functionallyrelated to ocular fixation and saccadic eye movements,respectively, and several subcortical and cortical areashave been shown to participate in both motor behaviors:brainstem, (Keller 1991); superior colliculus (Munoz andWurtz 1993); substantia nigra (Hikosaka and Wurtz1989); frontal eye field (Seagraves and Goldberg 1987);and supplementary eye field (Bon and Lucchetti 1992).The FEF is considered to be the final stage of corticalprocessing for visually guided purposive saccades(Bruce and Goldberg 1985; Dassonville et al. 1992). Fix-ation related activities have also been identified in theFEF and more specifically in the lateral FEF (Hanes etal. 1998). The specific visual and prefrontal connectionsof LIPd and LIPv as well as the specific projections oftheir projecting areas suggests a relative functional seg-regation between these two subregions of LIP. This issupported by a relative segregation of cells on the basisof their visual properties: cells with more central RFs arelocalized in LIPd whereas cells with more peripheralRFs are found in LIPv. Because of the overlap betweenthe preferred visual and oculomotor directions of LIPcells (Barash et al. 1991a), we would expect a relativesegregation of the oculomotor cell properties, with cellsin LIPd having activities in relation to fixation or tosmall saccades, and cells in LIPv being concerned withsaccades of larger amplitude. We would also expect that,by virtue of its connections with ventral stream visual ar-eas, LIPd could possibly specifically receive informationabout stimulus form and color, and thus provide a func-tional bridge between visual analysis and visual explora-tion. Recently reported electrical microstimulations ofthe intraparietal area do not provide enough evidence toclarify this point (Thier and Anderson 1998). The datapresented here thus corresponds, to our knowledge, tothe first electrophysiological evidence for the existenceof two segregated subdivisions of LIP, corresponding tothe previously anatomically described LIPd and LIPv.More in depth studies combining an electrophysiologicalapproach and a precise analysis of the cortical and sub-cortical projections of LIP are needed in order to addressthis issue.

Acknowledgements This work was supported by grants from theEuropean Community (HCM: ERBCHRXCT930267) and theHuman Frontier Science Program (RG71/96B).

142

References

Andersen RA, Essik G, Siegel RM (1985) Encoding of spatial lo-cation by posterior parietal neurons. Science 230:456–458

Andersen RA, Asanuma C, Essik G, Siegel RM (1990a) Cortico-cortical connections of anatomically and physiologically de-fined subdivisions within the inferior parietal lobule. J CompNeurol 296:65–113

Andersen RA, Bracewell RM, Barash S, Gnadt JW, Fogassi L(1990b) Eye position effects on visual, memory, and saccade-related activity in areas LIP and 7a of macaque. J Neurosci10:1176–1196

Asanuma C, Andersen RA, Cowan WM (1985) The thalamic rela-tions of the posterior inferior parietal lobule and the lateralprefrontal cortex in monkeys: divergent cortical projectionsfrom cell clusters in the medial pulvinar nucleus. J Comp Neu-rol 241:357–381

Baker JV, Petersen SE, Newsome WT, Allman JM (1981) Visualresponse properties of neurons in four extrastriate areas of theowl monkey (Aotus trivirgatus): a quantitative comparison ofmedial, dorsomedial, dorsolateral and middle temporal areas. JNeurophysiol 45:397–416

Barash S, Bracewell RM, Fogassi L, Gnadt JW, Andersen RA(1991a) Saccade-related activity in the lateral intraparietal ar-ea. II. Spatial properties. J Neurophysiol 66:1109–1124

Barash S, Bracewell RM, Fogassi L, Gnadt JW, Andersen RA(1991b) Saccade-related activity in the lateral intraparietal ar-ea. I. Temporal properties; comparison with area 7a. J Neuro-physiol 66:1095–1108

Ben Hamed S, Duhamel J-R, Bremmer F, Graf W (1996) Dynamicchanges in visual receptive field organization in the macaquelateral intraparietal area (LIP) during saccade preparation. SocNeurosci Abstr 22:635.12

Ben Hamed S, Duhamel J-R, Bremmer F, Graf W (1997a) Atten-tional modulation of visual receptive fields in the posterior pa-rietal cortex of the behaving macaque. In: Thier P, KarnathH-O (eds) Parietal lobe contribution to orientation in 3Dspace. Springer-Verlag, Berlin, pp 371–384

Ben Hamed S, Duhamel J-R, Bremmer F, Graf W (1997b) Repre-sentation of the visual field in the macaque lateral intraparietalarea (LIP) Soc Neurosci Abstr 23:604.2

Blatt GJ, Andersen RA, Stoner GR (1990) Visual receptive fieldorganization and cortico-cortical connections of the intraparie-tal area (area LIP) in the macaque. J Comp Neurol299:421–445

Bon L, Lucchetti C (1992) The dorsomedial frontal cortex of themacaca monkey: fixation and saccade related activity. ExpBrain Res 89:571–580

Bremmer F, Distler C, Hoffmann K-P (1997) Eye position effectsin monkey cortex. II. Pursuit- and fixation-related activity inposterior parietal areas LIP and 7a. J Neurophysiol 77:962–977

Bremmer F, Pouget A, Hoffmann K-P (1998) Eye position encod-ing in the macaque posterior parietal cortex. Eur J Neuro-physiol 10:153–160

Brotchie PR, Andersen RA, Snyder LH, Goodman SJ (1995) Headposition signals used by parietal neurons to encode locationsof visual stimuli. Nature 375:232–235

Bruce CJ, Goldberg ME (1985) Primate frontal eye fields. I. Sin-gle neurons discharging before saccades. J Neurophysiol53:603–635

Bullier J, Schall JD, Morel A (1996) Functional streams in occip-ito-frontal connections in the monkey. Behav Brain Res76:89–97

Bushnell MC, Goldberg ME, Robinson DL (1981) Behavioral en-hancement of visual responses in monkey cerebral cortex. I.Modulation in posterior parietal cortex related to selective vi-sual attention. J Neurophysiol 46:755–772

Cai D, DeAngelis GC, Freeman RD (1997) Spatiotemporal recep-tive field organization in the lateral geniculate nucleus of catsand kittens. J Neurophysiol 78:1045–1061

Cavada C, Goldman-Rakic PS (1989) Posterior parietal cortex inrhesus monkey: II. Evidence for segregated corticocorticalnetworks linking sensory and limbic areas with the frontallobe. J Comp Neurol 287:422–445

Colby CL, Gattass R, Olson A, Gross CG (1988) Topographicalorganization of cortical afferents to extrastriate visual area POin the macaque: a dual tracer study. J Comp Neurol 269:392–413

Colby CL, Duhamel J-R, Goldberg ME (1993) Ventral intraparie-tal area of the macaque: anatomic location and visual responseproperties. J Neurophysiol 69:902–913

Colby CL, Duhamel J-R, Goldberg ME (1996) Visual, presaccadicand cognitive activation of single neurons in monkey lateralintraparietal area. J Neurophysiol 76:2841–2852

Connor CE, Gallant JL, Preddie DC, Van Essen DC (1996) Re-sponses in area V4 depend on the spatial relationship betweenstimulus and attention. J Neurophysiol 75:1306–1308

Crist CF, Yamasaki DSG, Komatsu H, Wurtz RH (1988) A gridsystem and a microsyringe for single cell recording. J Neuro-sci Methods 26:117–122

Dassonville P, Schlag J, Schlag-Rey M (1992) The frontal eyefield provides the goal of saccadic eye movements. Exp BrainRes 89:300–310

DeAngelis GC, Ohzawa I, Freeman RD (1993a) Spatiotemporalorganization of simple-cell receptive fields in the cat’s striatecortex. II. Linearity of temporal and spatial summation. J Neu-rophysiol 69:1118–1135

DeAngelis GC, Ohzawa I, Freeman RD (1993b) Spatiotemporalorganization of simple-cell receptive fields in the cat’s striatecortex. I. General characteristics and postnatal development. JNeurophysiol 69:1091–1117

Duhamel J-R, Colby CL, Goldberg ME (1992) The updating ofthe representation of visual space in parietal cortex by intend-ed eye movements. Science 225:90–92

Duhamel J-R, Ben Hamed S, Bremmer F, Graf W (1995) The in-fluence of attention and intended eye movements on the visualexcitability of neurons in the lateral intraparietal area of ma-caque monkeys. Soc Neurosci Abstr 21:693.5

Duhamel J-R, Bremmer F, Ben Hamed S, Graf W (1997) Spatialinvariance of visual receptive fields in parietal cortex neurons.Nature 389:845–848

Felleman DJ, Van Essen DC (1991) Distributed hierarchical pro-cessing in the primate cerebral cortex. Cereb Cortex 1:1–47

Gnadt JW, Andersen RA (1988) Memory related motor planningactivity in posterior parietal cortex of macaque. Exp Brain Res70:216–220

Goldberg ME, Colby CL, Duhamel J-R (1990) Representation ofvisuomotor space in the parietal lobe of the monkey. ColdSpring Harb Symp Quant Biol 55:729–739

Gottlieb JP, Kusunoki M, Goldberg ME (1998) The representation ofvisual salience in monkey parietal cortex. Nature 391:481–484

Grunewald A, Linden JF, Andersen RA (1999) Responses to audi-tory stimuli in macaque lateral intraparietal area. I. Effects oftraining. J Neurophysiol 82:330–342

Hanes DP, Patterson WF II, Schall JD (1998) Role of frontal eyefield in countermanding saccades: visual, movement and fixa-tion activity. J Neurophysiol 79:817–834

Hays AV, Richmond BJ, Optican LM (1982) A Unix-based multi-ple-process system for real-time data acquisition and control.WESCON Conference Proceedings 2:1–10

Hikosaka O, Wurtz RH (1989) The basal ganglia. In: Wurtz RH,Goldberg ME (eds) The neurobiology of saccadic eye move-ments. Elsevier, New York, pp 257–282

Jones JP, Palmer LA (1997a) The two-dimensional spatial struc-ture of simple receptive fields in cat striate cortex. J Neuro-physiol 58:1187–1211

Jones JP, Palmer LA (1997b) An evaluation of the two-dimension-al gabor filter model of simple receptive fields in cat striatecortex. J Neurophysiol 58:1233–1259

Judge SJ, Richmond BJ, Chu FC (1980) Implantation of magneticsearch coils for measurements of eye position: an improvedmethod. Vision Res 20:535–538

143

Keller EL (1991) The brain stem. In: Carpenter RHS (ed) Visionand visual dysfunction: eye movements. CRC, Boca Raton, pp200–223

Li CS, Andersen RA (1994) Up-down asymmetry of saccade-re-lated and fixation activity in area LIP. Soc Neurosci Abstr20:773

Linden JF, Grunewald A, Andersen RA (1999) Responses to audi-tory stimuli in macaque lateral intraparietal area. II. Behavior-al modulation. J Neurophysiol 82:343–358

Lynch JC, Graybiel AM (1985) The differential projection of twocytoarchitectonic subregions of the inferior parietal lobule ofmacaque upon the deep layers of the superior colliculus. JComp Neurol 235:241–254

Lynch JC, Mountcastle VB, Talbot WH, Yin TC (1977) Parietallobe mechanisms for directed visual attention. J Neurophysiol40:362–389

Moran J, Desimone R (1985) Selective attention gates visual pro-cessing in the extrastriate cortex. Science 229:782–784

Motter BC (1993) Focal attention produces spatially selective pro-cessing in visual cortical areas V1, V2 and V4 in the presenceof competing stimuli. J Neurophysiol 70:909–919

Motter BC (1994) Neural correlates of attentive selection for coloror luminance in extrastriate area V4. J Neurosci 14:2178–2189

Munoz DP, Wurtz RH (1993) Fixation cells in monkey superiorcolliculus. I. Characteristics of cell discharge. J Neurophysiol70:559–575

Platt ML, Glimcher PW (1997) Responses of intraparietal neuronsto saccadic targets and visual distractors. J Neurophysiol1574–1589

Platt ML, Glimcher PW (1998) Response fields of intraparietalneurons quantified with multiple saccadic targets. Exp BrainRes 121:65–75

Raiguel S, Van Hulle MM, Xiao D-K, Marcar VL, Orban GA(1995) Shape and spatial distribution of receptive fields andantagonistic motion surrounds in the middle temporal area(V5) of the macaque. Eur J Neurosci 7:2065–2082

Raiguel S, Van Hulle MM, Xiao D-K, Marcar VL, Lagae L, OrbanGA (1997) Size and shape of receptive fields in the medial su-perior temporal area (MST) of the macaque. Neuroreport8:2803–2808

Schall JD, Morel A, King DJ, Bullier J (1995) Topography of vi-sual cortex connections with frontal eye field in macaque: con-vergence and segregation of processing streams. J Neurosci15:4464–4487

Schmued LC (1990) A rapid sensitive histochemical stain for my-elin in frozen brain sections. J Histochem Cytochem 38:717–720

Seagraves MA, Goldberg ME (1987) Functional properties of cor-ticotectal neurons in the monkey frontal eye fields. J Neuro-physiol 68:1967–1985

Seltzer B, Pandya DN (1986) Posterior parietal projections to theintraparietal sulcus of the rhesus monkey. Exp Brain Res62:459–469

Steinmetz MA, Connor CE, Constantinidis C, McLaughlin JR(1994) Covert attention suppresses neuronal responses in area 7aof the posterior parietal cortex. J Neurophysiol 72:1020–1023

Stricanne B, Andersen RA, Mazzoni P (1996) Eye-centered, head-centered, and intermediate coding of remembered sound loca-tions in area LIP. J Neurophysiol 76:2071–2076

Thier P, Anderson RA (1998) Electrical microstimulation distin-guishes distinct saccade-related areas in the posterior parietalcortex. J Neurophysiol 80:1713–1735

Tootell RBH, Switkes E, Silverman MS, Hamilton SL (1988)Functional anatomy of macaque striate cortex. II. Retinotopicorganization. J Neurosci 8:1531–1568

Ungerleider LG, Desimone R (1986) Cortical connections of visu-al area MT in the macaque. J Comp Neurol 248:190–222

Wörgötter F, Holt G (1991) Spatiotemporal mechanisms in recep-tive fields of visual cortical simple cells: a model. J Neuro-physiol 65:494–510

Wurtz RH (1969) Visual receptive fields of striate cortex neuronsin awake monkeys. J Neurophysiol 32:727–742

144