How the Brain Moves Your Eyes About

CHAPTER 9

Look and see: how the brain moves your eyes about

Peter H. Schiller Ł and Edward J. Tehovnik

Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA

Abstract: Two major cortical streams are involved in the generation of visually guided saccadic eye movements: the anterior and the posterior. The anterior stream from the frontal and medial eye fields has direct access to brainstem oculomotor centers. The posterior stream from the occipital cortices reaches brainstem oculomotor centers through the superior colliculus. The parietal cortex interconnects with both streams. Our findings suggest that the posterior stream plays an unique role in the execution of rapid, short latency eye movements called ‘express saccades’. Both the anterior and posterior streams play a role in the selection of targets to which saccades are to be generated, but do so in different ways. Areas V1, V2 and LIP contribute to decisions involved in where to look as well as where not to look. In addition, area LIP is involved in decisions about how long to maintain fixation prior to the execution of a saccade. Area V4 does not appear to be directly involved in eye-movement generation. In the anterior stream, the frontal eye fields, and to a lesser extent the medial eye fields, are involved in the correct execution of saccades subsequent to decisions made about where to look and where not to look.

Introduction studied quite a number of brain areas that include V1, V2, V4, the lateral intraparietal sulcus (LIP), the Our eyes are on the move most of the time during frontal eye ﬁelds (FEF), the medial eye ﬁelds (MEF), our waking hours. We make about three saccadic and the superior colliculus (SC). eye movements per second, some 170,000 a day, These were carried out on trained rhesus monkeys and about 5 billion in an average life time. With because their eye movements are similar to those each shift in gaze, the numerous stimuli in the vi- of humans. Fig. 1 shows records of eye movements sual scene impinge on new retinal locations, to only made by one of our animals while he was restrained one of which can the eye be shifted subsequently. and looked at the displays shown on the left. The This process of target selection necessitates the dis- monkey received periodic juice reward. The eye crimination of objects, the assessment of temporal movements are similar to those generated by humans order and the determination of spatial location. In as shown by Yarbus (1967). this presentation, some of the neural systems that are involved in the generation of visually guided sac- Neural structures involved in the generation of cadic eye movements are described. We have used saccadic eye movements several methods in our effort that include single-cell recordings, lesions and microstimulation. We have The exploration of the neural mechanisms of visually guided saccade generation is begun by demonstrat- ing, schematically, the central structures from which Ł Corresponding author: P.H. Schiller, Department of saccadic eye movements can be elicited by electrical Brain and Cognitive Sciences, Massachusetts Institute of microstimulation. As shown in Fig. 2, such saccades Technology, Cambridge, MA 02139, USA. can be elicited from quite a number of brain areas Tel.: C1-617-253-5754; Fax: C1-617-253-8943; at relatively low currents. As indicated by the arrows E-mail: [email protected] in the circles, stimulation of regions of the occipital

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 2

Fig. 1. Saccadic eye movements of a monkey, shown in the right panels, while the animal was viewing the stimulus displays shown on the left. cortex, such as V1 and V2, elicits saccades whose A different coding operation takes place in the me- vectors are constant, and are independent of starting dial eye fields where stimulation elicits a saccade eye position (Schiller, 1972, 1977, 1998). The vector that takes at the center of gaze to a particular lo- and amplitude of the saccade elicited depends on cation in craniotopic space (Schlag and Schlag-Rey, where one stimulates within the structure. When one 1987; Tehovnik and Lee, 1993). Different subregions records as well and finds the location of the recep- code for different spatial locations. In the parietal tive fields, stimulation takes the eye to the receptive lobe, in different subregions one can find either one field location prior to the eye movement. These areas or the other of these coding operations (Shibutani therefore carry a vector code. What is computed is et al., 1984; Kurylo and Skavenski, 1991; Thier and a retinal error signal between the center of gaze and Andersen, 1998). the location of the receptive field activated by the Although it was thought that cortical signals for target; the eye movement nulls this error. Like visual the generation of saccadic eye movements are con- cortex, the superior colliculus and the frontal eye veyed entirely through the superior colliculus, more fields also carry a vector code (Robinson and Fuchs, recent evidence indicates that this is not the case. 1969; Robinson, 1972; Schiller and Stryker, 1972). When the colliculus is removed, monkeys still make

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 3

Fig. 2. Eye movements elicited by electrical stimulation of five areas. Stimulation of the superior colliculus, the visual cortex, regions of the parietal cortex, and the frontal eye fields produces saccades whose amplitude and direction depends on the subregion stimulated in each of these areas. Stimulation of the medial eye fields and some regions of the parietal cortex elicits saccades that take the eye to a particular orbital location. pretty good visually guided eye movements (Schiller cortex and the colliculus (Lynch et al., 1985; Huerta et al., 1980; Keating and Gooley, 1988a). Further et al., 1987; Huerta and Kaas, 1990), there is one work established that collicular removal has a se- important additional piece of information in support lective effect on electrically triggered saccadic eye of the two-stream hypothesis depicted in Fig. 4. This movements from the cortex. What happens is de- additional evidence is that ablation of both the su- picted in Fig. 3. After ablation of the superior perior colliculus and the frontal eye fields eliminates colliculus, one can still elicit saccadic eye move- all visually guided saccadic eye movements (Schiller ments from the frontal and medial eye fields, but not et al., 1980). This is a deficit from which there is from posterior cortex (Schiller, 1977; Keating et al., virtually no recovery. 1983; Keating and Gooley, 1988b; Tehovnik et al., Given that there are quite a number of cortical 1994). areas involved in visually guided saccadic eye-move- These findings have led to the formulation de- ment control that appear form two major streams to picted in Fig. 4. According to this generalized the brainstem oculomotor centers, the question arises scheme, there are two major streams for the gen- as to what the functional contributions are of the eration of visually guided saccadic eye movements, cortical areas within each of these two streams. To the anterior and the posterior. The posterior stream address this question, in our recent work, we have gains access to the brainstem oculomotor centers concentrated on two methods. The first method ex- through the superior colliculus whereas the anterior amined the effects of selective lesions on the genera- stream from the frontal and medial eye fields has di- tion of visually guided saccadic eye movements. The rect connections to the brainstem. Although cortical second method examined the effects subthreshold areas are extensively interconnected, and the region electrical microstimulation in various cortical areas of the parietal cortex that resides in the lateral intra- on target selection and the execution of saccadic eye parietal sulcus has connections both with the frontal movements.

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 4

Fig. 3. After ablation of the superior colliculus, electrical stimulation no longer elicits saccadic eye movements from the occipital and parietal cortices, but continues to do so in the frontal lobe.

Fig. 4. Model of the hypothesized connections involved in visually guided eye-movement control. Two major streams control these eye movements: the anterior and the posterior. The anterior system has direct access to the brainstem oculomotor centers. The posterior stream reaches these centers through the superior colliculus.

Lesion studies streams contribute to different aspects of eye-movement generation. To block the posterior stream we The aim of our lesion studies was to determine removed the superior colliculus. We then compared how the structures within the anterior and posterior that with lesions of the frontal and medial eye ﬁelds.

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 5

The first task to be discussed is depicted in Fig. 5. Next we turn to a different task depicted in Fig. 7, When a central fixation spot comes on, the monkey which we refer to as the sequential task. In this looks at it. Then a single target appears somewhere case, we present two targets in succession and the on the monitor and the monkey has to look at it monkey’s task is to make a corresponding pair of to receive a drop of apple juice for reward. When saccades. The task is shown in Fig. 7. The dura- a monkey performs this task repeatedly, the latency tion of the targets and the interval between them distribution becomes bimodal as first shown by Fis- is systematically varied. We studied performance on cher and Boch (1983). An example of such a distri- this task before and after unilateral frontal and me- bution appears in the top panel of Fig. 6. The first dial eye field lesions. Fig. 8 shows eye-movement mode in this distribution that peaks at less than 100 records obtained with this task for two sequence du- ms is referred to as ‘express saccades’. Those falling rations 18 and 60 weeks after a left frontal eye field into the second mode are called ‘regular saccades’. lesion (Schiller and Chou, 2000a). The monkey does Throughout these studies, the head of the animals well toward the intact side, but does poorly, espe- was restrained during the experimental sessions. cially for the shorter sequence duration, for saccades So we can now ask what happens to the bimodal to the right. More detailed data showing percent cor- distribution of saccadic latencies when one makes rect performance after frontal eye field and medial lesions of either the superior colliculus or the frontal eye field lesions appear in Fig. 9. In this case, we ran eye fields. After a frontal eye field lesion, there is the same four sets of sequences for several weeks. really very little effect (Schiller et al., 1987). The Recovery is quite rapid with the medial eye field monkey still exhibits a bimodal distribution of sac- lesions, but much slower after frontal eye field le- cadic latencies, although sometimes with a small sions. But even after the apparent recovery, when a overall shift towards slightly longer latencies. By new set of sequences is introduced, a large deficit is contrast, a unilateral lesion of the superior colliculus still evident. Fig. 8 in fact shows the eye movement abolishes all express saccades as depicted in the bot- records obtained with a sequence to which the mon- tom panels in Fig. 6; after the left superior colliculus key was only exposed a few times after 18 and 60 had been removed, the monkey never again makes weeks. express saccades to the right. On the other hand, Next, a third task that is reported is shown in saccades to the left retain their bimodal distribution. Fig. 10. We call this the paired target task. After In addition, the regular saccades to the right have the monkey centers his gaze on the fixation spot, significantly longer latencies. two targets appear. He gets rewarded for looking at These findings suggest that the posterior stream is either target. In addition to presenting the two tar- essential for the generation of express saccades. gets simultaneously, the temporal asynchrony with

Fig. 5. Procedure to study eye movements made to single targets. First a ﬁxation spot appears (1); when the monkey ﬁxates it a target appears in one of several locations (2). A saccade made to the target is rewarded with a drop of apple juice.

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 6

Fig. 6. The distribution of saccadic latencies before and after left superior colliculus lesion. Saccadic latencies made to single targets form a bimodal distribution the first mode of which has been called ‘express saccades’. After a unilateral collicular lesion, monkeys no longer can make express saccades into the visual hemifield that had been represented by the removed colliculus. Regular saccades (the second peak in the distribution) have longer latencies after the collicular lesion. which they appear is also varied. This way we can This task, using only simultaneous presentation determine what the probability is of selecting one conditions, has been extensively used in clinical stud- target or the other as a function of the temporal ies showing that after unilateral cortical lesions the asynchrony between them. target presented in the hemifield contralateral to the

Fig. 7. Method for studying the execution of sequential saccades. After the animal ﬁxates the ﬁxation spot (1) two targets appear in succession (2 and 3). Target duration the interval between them is systematically varied.

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 7

Fig. 8. Eye movements shown 18 and 60 weeks after a left frontal eye field lesion on a set of two target sequences presented for two sequence durations, 217 and 117 ms. The four sequences used were A2–B1, E2–D1, A4–B5 and E5–D5. Performance is much poorer to the right, especially for the shorter sequence duration. The monkey was exposed to the particular set of sequences shown here only occasionally. damaged area is ignored (Bisiach and Vallar, 1988). equal probability choice shifts dramatically after a This has been termed the extinction phenomenon be- unilateral frontal eye field lesion. The target in the cause the target presented in the hemifield ipsilateral affected hemifield has to be presented much earlier to the damaged area ‘extinguishes’ the perception of than the target in the intact field to obtain equal the target in the intact hemifield; presented singly, a probability choice for the left and right targets. This target is perceived at either location in many patients. suggests that processing has been greatly slowed This task with using just two targets is the simplest down as a result of the lesion. Recovery is slow and one we could devise to study target selection. never complete. The effects are much smaller after Fig. 11 shows data obtained on this paired tar- medial eye field lesions and recovery is more rapid. get task after frontal and medial eye field lesions The inset on the right shows actual eye-movement (Schiller and Chou, 1998, 2000b). Plotted here is the records obtained pre-operatively with the targets ap- percent of time the left target is chosen as a function pearing either simultaneously or with 33 ms offsets. of the temporal asynchrony between the targets. Pre- The paired targets on any given trial appeared either operative performance is shown by the solid lines above or below fixation. In the simultaneous pre- and black squares; the probability with which the sentation case, the monkey makes about the same left and right targets are chosen is about the same number of saccades to the left and the right, whereas when the two targets appear simultaneously. This when the targets on the left appear, first most of the

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 8

Fig. 9. Performance on another set of four sequences with varied sequence duration before and after right medial and left frontal eye field lesions. The same set of sequences were used repeatedly for 11 weeks starting 1 and 2 weeks after the RMEF and the LFEF had been ablated. The sequences used were C1–A1, C1–E1, C5–A5 and C5–E5. The RMEF lesion produced only mild deficits that recovered rapidly. The LFEF lesion produced a major deficit that recovered slowly over time. However, as shown in Fig. 8, significant recovery occurred only when the animal was repeatedly exposed to the same set of sequences. saccades are made to the left; the converse is the case in generating express saccades. When it comes to when the targets appear first to the right of fixation. more complex tasks, it appears that the frontal eye Collectively, these lesion studies suggest then that fields in particular play an important role in temporal the posterior stream plays an important role execut- sequencing. The lesion data suggest that this struc- ing saccadic eye movements rapidly, and especially ture also plays a significant role in target selection.

Fig. 10. The paired target task. After ﬁxation, two targets appear. The monkey is rewarded for making a saccadic eye movement to either target. The targets are presented with various temporal asynchronies.

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 9

Fig. 11. Data obtained on the paired target task before and after LFEF and RMEF lesions. Percent of saccades made to the left targets is plotted as a function of the temporal asynchrony between the targets. In the intact monkey (pre-op), the left and right choices are made with equal probability when the two targets are simultaneous. After the LFEF lesion, the equal probability point shifts dramatically; the target in the affected hemifield has to be presented more than 100 ms before the target in the ipsilateral hemifield. There is gradual recovery over time. When tested over a year later, there is still a sizeable deficit after the LFEF lesion. The RMEF lesion produced much milder deficits that recovered fully by the 16th week. Pre-operative eye-movement traces are shown for targets presented simultaneously and with left and right target appearing first by 33 ms on the right side of figure.

Microstimulation studies the threshold level of the stimulation, expressed as the percent probability with which a saccade was To learn more about the role various brain structures elicited with electrical stimulation under our control play in target selection we have recently turned to condition in which the electrical stimulation was microstimulation (Schiller and Tehovnik, 2000). The applied 120–140 ms after the termination and no rationale of these experiments is that subthreshold target was presented. These control conditions were electrical stimulation should be able to influence tar- randomly interspersed with the presentation of the get selection in those cortical areas that are involved paired targets. This procedure for establishing the in this process. The nature of this influence should lowest possible threshold was based on our earlier reveal how various areas contribute to the process. work in which we had found that the efficacy with The procedures we used are depicted in Fig. 12. which a saccade could be elicited with electrical We first record the activity of single neurons in stimulation was lowest when stimulation was applied the area under study. We then map the receptive or well after the fixation spot was extinguished and no motor fields of the neurons from which we record, targets were presented. In the frontal eye fields, and we also examine their response characteristics. thresholds were three times and in the medial eye We then proceed to use the paired target task that are fields 16 times lower when there were no targets and presented with various temporal asynchronies. One stimulation was delivered after the fixation spot had of the targets is placed into the receptive or the motor been terminated (Tehovnik et al., 1999). field of the cells. We then proceed to electrically Saccadic latencies elicited by above threshold stimulate at low current levels to determine how electrical stimulation depended on the current levels target choice is altered. In addition, at each site we used. This was true for all the brain sites we had also determined to what extent saccadic latencies to studied. When the probability of eliciting saccades single targets are affected electrical stimulation. in the no target control condition was between 25 We most commonly stimulated for 80 ms at 200 and 75%, latencies ranged between 75 and 150 ms. Hz. Stimulation was typically initiated 30 ms after The majority of the effects reported on here were the appearance of the first target. This delay time obtained with current levels below which saccades was chosen because we had found that in areas were elicited under our control conditions. V1 and V2 single cells began discharging to the Also interspersed with the presentation of the visual stimuli we used with that latency. The current paired targets and the control condition were sin- levels we used are indicated for each curve generated gle targets that appeared at the same location as ?#1 in Figs. 13–16. Also indicated in these figures is the paired targets. Stimulation was administered on a

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 10

Fig. 12. Method used to study the effects of subthreshold electrical stimulation on target choice. After an electrode is lowered into a selected structure, the receptive field location of the neurons is plotted first. Then the area is stimulated at above threshold levels. The saccade elicited shifts the center of gaze to the receptive field. Thirdly, a target is placed into the receptive field to elicit the same direction and amplitude saccade that was elicited by electrical stimulation. Fourthly, two targets are presented with various asynchronies with one of them in the receptive field. We then proceed to determine to what extent subthreshold electrical stimulation alters they frequency with which the targets are selected. minority of trials with single targets as well which al- in Fig. 14a, stimulation of the upper layers of V1 typ- lowed us to determine the extent to which the latency ically produced interference by decreasing the prob- a visually triggered saccade could be influenced by ability with which the target in the receptive field of subthreshold stimulation. the stimulated neurons was selected; the magnitude Fig. 13 depicts the cortical areas we had studied. of the effect increased with increases in subthreshold They are areas V1, V2, V4, LIP, FEF and MEF. current. By contrast, stimulation in lower layers of In area V1, suprathreshold stimulation in the ab- V1 produced facilitation by increasing the probabil- sence of any targets elicited saccades that shifted ity with which the target in the receptive field was the center of gaze into the receptive field location chosen. This is shown for one site in Fig. 14b. of the stimulated neurons. Thresholds for eliciting At sites where stimulation produced an interfer- such saccades were considerably higher in the upper ence effect, stimulation administered in conjunction layers than in the lower layers where saccades were with single targets produced significant increases in often elicited with currents of less than 25 µA. Sub- saccadic latencies; on the other hand, stimulation at threshold stimulation administered with paired tar- sites where facilitation was obtained had little ef- gets produced two effects. Unexpectedly, as shown fect on saccadic latencies to single targets. For the

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 11

Fig. 13. A schematic showing the brain areas we had stimulated. They are V1, V2, V4, LIP, MEF and FEF. data shown in Fig. 14a and b, concurrently collected Differences in latencies for stimulated and unstimu- single target latency data are shown in Fig. 16, V1. lated trials are shown in Fig. 16, V2, for the same In area V2, similar results were obtained. Stim- data set that is graphed in Fig. 14c. For stimula- ulation in the lower layers typically produced fa- tion in the upper layers, which induced interference, cilitation, whereas stimulation in the upper layers saccadic latencies to single targets increased. By produced interference. In Fig. 14c, data are shown contrast, stimulation in the lower layers at the same for conditions in which stimulation was delivered current levels was above threshold and shortened to two locations in the same electrode penetration saccadic latencies. using 50 µA. In the lower layers, this current was In area V4, electrical stimulation never produced suprathreshold and resulted in driving eyes to the saccades using currents levels up to 200 µAand target in the receptive field. By contrast, the same failed to have any effect on target selection or current level in the upper layers decreased the proba- saccadic latencies. Data are plotted in Figs. 14d bility of choosing the target in the receptive field. and 16. Saccadic latencies to single targets appearing in In area LIP, saccades could be elicited from a the receptive field complemented these effects in V2. minority of sites (see Table 1). At such sites, sub- ?#2

TABLE 1 Number of sites studied in each cortical area and their effects

V1 V2 V4 LIP DMFC FEF Sites electrically stimulated 47 11 21 156 63 92 Sites from which saccades were elicited 46 11 0 8 11 50 Sites tested with single and paired targets 44 11 12 33 9 24 Sites yielding increased choice of target in RF 12 6 0 10 9 24 Sites yielding decreased choice of target in RF 31 4 0 9 0 0 Sites yielding increased ﬁxation times 0 0 0 14 0 0

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 12

Fig. 14. The probability with which the target presented in the receptive field is chosen as a function of the temporal asynchrony between the two targets and the currents of electrical stimulation used. (a) Interference in V1. (b) Facilitation in V1. (c) Facilitation in the lower and interference in the upper layers of V2. (d) No effect in V4. The insets show arrangement of targets in the visual field. Currents and probability of eliciting a saccade when no target appeared (control condition) are shown for each line. The total number of trials from which the data are derived in this figure was 3480. threshold stimulation in the paired target case either higher subthreshold currents increased single-target produced facilitation or interference. In Fig. 15a, latencies. data are shown from a site where stimulation pro- An additional effect in LIP was found at sites duced interference. Fig. 15b shows data for a another where neurons discharged with fixation (Lynch et site that yielded facilitation. al., 1977; Mountcastle et al., 1981); here currents of Saccadic latencies obtained in LIP to single tar- up to 200 µA failed to elicit saccades. The effect gets appear in Fig. 16. Latency differences shown in of stimulation at these sites was to prolong fixation. Fig. 16, LIP, sections a and b, were obtained concur- This effect differed from the already noted increases rently with the data collected as shown in Fig. 15a in saccadic latencies in V1, V2 and LIP by pro- and b. Single target latencies obtained at facilitatory longing fixation duration no matter where targets sites were not significantly altered by stimulation. appeared in the visual field. Data obtained from one On the hand, at sites that produced interference, such site are plotted in Fig. 16, LIP, section c. Sim-

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 13

Fig. 15. The probability with which the target presented in the receptive field is chosen as a function of the temporal asynchrony between the two targets and the currents of electrical stimulation used. (a) Interference in LIP. (b) Facilitation in LIP. (c) Facilitation in MEF (d) Facilitation in FEF. The insets show arrangement of targets in the visual field. Currents and probability of eliciting a saccade when no target appeared (control condition) are shown for each line. The total number of trials from which the data are derived in this figure was 4620. ilar increases in fixation time were obtained at 13 were lowest in this area. Fig. 15d shows data from other sites in LIP (Table 1). one site. Stimulation in the FEF also shortened sac- In the MEF, stimulation at 200 µAorlower cadic latencies to single targets presented in the mo- elicited saccadic eye movements at the minority tor field. The mean decreases in saccadic latencies of the sites tested. At the effective sites, stimulation obtained with single targets in the motor field, col- always produced facilitation. Data from one such site lected concurrently with the data shown in Fig. 15d, appear in Fig. 15c. Concurrently collected saccadic are shown in Fig. 16, FEF. latencies to single targets were shortened moderately Table 1 provides a tally of the number of sites at with stimulation as shown in Fig. 16, MEF. which the three effects on which we report had been In the FEF stimulation elicited saccades at the found. majority of sites. Subthreshold stimulation consis- These microstimulation results suggest that both tently produced facilitation. The effective currents the anterior and the posterior system are involved

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 14

Fig. 16. Differences between the mean saccadic latencies made to single targets with and without electrical stimulation. The data shown were collected concurrently with those that appear in Figs. 2 and 3 with the exception of those for LIP in Fig. 4, subsection c, where the data shown were obtained from a site which contained cells that responded with fixation. No data were collected here with paired targets since the increases in saccadic reaction times were not target location specific. The stars next to the right tick marks in the figure indicate saccadic latency differences that were statistically significant (P < 0:001). The mean saccadic latencies without stimulation varied between 105 and 146 ms with standard deviations of 7–12 ms. in target selection. In the posterior system, it ap- The first under #1, is to determine what the objects pears that decisions are made not only as to where are in the scene. This we know is carried out by to look, but also where not to look. This latter de- many structures that include V1, V2, V4, inferotem- cision process, where not to look, seems to be a poral cortex and LIP. Second, one target needs to be prominent feature of V1, V2 and LIP since, in these chosen to be looked at. This decision, based on the areas, stimulation at some of the sites decreased facilitatory effects we had found, appears to involve the probability of choosing the target that had been areas V1, V2, LIP, the FEF and the MEF. Our new placed within the receptive field of the stimulated finding showing that stimulation at some sites actu- neurons. ally interfered with target selection suggests that a Based on our lesion and microstimulation exper- decision also needs to be made as to which targets iments we can now proceed to specify what goes not to look at. This ‘de-selection’ process involves on in the process of target selection with saccadic the upper layers of V1 and V2 and some regions of eye movements in the various cortical areas we had LIP since at these locations stimulation typically de- studied. Following each saccade to a new location, creased the probability with which the targets in the we see five computations being carried out for gen- receptive fields were chosen. This interference effect erating the next saccade as schematized in Fig. 17. may well be a product of inhibitory feedback circuits

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 15

Fig. 17. Schematic representation of the computations involved in generating a saccadic eye movement to a selected target. After each saccade to a new location, the object to which the eye is directed is analyzed. At the same time, steps need to be taken for generating the next saccade. The ﬁve computations carried out are depicted as well as the structures involved. activated by our electrical stimulation. Fourth, the Acknowledgements spatial location of the targets needs to be computed to generate the correct saccade; this in cortex proba- Collaborators in the work described were I-han bly involves V1, V2, and the FEF in each of which Chou, Andreas Tolias, John Maunsell, Julie Sandell, we have nice topographic representation. Lastly, #5, Michael Stryker, Janet Conway, and Sean True. We a decision needs to be made as to when to initiate the are also indebted to Warren Slocum and Christina saccade which we believe lies largely in the domain Carvey for their contributions to this work. of area LIP. These considerations suggest that both the anterior and the posterior streams area centrally References involved in selecting targets with saccadic eye movements. One cannot help but be awed by the fact Bisiach, E. and Vallar, G. (1988) Hemineglect in humans. In: E. that this seemingly simple task of moving one’s eyes Boller and J. Grafman (Eds.), Handbook of Neuropsychology, about involves so many brain structures and so much Vol. 1, Elsevier, Amsterdam, pp. 195–222. computation. Fischer, B. and Boch, R. (1983) Saccadic eye movements after

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16 16

extremely short fixation times in the monkey. Brain Res., 260: Attention, Lawrence Erlbaum, London, Chapt. 1. 21–26. Schiller, P.H. and Chou, I. (1998) The effects of frontal eye field Huerta, M.F. and Kaas, J.H. (1990) Supplementary eye field and dorsomedial frontal cortex lesions on visually guided eye as defined by intracortical microstimulation: connections in movements. Nat. Neurosci., 1: 248–253. macaques. J. Comp. Neurol., 293: 299–437. Schiller, P.H. and Chou, I. (2000a) The effects of anterior arcuate Huerta, M.F., Krubitzer, L.A. and Kaas, J.H. (1987) Frontal eye and dorsomedial frontal cortex lesions on visually guided eye field as defined by intracortical microstimulation in squirrel movements in the rhesus monkey: I. Single and sequential monkeys, owl monkeys, and macaque monkeys. J. Comp. targets. Vision Res., 40: 1609–1626. Neurol., 265: 332–361. Schiller, P.H. and Chou, I. (2000b) The effects of anterior arcuate Keating, E.G. and Gooley, S.G. (1988a) Saccadic disorders and dorsomedial frontal cortex lesions on visually guided eye caused by cooling the superior colliculus or the frontal eye movements in the rhesus monkey: II. Paired and multiple field, or from combined lesions of both structures. Brain Res., targets. Vision Res., 40: 1626–1638. 438: 247–255. Schiller, P.H. and Stryker, M. (1972) Single-unit recording and Keating, E.G. and Gooley, S.G. (1988b) Disconnection of pari- stimulation in superior colliculus of the alert rhesus monkey. etal and occipital access to the saccadic oculomotor system. J. Neurophysiol., 3: 915–924. Exp. Brain Res., 70: 385–398. Schiller, P.H. and Tehovnik, E.J. (2000) The role of cortical Keating, E.G., Gooley, S.G., Pratt, S.E. and Kelsey, J. (1983) areas in target selection with saccadic eye movements. Soc. Removing the superior colliculus silences eye movements nor- Neurosci. Abstr., 26: 1075. mally evoked from stimulation of the parietal and occipital eye Schiller, P.H., True, S.D. and Conway, J.L. (1980) Deficits in eye fields. Brain Res., 269: 145–148. movements following frontal eye field and superior colliculus Kurylo, D.D. and Skavenski, A.A. (1991) Eye movements ablations. J. Neurophysiol., 44: 1175–1189. elicited by electrical stimulation of area PG in the monkey. Schiller, P.H., Sandell, J.H. and Maunsell, J.H.R. (1987) The J. Neurophysiol., 65: 1243–1253. effect of frontal eye field and superior colliculus lesions on Lynch, J.C., Mountcastle, V.B., Talbot, W.H. and Yin, T.C.T. saccadic latencies in the rhesus monkeys. J. Neurophysiol., 57: (1977) Parietal lobe mechanisms for directed visual attention. 1033–1049. J. Neurophysiol., 40: 362–389. Schlag, J. and Schlag-Rey, M. (1987) Evidence for a supplemen- Lynch, J.C., Graybiel, A.M. and Lobeck, L.J. (1985) The dif- tary eye field. J. Neurophysiol., 57: 179–200. ferential projection of two subregions of the inferior parietal Shibutani, H., Sakata, H. and Hyvarinen, J. (1984) Saccade and lobe of macaque upon deep layers of the superior colliculus. J. blinking evoked by microstimulation of the posterior associa- Comp. Neurol., 235: 241–245. tion cortex of the monkey. Exp. Brain Res., 55: 1–8. Mountcastle, V.B., Andersen, R.A. and Motter, B.C. (1981) The Tehovnik, E.J. and Lee, K.-M. (1993) The dorsomedial frontal influence of attentive fixation upon the excitability of the light- cortex of the rhesus monkey. Topographic representation of sensitive neurons of the posterior parietal cortex. J. Neurosci., saccades evoked by electrical stimulation. Exp. Brain Res., 96: 11: 1218–1235. 430–442. Robinson, D.A. (1972) Eye movements evoked by collicular Tehovnik, E.J., Lee, K.-M. and Schiller, P.H. (1994) Stimulation- stimulation in the alert monkey. Vision Res., 12: 1795–1808. evoked saccades from the dorsomedial frontal cortex of the Robinson, D.A. and Fuchs, A.F. (1969) Eye movements evoked rhesus monkey following lesions of the frontal eye fields and by stimulation of frontal eye fields. J. Neurophysiol., 32: 637– superior colliculus. Exp. Brain Res., 98: 179–190. 648. Tehovnik, E.J., Slocum, W.M. and Schiller, P.H. (1999) Be- Schiller, P.H. (1972) The role of the monkey superior colliculus havioural conditions affecting saccadic eye movements elicited in eye movements and vision. Invest. Ophthalmol., 11: 451– electrically from the frontal lobes of primates. Eur. J. Neu- 459. rosci., 11: 2431–2443. Schiller, P.H. (1977) The effect of superior colliculus ablation Thier, P. and Andersen, R.A. (1998) Electrical microstimulation on saccades elicited by cortical stimulation. Brain Res., 122: distinguishes distinct saccade-related areas in the posterior 154–156. parietal cortex. J. Neurophysiol., 80: 1713–1735. Schiller, P.H. (1998) The neural control of visually guided eye Yarbus, A.L. (1967) Eye Movements and Vision. Plenum Press, movements. In: J. Richards (Ed.), Cognitive Neuroscience of New York.

QUERIES: ?#1: Fig. 16, legend: Fig. 4 has been cited. Is this correct? (page 9) ?#2: Please note that Table 1 has been scanned. (page 11)

CICERO/GALAYAA B.V. / CASANOVA 9: pp. 1-16