Response Characteristics of Single Cells in the Monkey Superior Colliculus Following Ablation or Cooling of Visual Cortex

Transcription

1 Response Characteristics of Single Cells in the Monkey Superior Colliculus Following Ablation or Cooling of Visual Cortex PETER H. SCHILLER, MICHAEL STRYKER, MAX CYNADER, AND NANCY BERMAN iwassachusetts Institute of Technology, Cambridge, ltlassachusetts ELECTROPHYSIOLOGICAL STUDIES have shown that the characteristics of single cells in the superior colliculus depend to a considerable extent on cortical input. In the cat, the majority of studies reveal a dramatic loss of binocularity and direction selectivity in superficial collicular units following ablation of the visual cortex (1, 18, 25). In the ground squirrel, removal of the visual cortex renders most cells in the intermediate and deep layers of the superior colliculus unresponsive to visual stimuli (14); this also appears to apply to the cat (22). Work on the rabbit, however, indicates that in this species, ablation of visual cortex produces no discernible effects on collicular function (10). The present study has been undertaken to investigate the contribution of visual cortex to collicular function in the rhesus monkey (Macaca mulatta), a species whose visual svstem is believed to be rather similar to that of man. Neuroanatomical work has indicated that in the monkey, as in the cat, retina and visual cortex project densely on the superior colliculus (8). In the monkey, a foveate animal, an interesting specialization has been reported: anterograde-degeneration and autoradiographic studies indicate that the anterior part of the colliculus, representing the central 5 of the visual field, receives extremely sparse or no terminations from the retina and, conversely, heavy projections from visual cortex (5, 9, 26) Single-unit studies of the superior colliculus of the intact monkey have disclosed that, as in the cat, most cells have binocular Received for publication April 13, receptive fields. In contrast to the cat, however, in which about 70% of the cells studied are selective for direction of stimulus movement (1, 13, 23), in the monkey only a small percentage of cells has been found to have this attribute (2, 4, 20). In both species cells firing in relation to eye movement have been reported in the lower layers of the colliculus (20, 24, 27). In the monkey these cells discharge prior to saccades and show specificity in terms of the size and direction of rapid eye movement. Some of these units have dual properties; not only do they fire prior to certain saccades, but they also have visual receptive fields. Electrical stimulation in this region of the alert animal elicits conjugate saccades at very low current levels (19, 21). On the basis of our earlier work, we have suggested that the superior colliculus contributes to the monkey s ability to foveate visual targets by saccadic eye movements (20) To determine what the contribution of the visual cortex is to the function of the superior colliculus in the monkey, we obtained extracellular single-unit recordings from the superior colliculus after the cortex had either been ablated or reversiblv cooled. METHODS The work reported here is based on extracellular recording from the superior colliculus of monkeys prepared by three different techniques: 1) anesthetized, paralyzed monkeys with unilateral ablation of visual cortex; 2) anesthetized, paralyzed monkeys in whom part of visual cortex was reversibly cooled; and 3) alert monkeys with one eye surgically immobilized, in whom visual cortex was unilaterally removed.

2 182 SCHILLER, STRYKER, CYNADER, AND BERMAN Anesthetized animals with unilatcd removal of visual cortex Eight monkeys were studied. Visual cortex was removed by suction under aseptic conditions using Nembutal anesthesia 223, 122, 111, 56 29, 6, 5, and 5 days prior to recording from thl superior colliculus. The lesion on the dorsal surface extended to the posterior margin of the lunate sulcus. The lesion was extended to include the medial and lateral convolutions of the calcarine fissure. For recording, animals were first anesthetized with Pentothal. After tracheotomy and application of contact lenses (see also below), animals were placed in a stereotaxic frame. Two trephine holes were cut in the skull above the superior colliculus to permit stereotaxic lowering of microelectrodes into each colliculus. During the experiment animals were paralyzed by infusion of Flaxedil (40 mg/hr) diluted with 5% dextrose in lactated Ringer solution and were respirated with a mixture of N,O and 0, (2). End-tidal CO, was monitored and maintained at 4.0%. The methods for unit recording, receptivefield plotting, and stimulation were similar to those previously reported (2). Platinum-iridium electrodes were used; single cells were identified by conventional criteria (14). A slide projector mounted on a tripod was used to find the location of the receptive fields on a tangent screen facing the animal. For the quantitative study of unit responses to light an optic bench was used, which through a lens system, a shutter, and a mirror galvanometer could project stimuli to various parts of the visual field. Waveform generators were used to drive the shutter and mirror galvanometer (20). In two animals our aim was to determine the location of receptive fields in the visual field with special emphasis on the central 5O. Receptive-field locations relative to the center of the fovea are difficult to determine with sufficient accuracy relying solely on repeated plotting of the fovea with a reversible opthalmoscope. Therefore, two additional procedures to monitor residual eye movement were employed: a) recordings from a reference unit and b) a laser optical lever. For the reference unit the procedure was to lower two microelectrodes, one into each colliculus. While exploring one colliculus with repeated penetrations, the microelectrode in the other colliculus was kept in one place, and the receptive-field location of the reference unit or multiple-unit activity was checked repeatedly for each eye. Reference units were generally chosen in the anterior part of of the colliculus where receptive fields for single cells as well as for multiple-unit activity were small (< 1 ). Thus, movement of the rcference unit s receptive field on the tangent screen was a measure of residual eye movement accurate to within t.25. This method was compared with the optical-lever method and our repeated plotting of the fovea with a reversible opthalmoscope. For the optical-lever technique a laser beam (Edmund Scientific) was reflected from two small mirrors attached to the contact lens with epoxy glue. The contact lens was glued to the cornea1 stroma with tissue cement (histoacryln-blau, B. Braun, Melsungen, W. Germany) which was applied in a very thin layer around the edge of the lens (3). In order to expose the stroma, the cornea1 epithelium was removed under anesthesia by gentle scraping with a scalpel blade 2 min after application of tincture of iodine (2%) to the eye. The glue attached the contact lens firmly to the stroma and provided a clear optic medium which did not cloud with time. This method has been used successfully in other applications with the lens remaining attached and clear for several months (3). After the animal was placed in the stereotaxic frame, the laser beam was directed onto that portion of the contact lens which contained the two mirrors, reflecting two small spots of light onto the tangent screen. The plotting of each receptive field was accompanied by a check on the reflected laser-beam positions, thus permitting an assessment of eye movements, including rotation. In some of the animals, small lesions were placed in the superior colliculus to mark microelectrode positions. Subsequently sections were stained with Nissl or reduced silver stains (Nauta and Fink-Heimer methods). In animals with long survival times, retrograde cell changes in the lateral geniculate nucleus were studied to determine how complete the cortical ablation was. Anesthetized animals with reversible cooling of visual cortex In three monkeys a region of cortical area 17 was reversibly cooled during recordings in the superior colliculus. The recording method for the cooling experiments was identical to those described above. Cooling and rewarming was accomplished by use of a Peltier -type thermoelectric cooling probe (Cambion Inc., Cambridge, Mass.); a silver plate for thermal conduction was attached to the probe. A ZOmm-diameter circular extension of this plate, which fit into a trephine hole in the skull, was shaped to conform to the contour of the brain. Dura was not removed. In the center of the

3 CORTICOTECTAL PATHWAY circular plate was a 2-mm-diameter hole through which one could insert probes for recording and temperature measurement. This plate cooled a region of visual cortex representing a 4-8O area of the visual field. Because of the relatively small area cooled, it was necessary to align the recording and cooling sites topographically. To do this, in two of the monkeys we first recorded single cells in visual cortex by inserting a microelectrode through the center hole of the silver plate, and we plotted on the tangent screen the location of their receptive fields. Penetrations were then made in the-superior colliculus until units were obtained with receptive fields no further than 2--3O from the cortical receptive fields. This assured us that the colliculus electrode was within the region receiving projections from the area under the cooling probe. Residual eye movements were checked using the optical-lever technique in one animal and the method of repeated plotting of the fovea with a reversible opthalmoscope in the other two. Cooling was commenced by passing current through the thermoelectric device using a battery. The amount of current passed was monitored by an ammeter and controlled by variable resistors which were operated manually. A thermistor temperature probe in a 24G needle (Yellow Springs Instruments), placed either on the dura or slightly below the surface of the cortex, was used to measure brain temperature. In one animal bipolar EEG was recorded by attaching two wire electrodes to the thermistor. Cooling to between 10 and 22 C was accompanied by a flattening of the EEG signal. After cooling, cortex could be warmed to body temperature by passing current in the reverse direction. The whole procedure was quite rapid; a cycle going from 36 to 15 C and back to 36 C took about 5 min. Unless cooled to very low temperatures, cortex would recover rapidly, as determined by cortical EEG and the responses of units in the superior colliculus. On return to 36 C, or a few minutes thereafter, the cooled area appeared to be fully functional. In one animal we recorded from both colliculi while cooling one visual cortex. Alert mon kevs The methods used for alert recording were similar to those already described (20). One eye of the monkey was immobilized by transection of the third, fourth, and sixth cranial nerves. Eye-movement electrodes were implanted around the moving eye. During experimental sessions the animal s head was restrained. Recordings in the superior colliculus permitted study of receptive-field characteristics through the immo bilized eye and the st udy of eyemovement related activity via the moving eye. During these experiments the major emphasis was placed on the study of eye-movement cells in the deeper layers of the superior colliculus. Two animals were studied. RESULTS Recording in superior colliculus of paralyzed and anesthetized animals with unilateral visual cortex ablation One hundred thirty-six visually responsive units were studied in six animals. Of these, 30 were in the superior colliculus contralateral to visual cortex ablation and 106 in the ipsilateral colliculus. In two additional monkeys, receptive-field plots of single or multiple units at 75 sites were made to study in detail the topographic representation of the retina on the collicu- SUPERFICIAL LAYERS. Receptive-field-properties. The following aspects of receptivefield orga nization were studied size of activa ting region, center-surround organ ization, spatial summation, and the temporal response characteristics of units to onset and offset of a stimulus. Unilateral ablation of visual cortex had no discernible effect on the receptive-field properties of units in the colliculus contralateral to the ablation. Ipsilateral to the ablation the response characteristics of single cells in stratum griseum superficiale and the upper region of stratum opticum were affected onlv to a small extent. Figure 1 shows the responses to a flashing spot of two units in the same monkey, one studied in the colliculus contralateral and the other ipsilateral to cortical ablation. These histograms show that as the size of the stimulus is increased, first there is spatial summation followed by progressive surround inhibition. The responses are typical of normal collicular units, showing a transient burst to both onset and offset of the stimulus. The small differences in the temporal distribution of the responses in the two sets of histograms are within the range of variation among normal units. While the majority of units studied in the superficial colliculus ipsilateral to visual cortex ablation had receptive-field

4 184 SCHILLER, STRYKER, CYNADER, AND BERMAN INTACT SIDE 80- ON OFF 1.6 sot ABLATED SIDE ON OFF FIG. 1. ~4. response histograms obtained from a superior colliculus unit contralateral to the side of visual cortex ablation. Stimulus duration, 1 set repeated every 2 sec. Bin width, 20 msec; 30 repeated stimulus presentations per histogram. Ordinate: total number of discharges per bin. Onset and offset of stimulus occurs at the left edge of the letter 0. B: response histograms obtained from a superior colliculus unit on the side ipsilateral to the visual cortex ablation. Twenty repeated stimulus presentations per histogram. Other parameters as in A. properties indistinguishable from those found in intact animals, some showed subtle differences in their receptive-field organization in that they no longer gave uniform on-off bursts throughout the receptive field. After the ablation such visual receptive fields appeared patchier : some regions produced strong off-responses with small or no on-bursts; the reverse was true for other regions. When a stimulus producing the most vigorous response was placed in the center of the field this on-off imbalance was frequently evident. Such receptive fields did not, however, show onoff center-surround organization of the sort commonlv seen in the retina. We have not seen such patchy receptive fields in the colliculus of normal animals. Response histograms obtained from at least 20 presentations of an optimal size circular spot centered in the receptive field allowed us to classify the units we studied into five categories: those in which the number of discharges elicited by the onset and offset of the stimulus were equal within 50% (category 3); those in which the onand off-responses differed by more than 50(;?, (categories 2 and 4); and those where only an on- or off-response was obtained (categories 1 and 5). Figure 2 compares these distributions for units in superior colliculi with normal and ablated overlying visual cortices. Data from the intact side of unilaterally ablated visual cortex animals were pooled with those from intact animals studied previous to this report under identical conditions (2). The shift toward a greater on-off imbalance, which is evident primarily in the increase of the size of the off-burst, is statistically significant at the

5 CORTICOTECTAL PATHWAY 185 INTACT SIDE ABLATED SIDE INTACT SIDE IOO- ABLATED SIDE w I- z I) : l- z W 0 40 a W CL I I ON- OFF RESPONSE lh FIG. 2. Distribution of units in the superior colliculus on the side of an intact and ablated visual cortex, based on the total number of discharges to the onset and to the offset of a disc of light in the center of the receptive field using 20 or more repeated stimulus presentations per unit. Abscissa: 1, on-response only; 2, on-response greater than offresponse by more than 507&; 3, on- and off-responses equal within SO%, 4, off-response greater than onresponse by more than SOq,; 5, off-response only. Number of units shown above each bar..oo 1 level (two-tailed Kolmogorov-Smirnoff test). The units which showed this effect did not seem to have a clear-cut distribution either in terms of the location of the receptive fields or in terms of the depth of recording. Animals studied with different survival times after visual cortex ablation did not show any notable differences. The effects for all the variables studied appeared to be similar among the six monkeys. Ocular dominance. A similarly subtle effect of visual cortex ablation was observed for ocular dominance. The majority of units studied both ipsilateral and contralateral to the ablation were binocular (94 and 99.5?& respectively). A slight shift in the ocular dominance distribution was observed on the ablated side. This is shown in Fig. 3, where data from the normal side were obtained in part from previous experiments (2). The increase in the spread in ocular dominance, though small, is significant at the,001 level (two-tailed Kolmogorov-Smirnoff test). Fovea1 refwesentation. Anatomical studies have indicated that retinal projections to the anterior part of the superior colliculus, which represents the central 5 of vision, are absent or very sparse in the mon l- z a W n 20-4 I_ I OCULAR DOMINANCE FIG. 3. Ocular dominance distributions for units in superior colliculi with overlying visual cortex intact or ablated. Abscissa: 1, response from contralateral eye only; 2, strong contralateral preference; 3, weak contralateral preference; 4, equal responses from both eyes; 5, weak ipsilateral preference; 6, strong ipsilateral preference; 7, response from ipsilateral eye only. Number of units shown above each bar. key (5, 9, 26). In two additional animals an attempt was made, therefore, to determine whether cells in this area depend solely on the geniculostriate pathway for their visual input. If this were the case, cells in fovea1 colliculus should be largely unresponsive to visual stimulation following ablation of visual cortex. Repeated plotting of the fovea with a reversible opthalmoscope, even before and after plotting each receptive field, is difficult to make sufficiently reliable for the absolute determination of receptive-field locations with respect to the fovea. However, the optical-lever and reference-unit techniques (see METHODS) provided a reliable measure of residual eye movement, thus allowing an accurate determination of receptive-field locations relative to one another. Since in the intact monkev there appears to be no significant overlap in the visual hemifields represented in the two colliculi (2, 19, 20), the line separating the receptive-field centers of units recorded in onecolliculus from those of units recorded in the other should represent the vertical meridian. Recordings were made in both colliculi of two animals following unilateral ablation of visual cortex. In one animal, studied 223 days after the cortical lesion, receptive fields of 27 units on the side ipsilateral

6 186 SCHILLER, STRYKER, CYNADER, AND BERMAN to the lesion and 21 units contralateral were plotted. In the other animal, studied 56 days after the cortical lesion, 17 receptive fields were plotted on the ipsilateral side and 10 on the contralateral side. Figure 4 shows the receptive-field plots from these two animals, corrected for drifts in eye position as determined by the opticallever and reference-units methods. Vertical and horizontal lines are drawn through the opthalmoscope plots of the fovea. The vertical meridian determined in this way, which assumes negligible torsional deviation, agrees well with that determined from receptive-field locations. This plot shows that after ablation of visual cortex, units representing the central 5 of the visual field are still numerous in the colliculus. These animals were studied a relatively long time after visual cortex ablation. How- ever, the central 5 was clearly represented in all eight of the monkeys studied after visual cortex ablation. Furthermore, every penetration in this area yielded units with brisk visual responses confined to normalsized receptive fields. Therefore this area appears to receive essentially normal visual input even after removal of visual cortex. As will be seen, similar results were also obtained with cooling. DEEP LAYERS. The most dramatic effect of visual cortex ablation became evident only below the superficial layers of the superior colliculus. At some point in the stratum opticum, it became impossible to activate units by visual stimulation. The multipleunit response, which is so characteristic in much of the superior colliculus of the intact animal, disappeared. Single units were still INTACT SIDE ABLATED SIDE FIG. 4. Receptive-field locations of 75 units in the superior colliculi of two monkeys with unilateral visual cortex ablation. Receptive fields drawn with heavy lines are from units obtained contralateral to the side of the lesion; receptive fields drawn with thin lines are from collicular units ipsilateral to the side of the lesion. Discontinuous lines represent fields from one monkey; continuous lines are fields obtained from the other.

7 CORTICOTECTAL PATHWAY 187 found, and they fired spontaneously, but we could not drive them using any sort of visual stimuli we could invent. Stratum opticum then is a transitional zone in this respect. Visual responses are still evident in the dorsal region but not in the ventral. In order to verify the depth at which units could no longer be driven by visual stimuli, small electrolytic lesions were placed at this point in two animals, studied 5 and 29 days after cortical ablation. Lesions were also placed in the colliculus contralateral to the ablation at the site where responses to light could no longer be elicited. An example of two such lesions is shown in Fig. 5. Visual cortex was removed over the superior colliculus on the right side. It can be seen that visual responses were obtained quite deep on the intact side; on the ablated side visual responses could not be obtained below the upper part of stratum opticum. In addition to localizing the recording sites by anatomical methods, we also inspected the perfused brains of animals to determine the extent of cortical ablation. It appeared that virtually all of area 17 was removed in these animals with minimal sparing in some cases in the deep concavity of the calcarine fissure in the region which falls anterior to a line in the coronal plane bisecting the lunate sulcus. Most of areas 18 and 19 were also ablated. In two animals with long survival times, retrograde cell changes were studied in the lateral geniculate nucleus, sections of which were stained by the Nissl method. The nucleus showed nearly complete degeneration sparing only a small number of cells. Most of the spared cells were in the medial and lateral parts of the ventral region of the geniculate. One of the animals from which we recorded was sacrificed 6 days following cortex ablation. Anterograde degeneration was studied by staining sections through the superior colliculus with the Fink-Heimer method. This showed a dense degeneration pattern similar to that reported by Kuypers (8) and Wilson and Toyne (26), which was heaviest in stratum opticum. Stratum griscum superficiale, by comparison, showed much lighter terminal degeneration. Recording in superior colliculus of paralyzed and anesthetized animals with reversible cooling of visual cortex In three monkeys a total of 64 superior colliculus cells were studied under conditions of cooling visual cortex. After locating units in that region of the superior colliculus which represented the same area FIG. 5. Photomicrograph of a section through the superior colliculi showing the site of two lesions placed at the point where visual responses could no longer be elicited. Left side is intact. Visual cortex above the superior colliculus on the right was ablated 5 days prior to the experiment.

8 -it-it- CElKiG l. ON 188 SCHILLER, STRYKER, CYNADER, AND BERMAN of the visual field as the region of cortex which was to be cooled, units were repeatedly stimulated with an optimal stimulus display while cortex was reversibly cooled. Of these units 72% were within 5 of the center of the fovea. The results obtained under these conditions complement those obtained with cortical ablation. The superficial layers were largely unaffected by cooling, while in the deeper layers units stopped responding to light when cortex was cooled. Figure 6 shows two units having their receptive fields in the same area of the visual field. The first unit was obtained near the surface of the colliculus, 220 p below the first audible signs of penetrating this structure with the microelectrode. Cooliqg had no discernible effect on the responses of this unit. The second cell, encountered 400 p below the surface, stopped responding to visual stimuli when cortex was cooled, and returned to normal responsivity after cortex was warmed again. Only 1 of the 14 superficial units studied showed a clear change during cooling to the onset and offset of a flashing spot in the center of the field. This cell showed a 50% decrement in the number of on- s--j--+- COOLED II OFF STIMULUS 1 I 500mrec FIG. 6. Response characteristics of two units in the superior colliculus before, during, and after cooling of visual cortex. A: unit 220 p below the surface of the colliculus. The lower trace represents shutter voltage. Stimulus: flashing spot centered in receptive field. B: unit 400 p below the surface of the colliculus. Lower trace is mirror galvanometer voltage. Stimulus: sweeping spot. discharges when cortex was cooled as compared with the responses obtained before and after cooling. Of the 50 deeper units studied, 45 had receptive fields located in that area of the visual field which was represented by visual cortex under the cooling probe. The visually elicited responses of all of these units were effectively disrupted during cooling. A more detailed picture of such a unit is shown in Fig. 7. This cell was located 1 mm ON OFF I second FIG. 7. Response histograms obtained from a single cell in the superior colliculus during cooling and rewarming of visual cortex. Each histogram represents 20 repeated presentations of.75o disc in the center of the receptive field. Onset and offset of the stimulus occur at the left edge of the letter 0.

9 CORTICOTECTAL PATHWAY 189 below the surface. This figure shows a series of response histograms taken in periodic samples during cooling and warming. The response to both onset and offset of the stimulus declined as cortex became cooler until the unit was completely unresponsive. Responses stopped when the temperature of the cortex, as measured by a thermistor 2 mm below the surface of the brain, reached 17 and was held there. After cortex was warmed this unit became slightly hyperresponsive. This was observed in some units but not in others. Two observations indicate that the effect of cooling on colliculus units was exerted through local suppression of the region of cortical area 17 immediately under the cooling probe. First, the superficial layers of the colliculus, which were closer to the cortical area cooled, were largely unaffected by cooling. This was true even at the end of the one experiment in which we accidently cooled the cortex to 3 C causing irreversible damage; the superficial unit recorded did not change its characteristics, while the visual response of the deeper units disappeared permanently. Second, the level of cooling required to eliminate the visual response of deeper units increased with increasing distance of the unit s receptive field from the area of the visual field represented near the center of the cooling probe. Receptive fields represented near the center of the cooling probe were readily suppressed by cooling; receptive fields more than 7 away appeared to be unaffected by cooling at the levels we employed. One of the questions these findings raise is whether or not a clear functional distinction can be made between the properties of cells which are and which are not influenced by cooling. While the aim of this study was not to answer this question in detail, some qualitative differences were noted. In general, cells unaffected by cooling had receptive-field properties previously found to be associated with superficial cells, while those influenced by cooling had properties associated with deep cells. Thus, the superficial cells, immune to cooling, tended to produce smaller signals with the same electrode, were more difficult to isolate, had smaller receptive fields, and responded quite consistently to repeated stimulation. Bv con trast, the deeper cells, whose responses were modified by cooling, were easier to isolate, had larger receptive fields, and showed response variability to repeated stimulation. The basic property of giving both on- and off-responses was maintained in both groups, as was the lack of specificity for stimulus shape or orientation. In order to determine whether or not cooling of visual cortex affects the contralateral superior colliculus, in one monkey two penetrations were made in the right colliculus while the cooling plate was on the left visual cortex. Five units were studied, three of which had their receptive fields 3 from the center of the fovea on the side contralateral to the effective cooling area. None of these units was affected by cooling. Recording in superior colliculus of alert animals with unilateral ablation of visual cortex SUPERFICIAL LAYERS. Thirty-eight visually responsive units were studied in the superficial layers of the superior colliculus of two alert monkeys on the side ipsilateral to cortical removal. The receptive-field properties of units, as studied through the immobilized eye of these animals, were similar to those found in the acutely prepared monkeys. Thus, the majority of the units in the alert animal also had properties indistinguishable from those seen in intact monkeys (2, 20). Ten units were found which had an abnormal patchiness in their receptive-field organization, giving unequal on- and off-bursts to a stimulus flashed in the center of the receptive field. An example of a unit discharging more vigorously to off than to on is shown in Fig. 8. The maintained activity of units in the alert animals was typically higher than in the acute animals. DEEP LAYERS. After the electrode advanced in the colliculus ipsilateral to the cortical lesion past some point in the stratum opticum, many units were encountered which could not be driven reliably by any visual stimuli we employed. A few of these units showed occasional high-frequency discharges when the monkev was looking around, but they did not seem to have vi-

10 190 SCHILLER, STRYKER, CYNADER, AND BERMAN 60- ON OFF 2 set FIG. 8. Response histogram obtained from a single unit in the superficial superior colliculus of an alert monkey. Thirty repeated presentations of a.750 spot in the center of the receptive field. Onset and offset of stimulus occur at the left edge of the letter 0. sual receptive fields although they did show less activity in the dark. Others maintained a variable activity but did not appear to be influenced by eye movements in light or in darkness or by any form of visual stimulation. Slightly deeper, eye-movement cells similar to those described previously were encountered (20). Thirty-four eye-movement cells were studied; of these twenty-seven were ipsilateral and seven contralateral to the side of visual cortex ablation. On the side contralateral to the lesion thev were indistinguishable from eye-movement cells seen in intact animals. On the ipsilateral side the activity of such cells prior to eye movement also appeared normal. This is shown in Fig. 9. Stimulation through the microelectrode produced saccades at similar current levels to those required to elicit eye movements in intact animals (2 l), and these elicited saccades duplicated the characteristics of the spontaneous saccades specifically associated with unit discharge. The motor fields of eye-movement units (20, 21, 27) on the ablated side were also normal, as shown in Fig. 10. Following the cortical lesion, however, one clear difference in the eye-movement units was found: they lacked visual receptive fields. In the intact animal, we have previously distinguished two types of eyemovement cells: the more superficially located ones, which have practically no spontaneous activity, a high percentage of which have visual receptive fields; and the deeper units, with relatively high maintained activity, which do not appear to have visual receptive fields (21). Of the 27 eyemovement units recorded ipsilateral to the cortical lesion, 21 were of the first, more superficial type with low spontaneous activity, the majority of which could be expected in the normal animal to have visual RECORDING AND STIMULATION, ABLATED SIDE I 500msec I RECORDING AND STIMULATION, INTACT SIDE FIG. 9. Recording and stimulation in the deeper layers of the superior colliculus. Responses of two eyemovement cells from the same animal are shown, one ipsilateral and the other contralateral to the side of visual cortex ablation. The first two columns show saccade-associated unit activity. The third column shows saccades elicited by stimulation through the microelectrode immediately following recording: 70- msec train of 5 pa,.5-msec pulses at 300 Hz.

11 CORTICOTECTAL PATHWAY FIG. 10. Motor field of an eye-movement unit in the superior colliculus of an alert monkey ipsilateral to the side of the lesion. This plot depicts the size and direction of saccades associated with unit firing. Open circles show saccades without accompanying unit discharge; filled circles represent saccades which were preceded by unit activity. receptive fields. Only 1 of these 21 units had a visual receptive field. These findings suggest that the ablation of visual cortex results in the virtual elimination of visual input to the deeper layers of the superior colliculus, including the eye-movement cells. DISCUSSION The results of this study show that ablation or cooling of visual cortex disrupts the transmission of visual information to the deeper layers of the superior colliculus. The major findings in the colliculus lacking visual cortex input are three: 1) the entire visual half-field is represented on the surface of the colliculus, despite anatomical claims of the absence of retinal projections to the perifoveal representation; 2) visual responsiveness disappears in units located below the superficial layers of the colliculus; and?) eye-movement units in the deeper layers of the colliculus still discharge before particular eye movements, but they no longer have visual receptive fields. These results lend themselves to an interesting comparison with three other species studied in this manner: the cat, the ground squirrel, and the rabbit. In the rabbit, vi- sual cortex does not appear to play a role in determining the receptive-field properties of collicular cells (10). In the cat, collicular receptive fields are similar to those of the monkey in many respects; most cells in the intact animal are binocular, show little or no shape specificity, and respond transiently to both onset and offset of a visual stimulus. The major difference is that in the cat the majority of the units (7073 are direction selective (1, 23). After cortical cooling or ablation, they are no longer directional or binocular (1, 18, 25). 0 ther aspects of receptive-field organization in the superficial layers are largely unaffected. A recent study by Stein and Arigbede (22) indicates that in the deeper layers, as in the monkey, cooling of visual cortex renders single cells unresponsive to visual stimuli. Studies on the ground squirrel yield a rather different result (14). This animal has an all-cone retina with crossed optic nerves. The superior colliculus has units in the upper layers which are direction selective and show little shape specificity. The deeper layers are reported to have a predominance of hypercomplex cells. Ablation or cooling of visual cortex does not affect the superficial cells; direction selectivity is still present, presumably obtained from the direction-selective retinal ganglion cells. The deeper layers, however, are affected dramatically; cells can no longer be driven by visual stimuli, although they retain spontaneous activity. The attribute of directionality in the cat is cortically mediated, while in the ground squirrel, as in many other rodents, it is already present in the retina. Binocularity in the cat superior colliculus appears to be mediated by visual cortex, while in the monkey it seems to have emerged without much reliance on connections from visual cortex. These findings are in harmony with the anatomy of this system. In the cat the retinal projection to the superior colliculus is primarily contralateral, while in the monkey the eye projects equally to both colliculi. In the deeper layers, disruption of the corticotectal input from visual cortex has similar effects in the cat, ground squirrel, and monkey. Of the animals studied, the rabbit appears to be the only exception.

12 192 SCHILLER, STRYKER, CYNADER, AND BERMAN The comparison among different species must largely be restricted to the sensory aspect of the superior colliculus, since those properties which are related to motor output have not been extensively studied. Single-unit activity related to eye movement has been found in both cat and monkey (20, 21, 24, 27), but it is only in the latter species that a clear relationship has been shown to exist between the sensory and motor maps of this structure (19, 21, 27). The fact that the eye-movement responses of cells in the deeper layers of the monkey superior colliculus are still present after ablation of visual cortex is consistent with the observation that some of these cells respond before eye movement in the dark and with calorically induced vestibular nystagmus (21). Since the response characteristics of these cells depend only in part on visual input, they must receive extensive projections from other, nonvisual structures involved in the control of eye movement. When our results are compared with the recent anatomical studies on the nature of cortical and retinal input to the superior colliculus, an apparent contradiction emerges. The anatomical studies have shown that in the monkev the retina does not project to the anterior portion of the colliculus or does so only very sparsely (5, 9, 26). By contrast, fibers from visual cortex project to the entire colliculus, showing especially dense termination in the anterior region. Our results do not fit these findings; after ablation of visual cortex, units having receptive fields in the fovea1 area are still plentiful and have properties similar to units with receptive fields further out in the visual field. Several possibilities may be considered in dealing with this contradiction. One is that our recordings are in error. We doubt this, for care was taken in our work not only to verify the fovea1 area and to monitor residual eye movements, but also to obtain a sufficient sample in our mapping procedures so that a large portion of the visual field was covered with receptive fields, counteracting the possible effect of any error in plotting the fovea. A second possibility, anatomical reorganization following the cortical ablation, seems also unlikely to account for our find- ing collicular receptive fields representing the central retina, since cooling the cortex left the superficial cells within the central 5 of the visual field unaffected. A further possibility is that the anatomical methods have not to date been sufficiently sensitive to detect retinal terminals in the anterior part of the colliculus representing central vision, or that the survival times used have been wrong for this area. Whether or not this is so remains to be seen. A final possibility is that the fovea1 area in the colliculus is innervated not directly from the retina but from a structure other than visual cortex; it is this other structure, perhaps the lateral geniculate nucleus, which receives the visual input and then relays it to the colliculus. While this alternative cannot be ruled out, it does not seem very likely to us. Were it so, our recordings should probably have reflected some inherent differences in the central and more peripheral representations of the visual field in the colliculus. We have not seen this. The onlv clear difference between central and peripheral representations is receptive-field size. Furthermore, any candidate for such a structure should itself have a binocular projection from the retina, since the fovea1 representation is no less binocular following the cortical lesion than are the more peripheral representations in the colliculus. Most of the known sites of retinal terminations in the monkev other than the lateral geniculate nucleus and the superior colliculus are chiefly contralateral (5). Although the cooling c and ablation methods in these studies demonstrate the dependence of collicular function on visual cortex, they provide only margina.l cues about the exact nature of the cortical information which reaches the colliculi. In the monkey one may consider several possibilities. One is that the visual receptivefield properties of deeper collicular cells are determined by visual cortex. This would seem to necessitate a great deal of convergence from the cortex so that several cells with preferences for different orientations would impinge on a single collicular cell; in this way the lack of orientation specificity could come about. Recent work suggests that in the cat there is indeed a

13 CORTICOTECTAL PATHWAY great deal of corticocollicular convergence (11, 12). The equivalence of on-off responses and binocularity would also be produced by cortical convergence, since in the monkev most cortical cells do not seem to produce equivalent on-off responses to flashes and only 50y0 of them are binocular (6, 7, 17). Conversely, one could assume that there is a population of binocular cortical cells which have properties similar to those observed in the colliculus. A recent study by Poggio (17) suggests that in the monkey area 17 there may be a higher proportion of nonoriented cells than has previously been reported. Recent work on the cat has attempted to identify, by antidromic activation, those cortical cells that send their axons to the colliculus. These experiments have shown that most such cells fall into the category of complex neurons (15, 16) which have orientation and direction specificity. Unfortunately none of these considerations can answer how the superficial cells in the colliculus might influence the cells in the deeper layers under normal conditions. Perhaps one should consider the possibility that cortex modulates the flow of visual information from the superficial down to the deeper layers. Such a hypothesis might not necessitate a multiple convergence notion, since in this case the properties of the deeper cells may be assumed to be derived from the superficial cells. SUMMARY 1. This study investigated the influence of corticotectal connections by studying the response characteristics of single cells in the superior colliculus of the monkey during cooling or after ablation of visual cortex. 2. The receptive-field properties of single cells in the stratum griseum superficiale and in the dorsal part of the stratum opticum are largely unaffected by ablation of visual cortex. Some cells do not, however, respond to visual stimuli throughout their receptive fields as uniformly as do collicular cells in the intact animal. Most cells are still binocular but the ocular dominance distribution is slightly broader. 3. In the deeper layers of the superior colliculus visual responses can no longer be elicited from single cells after visual cortex ablation. 4. In agreement with these findings, cooling of visual cortex has little or no effect on the cells in the superficial layers of the superior colliculus but disrupts visual responses below this region. 5. The central 5 of the visual field continues to be represented on the surface of the colliculus following either ablation or cooling of visual cortex, despite anatomical evidence for the virtual absence of a fovea1 retinotectal projection. 6. Eye-movement cells in the deeper layers of the superior colliculus of the alert monkey are still present after ablation of visual cortex but they no longer have visual receptive fields. 7. The results suggest that visual cortex plays a prominent role in controlling the flow of information to the deeper layers of the superior colliculus of the rhesus monkey.,\cknowledgments We thank Ms. Susan Volman and Ms. Cynthia Richmond for their assistance. This research was in part supported by National Institutes of Health Grants EY00676 and EY00 756, a Sloan Foundation Grant, and Public Health Services Training Grant GM Present address of M. CY nader: Dept. of ogy, Dalhousie Universi tyy Halifax, Nova Psvchol- Scotia. REFERENCES 1. BERMAN, N. AND CYNADER, M. Comparison of receptive-field organization of the superior colliculus in Siamese and normal cats. J. Physiol., London 224: , CYNADER, M. AND BERMAN, N. Receptive-field organization of monkey superior colliculus. J. Neurophysiol. 35: , DOANE, M. G. AND DOHLMAN, C. H. Physiological response of the cornea to an artificial epithelium. Exptl. Eye Res. 9: , GOLDBERG, M. E. AND WURTZ, R. H. Activity of superior colliculus in behaving monkey. I. Visual receptive fields of single neurons. J. Neurophysiol. 35: , HENDRICKSON, A., WILSON, hl. E., AND TOYNE, M. J. The distribution of optic nerve fibers in Macaca rnulatta. Brain Res. 23: , HUBEL, D. H. AND WIESEL, T. N. Receptive fields and functional architecture of monkey striate cortex..i. Physiol., London 195: , HUBEL, D. H. AND WIESEL, T. N. Stereoscopic

14 SCHILLER, STRYKER, CYNADER, AND BERMAN vision in the macaque monkey. Nature 225: 41-42, KUYPERS, H. G. J. M. AND SWARCBART, M. Discussion. In: Interhemispheric Relations and Cerebral Dominance, edited by V. B. Mountcastle. Baltimore: Johns Hopkins Press, 1962, p. 115-l 16. LUND, R. D. Synaptic patterns in the superficial layers of the superior colliculus of the monkey, Macaca mulatta. Exptt. Brain Res. 15: , MASLAND, R. H., CHOW, K. L., AND STEWART, D. L. Receptive-field characteristics of superior colliculus neurons in the rabbit. J. Neurophysiol. 34: , MCILWAIN, J. T. Retinotopic fidelity of striate cortex-superior colliculus interactions in the cat. J. Neurophysiol. 36: , MCILWAIN, J. T. Topographic relationships in projection from striate cortex to superior colliculus of the cat. J. Neurophysiol. 36: , MCILWAIN, J. T. AND BUSER, P. Receptive fields of single cells in the cat s superior colliculus. Exptl. Brain Res. 5: , MICHAEL, C. R. Visual receptive fields of single neurons in superior colliculus of the ground squirrel. J. NeurophysioZ. 35: , PALMER, L. A. An Electrophysiological Study of a Corticotectal Pathway in the Cat and the Effects of Its Removal on Single Units in the Superior Colliculus (Dissertation). Philadelphia: University of Pennsylvania, PALMER, L. A., ROSENQUIST, A. C., AND SPRAGUE, J. M. Corticotectal systems in the cat: their structure and function. In: Corticothalmic Projections and Sensorimotor Activities, edited by T. L. Frigyesi, I. Rinvick, and M. D. Yahr. New York: Raven, POGGIO. G. F. Spatial properties of neurons in striate cortex of unanesthetized macaque monkey. Inuest. Opthalmot. 11: , RIZZOLATTI, G., TRADARDI, V., AND CAMARDA, R. Unit responses to visual stimuli in the cat s superior colliculus after removal of the visual cortex. Brain Res. 24: , ROBINSON, D. A. Eye movements evoked by collicular stimulation in the alert monkey. Vision Res. 12: , SCHILLER, P. H. AND KOERNER, F. Discharge characteristics of single units in superior colliculus of the alert rhesus monkey. J. Neurophysiol. 35: , SCHILLER, P. H. AND STRYKER, M. Single-unit recording and stimulation in superior colliculus of the alert rhesus monkey. J. Neurophysiol. 35: , STEIN, B. E. AND ARIGBEDE, M. 0. Unimodal and multimodal response properties of neurons in the cat s superior colliculus. Exptl. Neural. 36: , STERLING, P. AND WICKELGREN, B. G. Visual receptive fields in the superior colliculus of the cat. J. Neurophysiol. 32: 1-15, STRASCHILL, M. AND HOFFMAN, K. P. Activity of movement sensitive neurons of the cat s tectum opticum during spontaneous eye movements. Exptl. Brain Res. 11: , WICKELGREN, B. G. AND STERLING, P. Influence of visual cortex on receptive fields in the superior colliculus of the cat. J. Neurophysiol. 32: 16-23, WILSON, M. E. AND TOYNE, M. J. Retino-tectal and cortico-tectal projections in Macaca mu- Zatta. Brain Res. 24: , WURTZ, R. H. AND GOLDBERG, M. E. Activity of superior colliculus in behaving monkey. III. Cells discharging before eye movements. J. Neurofihvsiol. 35: , 1972.