THE JOURNAL OF COMPARATIVE NEUROLOGY 314~534-544 (1991)

Receptive Field Properties of Somatosensory Neurons in the Cat Superior Colliculus H. RUTH CLEM0 AND BARRY E. STEIN Department of Physiology, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia 23298-0709

ABSTRACT In general, knowledge of the internal organization of receptive fields has played an important role in shaping current understanding of sensory physiology. Such knowledge is particularly important for understanding the function of the superior colliculus, since this structure is at once implicated in spatial localization and has relatively large receptive fields. While this issue has been addressed in the visual and auditory modalities represented in the superior colliculus, there are no previous studies of its somatosensory receptive field organization. Here, the properties of somatosensory receptive fields in the cat superior colliculus were studied quantitatively to determine whether they contain internal non-homogeneities that might aid in the determination of stimulus detail. Of special interest was the possibility that these comparatively large receptive fields would contain areas of differential excitability that could aid in spatial resolution, that within-field spatial summation and/or inhibition would be exhibited, and that the borders of the excitatory receptive field would be flanked by inhibitory regions. The data demonstrate that while inhibition beyond the receptive field borders is a rarity, these somatosensory receptive fields nearly always contain a well-defined area of maximal sensitivity within which the size of the stimulus is a critical feature in determining the magnitude of the response. These best areas are systematically distributed across receptive fields as a function of their location in the structure, and indicate that the resolution of stimulus location and size may be greater than expected on the basis of receptive field size alone. Key words: tactile, in-field summation, surround inhibition, spatial resolution

Somatosensory receptive fields are organized in detailed map-like representations of the body surface in various areas of the mammalian brain. Many of these representations are believed to underlie the discriminative and affective functions critical to tactile perceptions. However, the somatosensory map in the superior colliculus is believed to be more closely linked to immediate attentive, orientation and localization functions, than to other tactile functions. To this end, there is an expected expansion of the representation of the face and forelimb, regions particularly important in the orientation to and manipulation of objects (e.g., food) and the rapidly adapting response characteristics that seem best suited for dealing with brief, changing stimulus characteristics (Gordon, '73; Stein et al., '76; Clemo and Stein, '84, '86a). Similarly, there are few deep receptor inputs here, and the neurons categorized as "cutaneous" are optimally activated by stimuli that are moved across the hair and skin rather than those that vertically indent the tissue. While somatosensory receptive fields of many superior colliculus neurons are large (i.e., covering one or more body

o 1991 WILEY-LISS. INC.

regions; see Meredith et al., '90, '91) and are arranged in a coarse topography (Stein et al., '76; Nagata and Kruger, '79; Clemo and Stein, '84, '86a), these features have been thought to represent the sacrifice of spatial resolution at the single neuronal level for the greater sensitivity of detection produced by activating many neurons with overlapping receptive fields (Stein et al., '76). Nevertheless, precision in localizing a stimulus may result from the fact that the population of simultaneously activated neurons will have receptive fields with only a small region of the body surface in common, a principle similar to that postulated for visual superior colliculus neurons (see McIlwain, '75; Meredith and Stein, '90). Although somatosensory neurons in the superior colliculus have preferences for the axis and velocity of movement across the receptive field, can code stimulus intensity, respond differently to stimuli of Accepted September 11, 1991. Address reprint requests to H. Ruth Clemo, Dept. of Physiology, Medical College of Virginia, Virginia Commonwealth University, Box 551, MCV Station, Richmond, VA 23298-0709.

SOMATOSENSORY PROPERTIES IN SUPERIOR COLLICULUS different sizes, and exhibit within-field heterogeneity (Stein et al., ’76; Nagata and Kruger, ’79; Clemo and Stein, ’86a,b; Clemo and Stein, ’87), little is known about the internal organization of their receptive fields. It seemed possible that these somatosensory neurons, like their neighboring auditory neurons, contain “best areas” within their receptive fields which could considerably increase their spatial resolution (Clemo and Stein, ’86a,b). In this study, the internal organization and spatial analyzing properties of somatosensory receptive fields in cat superior colliculus were studied quantitatively. It was of specific interest to determine whether these receptive fields were organized to produce within-field spatial summation and/or inhibition, whether best areas could be discriminated, and whether their excitatory regions are bordered by inhibitory zones. Preliminary findings have been presented in abstract form (Clemo and Stein, ’86b, ’88).

MATERIALS AND METHODS All of the following procedures were performed in compliance with the Guide for Care and Use of Laboratory Animals (NIH Publication No. 86-23) a t Virginia Commonwealth University, which is accredited by the American Association for Accreditation of Laboratory Animal Care (AAALAC1.

Surgical preparation Each cat was anesthetized with sodium pentobarbital (40 mg/kg, i.p.1 and maintained with small supplemental doses ( 5 mgikglhour, i.v.). Body temperature was maintained throughout the surgical preparation at approximately 38°C with a heating pad. The animal’s head was immobilized in a Kopf stereotaxic head-holder and a cranial opening was made over the visual cortex allowing access to the superior colliculus. A metal cylinder was then implanted over the opening and secured to the skull with dental acrylic and anchoring screws (McHaffie and Stein, ’83). During recording, the implanted cylinder was fitted with an X-Y slide that allowed the precise location of electrode penetrations. Following surgery, the animals received analgesics (Meperidine, 5-10 mg/kg) and antibiotics (Flocillin, lcc/day) as necessary and were allowed to recover for 10-14 days before the first recording session.

Recording For recording, the metal implant was attached to a support arm mounted to a stereotaxic frame, thereby supporting the body without pressure points. Despite the absence of wounds or pressure points during recording, the animals were anesthetized with ketamine hydrochloride (20 mg/kg). They were immobilized with gallamine triethiodide (1.5 cc, initial dose, followed by 0.5 cc/hour) administered intravenously and artificially ventilated with a mixture of 75% nitrous oxide and 25% oxygen; the expiratory CO, was monitored with a Beckman CO, analyzer and kept at approximately 4%. Body temperature was maintained at 38°C with a heating pad. EEG activity was recorded through an anchoring screw over the frontal cortex and routed through an amplifier to an oscilloscope and monitored continuously. Anesthetic levels were adjusted when necessary to maintain a synchronous, high amplitude EEG. Standard extracellular recording procedures (glasscoated tungsten microelectrodes) were used to record from superior colliculus neurons. When searching for a somato-

535

sensory neuron, the electrode was slowly advanced while delivering a variety of innocuous mechanical stimuli. These included the manual distortion of the skin and underlying tissue, rotation of the joints, air puffs and light brushing across the hairs or skin with camel’s hair brushes or von Frey hairs. When a responsive neuron was isolated, it was identified as cutaneous (skin or hair) or deep (muscle or joint) and as high or low velocity by using established criteria (Iggo, ’77). It was also categorized as sustained or transient, depending upon whether or not it responded to a maintained stimulus with a maintained discharge. The receptive field was mapped by using the minimal effective stimulus (see Mountcastle, ’57), and classified as small (area = 50 cm2).Receptive fields were also assigned to one of the six regions into which the body surface was divided: face, forepaw, forelimb, hindpaw, hindlimb and trunk/belly/tail. An electronically controlled (Ling 102A) mechanical probe was then used to deliver controlled tactile stimuli. The probe was used: (1)to determine the threshold stimulus velocity and displacement amplitude that would activate each neuron at various sites throughout the receptive field. This was accomplished by a series of tests that consisted of a minimum of ten trials (1per 15 seconds) for each of several velocities (0-450 mm/ms) and displacement amplitudes (0.5-5 mm). At each site the probe tip was positioned so that it just touched the cutaneous surface, and by manipulating velocity and amplitude of displacement, the threshold and optimal stimulus were determined. The threshold was defined as a response to 50% of the stimulus presentations (10 trialsilocus), the optimal stimulus was defined as that which evoked the greatest number of impulsesiresponse (optimal response); (2) to determine whether the receptive field contained subareas of different sensitivities. The receptive field was divided into a grid system of squares, with each square 5-10 mm in diameter; the tactile probe was positioned within each square and the optimal stimulus was delivered. The average optimal response at each locus was measured and the locus from which the highest number of impulses was evoked was considered the point with the “best” response. Optimal responses at other loci were calculated as a percentage of this “best” response, and those in which the response was greater than 70% of the best response were considered to be within the “best area.” Once these tests were completed, similar tests with “suboptimal” stimulus velocities were conducted; and (3) to determine whether these neurons would show spatial summation (i.e., a greater response to larger stimuli), spatial inhibition (i.e., weaker responses to larger stimuli), or surround inhibition (i.e., simultaneously stimulating inside and outside the exicitatory receptive field would produce fewer impulses than stimulating only within the field). These spatial characteristics were examined by using one or two mechanically controlled tactile stimulators. Responses to simultaneous stimulation of two points within the receptive field were determined (within field summation/inhibition) as well as responses to one probe within and one outside of the receptive field (surround inhibition). Best areas and suboptimal areas were examined. A typical testing paradigm was as follows: (1) one probe delivered stimuli within the best area, and the threshold velocity and amplitude of displacement required to elicit a response were determined; (2) a second probe delivering stimuli identical to the first was placed successively at different sites either within the receptive field or

536

H.R. CLEM0 AND B.E. STEIN High Velocity

Low Velocity

1

/

/

f

Impulse Nmber

Impulse Frequency (Hz)

/

Peak Impulse Number

I

A /

Peak Impulse Frequency (Hz)

>

7 0 % of best response

4 0 69% of best response

c? 0

4

-

l

10-39% of best response

0 - 9 % of best response

Fig. 1. Somatosensory receptive fields have nonhomogeneous organizations. The responses evoked by high (250 mmisecond, left) and low (20 mmisecond, right) velocity stimulation of six sites within a receptive field are shown as pairs of rasters and peristimulus time histograms. The downward portion of the electronic trace above each pair of raster-histograms represents the movement of a mechanical stimulus that depressed the hairs (4 mm amplitude) without direct displacement of the skin. The rasters display responses to each of 10 stimulus presentations and each dot in the raster represents one impulse. The asterisk marks the site from which the greatest number of impulsesiresponse could be elicited (“best response”). The differential

sensitivity of the receptive field is indicated by shading and is explained by the key at the bottom. The best area of the receptive field is shown in black and yielded the highest number and frequency of impulses. It was flanked by areas in which the stimulus effectiveness was progressively degraded (cross hatched and dotted), and finally by areas in which responses were rarely evoked. This receptive field organization was apparent with each of the four response measures used (center: e.g., impulse number, etc.) and over a wide range of effective velocities. However, note that at the lower velocities (right),the internal organization could only be seen by using impulse frequency as the response measure.

outside its excitatory borders; and (3) responses to stimulation of each site independently, followed by simultaneous stimulation of both sites, were collected and compared. Spatial summationiinhibition was also examined by changing the size of the stimulus applied at the same receptive field locus. Different-sized probe tips were attached to the tactile stimulator with diameters of 0.5, 1, 1.5, 5 , 10, 15, and 25 mm and each was used to deliver a series of 10

successive stimuli at the optimal velocity and at the same receptive field locus. Neuronal responses to these stimuli were amplified and routed to a MINC PDP 11/23 computer for storage and subsequent analysis. The computer was used to construct peristimulus time histograms and response rasters, as well as to count impulses and calculate response latency. To facilitate histological identification of the recording sites, a

SOMATOSENSORY PROPERTIES IN SUPERIOR COLLICULUS

537

A.

c.

r ’ O

-.8

a

0

D In

-.6

$m

5001

-.t: 400 - 300 E

E

h

._ 4.

0

-.4

t

O

t 1

2

3

4

5

U

-.2

4 g F -..

-0 ‘

tI

-

5

0

200 -

r

cP

100-

0

i.

6

Receptive Fig. 2. Receptive field sensitivity degrades monotonically outward from the best area. In A, the receptive field (figurine) is illustrated above rasters and histograms of the responses elicited at six stimulation sites. The greatest sensitivity is apparent in the best area (positions 1-21 in terms of average number of impulses evoked and response

small current (12 PA, 14 seconds) was passed through the recording electrode at the termination of an electrode penetration. At the end of the experiment, the animal was given a lethal dose of sodium pentobarbital (80 mgikg, i.v.), and perfused through the aorta with physiological saline followed by 10% formalin. The brain was removed and frozen sections through the superior colliculus were mounted and stained with cresyl violet. Electrode tracks and recording sites were then reconstructed.

RESULTS A total of 140 somatosensory neurons were studied in ten cats; they were located in all deep laminae (below stratum opticum) of the superior colliculus and were spread throughout its rostro-caudal and medio-lateral extent. Many of the neurons responded with a train of impulses (average response duration = 40 ms) when the stimulus was brushed across the cutaneous surface or vertically indented the skin,

1

1 2

3

4

5

6

Field Position probability (B, lower left), as well as threshold velocity (C, lower right). A gradual reduction in the sensitivity to mechanical stimuli is apparent as progressively more eccentric points are tested. Vertical bars through data points indicate standard errors of the mean in this and subsequent figures.

and often they responded with a brief burst of activity to its removal. All neurons responded to a wide range of stimulus velocities, but for the most part (79%) their optimal velocities were high ( > 5 0 mm/second) and they responded poorly, if at all, to stimuli of low velocity ( < 20 mmisecond).

Receptive fields are non-homogeneous Rarely (n = 3) was a neuron equally responsive to the same stimulus at different points within its receptive field. Rather, a specific region yielded a substantially higher number and frequency of impulses than did other subregions to the same stimulus. This was quite evident even during the initial manual mapping of receptive fields. Usually this “best area” had a particularly effective point within it where the “best response” was elicited, and the entire area was located away from the geographic center of the receptive field, The best area was flanked by an area in which the effectiveness of the stimulus was progressively

H.R. CLEM0 AND B.E. STEIN

538 ..

. .

..

.

.

.-.. -. .-

.

.

.. . . ... .. . .. -. .

... . .

.

-.. . . .. .

. .

.

H

0

0 0.5mm

2.5 mrn

5 mm

10 mm

0

-

500 6

500

1

N

N

I

I 400-

v

400

, 0

> c u

c

a, 3

a 300

-5 0-

300

??

E

LL

a,

200

$ -

200

E

3

a E

100

-

100

IX

d

01, , I 0.5 2.5 5

Lo

5 a,

a

IX

rb

15

o;

0

0525 5

10

15

20

0

Probe Size (mmdiameter) Fig. 3. Within-field spatial summation. This neuron responded optimally to high velocity (250 mmisecond) displacement of guard hairs within its receptive field; shown here are its responses to vertical depression of the hairs. Different-sized probe tips were used and are shown below each raster-histogram. They were centered at the same site as shown in the figurine. Stimulation with the smallest (0.5 mm) probe tip evoked the weakest response and the lowest response

reliability (60%), while the strongest and most reliable response (100% probability) was evoked using the largest (20 mm) probe tip. Note that although there was also an increase in the response duration with larger probe sizes, the latency and duration of the peak response remained constant. Quantitative evaluations of the effect of the stimulus size on various response measures are shown graphically below.

degraded, followed by an even more peripheral area in which responses were rarely evoked (Fig. 1). For quantitative evaluation of the spatial heterogeneity of excitability, identical stimuli were presented sequentially at predetermined points throughout the receptive field. To ensure the greatest sensitivity of the test, the stimulus selected for each neuron was the one that evoked the greatest number of impulses (i.e., the “best response”). This stimulus, operationally defined as having optimal size and velocity, always proved to be the most effective stimulus in other areas of the receptive field as well. The response it elicited from each test site was expressed as a percentage of the best response. For the most part, subregions within the receptive field could be grouped into 3-4 divisions. The first, or “best area,” was defined as that region containing sites from which the optimal response evoked was 70% or more of the best response. This region varied in size, but almost always formed a spatially contiguous zone. Two and sometimes three additional areas were identified from which progressively lower proportions of the best response were evoked: 40-69%, 10-39%, and 1-9%. In each of 28 neurons studied in detail, there were significant (P < 0.05) differences in the average number of impulses evoked from each of these

regions. A typical example is illustrated in Figure 1. Here, the small (% the length of the receptive field), eccentrically placed best area was flanked by two progressively larger but less sensitive areas. The average number of impulses elicited at a point near the receptive field border was quite high (23%of the best response) and a sharp, continuous boundary marked the edge of the excitatory receptive field. To determine if the choice of response measures used in these experiments might influence the representation of receptive field organization, the mean and peak impulse numbers and frequency of the discharge trains evoked from each neuron at different stimulus velocities were calculated and compared. When the optimal stimulus velocity was employed, it did not matter which response measure was used to determine the spatial organization of the receptive field; they were identical. However, at non-optimal velocities, only the mean number and frequency of the response were useful for discriminating among receptive field subregions, and this is shown in the illustrative example provided (Fig. 1). Just as the number of impulses evoked by the optimal stimulus varied with stimulus position in the receptive field, so did threshold velocity (the minimum velocity required to elicit a response on at least 50% of the trials),

SOMATOSENSORY PROPERTIES IN SUPERIOR COLLICULUS

539

and both produced the same spatial profile: a low threshold stimulus size outside the best area produced comparatively best area flanked by areas with progressively higher thresh- small and statistically insignificant changes in activity. olds (see Fig. 2). As shown in the case illustrated in Figure Large receptive fields had generally poorly defined internal 2, a comparatively steep rise in threshold velocity and a organizations and exhibited little ability to deal with signifconcomitant decline in evoked activity and response proba- icant changes in stimulus size; large stimuli were the most bility occurred as the stimulus was positioned from the best effective, while small stimuli tended to elicit strong rearea towards the periphery, thus making it possible to sponses only from the best area. The second test method (2 simultaneous stimuli within define a sharp excitatory receptive field border. This was generally the case when the receptive field was small or the receptive field) revealed that significant (P < 0.05) when the borders proved to be less than 1 cm from the best spatial summation could be produced (n = 8 / 16) by combinarea. However, in those large receptive fields in which there ing a stimulus within the best area with a stimulus outside was a gradual decline in responsiveness over a wide area so it. This indicates the suprathreshold enhancement capabilthat a fourth, least excitable area of the receptive field was ity of eccentric regions. A typical case is illustrated in evident, there was considerably more doubt in determining Figure 5 . Here, a small probe (5 mm diameter) applied to the best area (neck) produced a consistent but weak rethe borders. Large receptive fields tended to have the largest best sponse that was significantly (P < 0.05) enhanced by an areas, but these areas constituted proportionately less of eccentric stimulus (Fig. 5A). Increasing the size of the the total receptive field than was the case in small receptive stimulus applied within the best area (Fig. 5B) also signififields. For example, the smallest receptive fields (average cantly (P < 0.05) enhanced the response. area = 6 cm') were located on the forepaw and contained Spatial inhibition the smallest best areas (X = 1.5 cm'), yet they constituted 25% (range = 13-31%) of the receptive field. IntermediateA small number (4121) of neurons showed spatial inhibition-a monotonic decrease in the activity evoked as the = 20 cm2),such as those found on sized receptive fields size of the probe increased (Fig. 6) without evidence of the forelimb, had larger best areas = 2.7 cm'), but they constituted a Comparatively small (14%, range = 6-27%) spatial summation at either extreme of the range of stimuportion of the receptive field. The largest receptive fields lus sizes used. Furthermore, it was not possible to predict encompassed the entire contralateral trunk and limbs and whether a neuron would show spatial summation, spatial contained the largest best areas ( g = 29.9 cm'), which inhibition, or no change in response to these treatments made up the smallest proportion (8%, range = 1-13%) of based on its receptive field location or size. In the example in Figure 6, the receptive field was very similar in location the receptive field. The location of best areas followed the somatotopy so and size to that shown in Figure 3, and in both cases the that their positions shifted within receptive fields from the neurons were cutaneous (hair), rapidly-adapting, and prefront to rear of the body as neurons were sampled from ferred high velocity stimuli. Nevertheless, while the neuron rostra1 to caudal in the superior colliculus. The systematic in Figure 3 showed statistically significant spatial summashift in best areas was apparent despite the presence of tion, the one in Figure 6 showed statistically significant large overlapping receptive fields encountered across much of the structure.

(w

(w

Spatial summation Two methods were also used to determine if stimuli of different sizes would affect responses differently. In the first, neurons (n = 26) were tested with stimuli of graded sizes that were applied at the same locus either within or outside the best area of the receptive field. In the second, neurons (n = 16) were examined with two identical stimuli that were applied simultaneously within the receptive field. With the first method, increasing the size of the stimulus (probe diameter range: 0.5-20 mm) applied within the best area of the receptive field affected the number of impulses evoked in most (21126) of the neurons tested. The majority of these (17/21), exhibited a monotonic increase in impulse number and frequency as the probe diameter increased, without any evidence of spatial inhibition at either extreme of the range of stimulus sizes used. In the example in Figure 3, increasing the size of the probe tip 40-fold (from 0.5 to 20 mm) resulted in a 388%increase in the number of impulses. In several (n = 7) of the neurons exhibiting spatial summation within their best areas, the same tests were repeated in more eccentric receptive field locations. Spatial summation was evident less frequently in these regions (n = 3/7), and in these examples, the small-intermediate sized receptive fields made it difficult to rule out mechanical transmission to the best area as the stimulus increased in diameter. In the remaining neurons (n = 417, spatial summation was never noted outside the best area, as illustrated in the example in Figure 4A. Even a 40-fold increment in

Fig. 4. Most neurons displayed spatial summation inside their best area but not at eccentric positions. Here, the best area is indicated as the shaded area, and the stimulation points as filled circles. As the probe size was increased within the best area from 0.5 to 20 mm, there was a dramatic increase in the number of impulses elicited. This same effect was not observed outside the best area.

540

H.R. CLEM0 AND B.E. STEIN

Fixed SiteA

Test Site

A and Test Site

B

6-

E .- 5-

c -%

w

\

u)

z

4-

/-----

3

3-

.c 0

........ ...... ..... ..... .......... . -.. . . . ...

... .- . ...... : .... ....... . .. .. .

.. .. .

2 2-

...

1

1

L

s

z

I-

0-

20

15

5

2.5

Probe Size (mm diameter) Fig. 5. Within-field summation. In A, stimulation loci within (A,1,2) and outside (3,4) the receptive field (shaded area of the figurine) are indicated. The responses to comhined stimulation of “A” (the fixed site) and each of the test sites (1-4) are shown as histograms and rasters to the left and in the bar histogram in the lower right. When the fixed stimulus a t A was comhined with a stimulus inside the receptive field, the average number of responses evoked was significantly great-

er (P < 0.05) than the response to stimulation to any single site; stimuli outside the receptive field had no significant influence. In B, increasing the diameter of the stimulus probe (indicated by the concentric rings on the figurine) at site 1 also resulted in significant (P < 0.05) response summation (bar histogram, left; histogram/ rasters bottom right). Scale bar in ms.

541

SOMATOSENSORY PROPERTIES IN SUPERIOR COLLICULUS

7 7 7 ....

... ... ..-. ..... ... . .... ._.

,. . . ...

5mm

0

. . . ._.-.. .... .. . .

.

....... .....

.

.

.. .. . ... . . . . . ........ .-. . . . . . . . . ,.

300

-0

N

I 200> c V a 3 , 0-

E?

LL

100

-g 3

a

IX

E -

‘1

IX

O 5

L 0

2’

10

0

5

10

1 2

Probe Size (mm diameter) Fig. 6. Within-field spatial inhibition. Although superior colliculus receptive fields generally lacked within-fieId inhibition, four examples of this phenomenon were noted. In this example, stimuli applied at the optimal velocity (250 mmisecond) displaced the hairs within the

receptive field and produced a monotonic decrement in the number and frequency of impulses evoked as the probe size increased from 5-20 mm in diameter.

spatial inhibition. The only consistent observation was that both effects (either spatial summation or inhibition) were most pronounced at optimal stimulus velocities.

sponses of 68% (n = 19/28) of the neurons were unaffected by stimulation outside the receptive field borders. This result was independent of the specific stimulus site (Fig. 7). In a few neurons (n = 9/28), stimulation outside the receptive field did result in a significant (P < 0.05) increase (n = 7/28) or decrease (n = 2/28) in the number of impulses elicited from the best area, but no spatially continuous inhibitory or subthreshold excitatory regions outside the receptive field could be defined. Furthermore, in each case, displacement of the “modifying” stimulus by as little as 1 mm eliminated the effect. The second method also provided very few examples of surround inhibition. In only a few examples (4/ 15)was there a consistent decrease in the number, frequency, and reliability of the stimulus when it straddled the receptive field border (see Fig, 8).

Receptive field surrounds Inhibitory areas surrounding receptive fields have been observed frequently in visual and auditory neurons in the superior colliculus (Middlebrooks and Knudsen, ’87; Ogasawara et al., ’84).Presumably, the same property would be evident in their neighboring somatosensory neurons. Therefore, the influence of areas bordering the excitatory receptive field was examined in 43 cutaneous neurons. In 28 of these, the number of impulses evoked by stimulating a single site within the receptive field was compared with that evoked by pairing that stimulus with one at another site outside the receptive field. In the remaining 15, the number of impulses evoked by a probe presented just inside the border was compared to that evoked when a probe twice its size was positioned so that it straddled the border with half inside and half outside the receptive field. Little evidence was observed in the majority of neurons studied with either method to indicate the presence of suppressive surrounds. Using the first method, the re-

DISCUSSION Previous studies of somatosensory neurons in the cat superior colliculus identified them as having relatively large receptive fields organized into a coarse somatotopic map (Stein et al., ’76; Nagata and Kruger, ‘79; Clemo and Stein, ’84; Meredith et al., ’90, ’91).The data suggested that such

542

H.R. CLEM0 AND B.E. STEIN

Test Site

Fixed Site A

A and Test Site

I

2

1

1

1

1 I

3

4

41

1

1

I

1

L U L L

Fig. 7. Somatosensory receptive fields generally lack surround inhibition. In this typical example, the effect of stimulating a site outside the excitatory borders of the receptive field (A) on the responses elicited from within (sites 1-3) and outside (site 4) the receptive field

was examined. Note that pairing A with another stimulation site produced no significant change in the number of impulses evoked. Scale bar in ms.

receptive fields lack precision in specifying the locus of a stimulus on the body surface. While the present results also demonstrate that these receptive fields are large compared to those in other areas of the central nervous system, the data also indicate that their capability to aid in stimulus localization may be greater than expected on the basis of receptive field size alone. The large areas of overlap that characterize the somatosensory map give rise to ambiguity regarding the specific location of a stimulus (Meredith et al., '90, '91). However, the presence within receptive fields of best areas which shift systematically (and more rapidly than do the receptive field borders) across the superior colliculus may provide further resolution to the somatotopic map. Best areas were identified in nearly all of the somatosensory receptive fields studied, and they varied with the size of the receptive field: the smaller the receptive field, the smaller the best area in absolute size, but the greater its proportion of the receptive field. Consequently, as receptive field size increased, the proportion of that field devoted to the best area decreased. Best areas also appear to

provide a better substrate for within-field spatial sensitivity than do adjacent areas of the receptive field. Increasing stimulus size within the best area resulted primarily in an enhanced response, indicative of spatial summation, although in a few cases spatial inhibition was noted. Spatial analysis features were far less robust when stimuli were presented outside the best area, and whatever trends toward spatial summation or inhibition were evident rarely reached statistical significance. Neurons representing the face and forelimb, where the small receptive fields were found, appeared to have proportionally better within-field spatial analysis capacities than neurons representing the trunk and hindquarters, and their best areas were proportionately the largest. While best areas appear to play an important role in some spatially-related properties (e.g., spatial summationiinhibition), they have little influence on others, such as direction selectivity (Clemo and Stein, '87) and surround inhibition, which are not characteristic of somatosensory neurons in the superior colliculus. However, this is not surprising

SOMATOSENSORY PROPERTIES IN SUPERIOR COLLICULUS

543

al., '81; Ogasawara, '81; Ogasawara and Kawamura, '82; Cooper and Dostrovsky, '851, and descending inputs are derived from somatosensory cortices, SIV and para-SIV in the anterior ectosylvian sulcus, and from the rostra1 suprasylvian sulcus (Stein et al., '83; Clemo and Stein, '84). Although most of these inputs have not been systematically examined in terms of their influence on superior colliculus neurons, the corticotectal inputs have been shown to be essential for normal excitability, and in some cells, for maintaining the integrity of receptive field borders (Clemo and Stein, '86a). It is possible that the heightened sensitivity of subregions of superior colliculus receptive fields reflects the areas of greatest overlap among these converging afferents.

7

ACKNOWLEDGMENTS The authors thank Dr. M.A. Meredith for his comments regarding this work, N. London for editorial assistance, and J. Nelson for histological assistance. Supported by NINCDS grant EY05554 and NSF grant BNS 8719234.

LITERATURE CITED 0

0

0

s

Fig. 8. A rare example of surround inhibition. T h e neuron described in Figure 1was examined with stimuli that bordered o n or overlapped t h e excitatory receptive field. These tests revealed inhibition from beyond the receptive field. T h e first raster-histogram pair shown here illustrates t h e responses t o a small probe (shaded circle) presented within t h e receptive field, t h e second pair represents the responses t o a larger probe (oval) t h a t includes t h e first site b u t extends beyond t h e excitatory region, a n d t h e third pair represents t h e absence of a response to stimulation j u s t outside the excitatory regions (open circle). T h e oval stimulus that overlapped the borders of t h e receptive field resulted in a decrease in t h e number of impulses (-86%) evoked, as we11 as in their frequency and reliability, as shown in t h e graph at t h e lower right. S = spontaneous activity.

given the rarity of the occurrence of these properties elsewhere in the somatosensory system. Surround inhibition has been noted in a small group of neurons in the cuneate nucleus that project to the thalamus (Janig et al., '77; Aoki, '81; Dykes et al., '82), and it becomes progressively less common, less effective, and less intense at progressively higher levels of the nervous system (Mountcastle, '57; Mountcastle and Powell, '59; Towe and Kennedy, '61; Poggio and Mountcastle, '63; Baker, '71; Baker et al., '71; Innocenti and Manzini, '72; Iwamura and Inubushi, '74; Tsumoto and Nakamura, '74; Janig et al., '79; Laskin and Spencer, '79; Gardner and Costanzo, '80a; Sur, '80). Similarly, direction selectivity among somatosensory neurons is relatively infrequent throughout the central nervous system (Mountcastle, '57; Gordon and Manson, '67; Hyvarinen and Poranen, '78; Gardner and Costanzo, '80b). It is not known at present how the internal organization of somatosensory receptive fields in the superior colliculus reflects the convergence of the numerous tectopetal afferents. Ascending inputs come from subdivisions within the dorsal column, trigeminal and lateral cervical nuclei that are distinct from those that project to the thalamus (Gordon and Jukes, '64; Baleydier and Mauguiere, '78; Blomqvist et al., '78; Edwards et al., '79; Berkley et al., '80; Huerta et

Aoki, M. (1981) Afferent inhibition on various types of cat's cuneate neurons induced by dynamic and steady tactile stimuli. Brain Res. 221257-269. Baker, M.A., C.F. Tyner, and A.L. Towe (1971) Observations on single neurons in sigmoid gyri of awake, unparalyzed cats. Exp. Neural. 32388-403. Baleydier, C., and F. Mauguiere (1978) Projections of the ascending somesthetic pathways to the cat superior colliculus visualized by the horseradish peroxidase technique. Exp. Brain Res. 31 :43-50. Berkley, K.J.,A. Blomqvist, A. Pelt, and R. Flink (1980) Differences in the collateralization of neuronal projections from the dorsal column nuclei and lateral cervical nucleus to the thalamus and tectum in the cat: An anatomical study using two different double-labeling techniques. Brain Res. 202:273-290. Blomqvist, A,, R. Flink, D. Bowsher, S. Griph, and J. Westman 11978) Tectal and thalamic projections of dorsal column and lateral cervical nuclei: A quantitative study in the cat. Brain Res. 141:335-341. Clemo, H.R., and B.E. Stein (1984) Topographic organization of somatosensory corticotectal influences in cat. J. Neurophysiol. 51:843-858. Clemo, H.R., and B.E. Stein (1986a) Effects of cooling somatosensory cortex on response properties of tactile cells in the superior colliculus. J. Neurophysiol. 55t1352-1368. Clemo, H.R., and B.E. Stein (1986b) Influence of stimulus location and size in superior colliculus somatosensory cells. Sac. Neurosci. Abstr. 12:327. Clemo, H.R., and B.E. Stein (1987) Responses to direction of stimulus movement are different for somatosensory and visual cells in cat superior colliculus. Brain Res. 405:313-319. Clemo, H.R., and B.E. Stein (1988) Examination of receptive field surrounds in somatosensory neurons in cat superior colliculus. Soc. Neurosci. Abstr. 14:125. Cooper, L.L., and J.O. Dostrovsky (1985) Projection from dorsal column nuclei to dorsal mesencephalon. J. Neurophysiol. 53,183-200. Dykes, R.W., D.D. Rasmusson, D. Sretavan, and N.B. Rehman (1982) Submodality segregation and receptive-field sequences in cuneate, gracile, and external cuneate nuclei of the cat. J. Neurophysiol. 47:389-416. Edwards, S.B., C.L. Ginsburg, C.K. Henkel, and B.E. Stein (1979) Sources of subcortical projections to the superior colliculus in the cat. J. Camp. Neurol. 184:309-330. Gardner, E.P., and R.M. Costanzo (1980a) Spatial integration of multiplepoint stimuli in primary somatosensory cortical receptive fields of alert monkeys. J. Neurophysiol. 43:420-443. Gardner, E.P., and R.M. Costanzo (1980b) Neuronal mechanisms underlying directional selectivity of somatosensory cortical neurons in awake monkeys. J. Neurophysiol. 43: 1342-1354. Gordon, B. (1973) Receptive fields in deep layers of cat superior colliculus. J. Neurophysiol. 36: 157-178. Gordon, G., and G.J. Jukes (1964) Dual organization of the exteroceptive components of the cat's gracile nucleus. J. Physiol. (Lond.) 173.263-290.

544 Gordon, G., and J.R. Manson (1967) Cutaneous receptive fields of single nerve cells in the thalamus of the cat. Nature (Lond.) 215597-599. Huerta, M.F., A. Frankfurter, and J.K. Harting (1981)The trigeminocollicular projection in the cat: Patch-like endings within the intermediate gray, Brain Res. 211:l-13. Hyvarinen, J., and A. Poranen (1978)Movement-sensitive and direction and orientation-selective cutaneous receptive fields in the hand area of the postcentral gyms in monkeys. J. Physiol. (Lond.) 283:523-537. Iggo, A. (1977) Cutaneous and subcutaneous sense organs. Br. Med. Bull. 33:97-102. Innocenti, G.M., and T. Manzini 11972) Response patterns of somatosensory cortical neurones to peripheral stimuli. An intracellular study. Arch. Ital. Biol. 110:322-347. Iwamura, Y., and S. Inubushi (1974) Regional diversity in excitatory and inhibitory receptive-field organization of cat thalamic ventrobasal neurons. J. Neurophysiol. 37:910-919. Janig, W., T. Shoultz, and W.A. Spencer (1977) Temporal and spatial parameters of excitation and afferent inhibition in cuneothalamic relay neurons. J. Neurophysiol. 40:822-835. Janig, W., W.A. Spencer, and S.G. Youngkin (1979) Spatial and temporal features of the afferent inhibition of thalamocortical relay cells. J. Neurophysiol. 42: 1450-1460. Laskin, S.E., and W.A. Spencer (1979) Cutaneous masking. 11. Geometry of excitatory and inhibitory receptive fields of single units in somatosensory cortex of the cat. J. Neurophysiol. 42:1061-1082. McHaffie, J.G., and B.E. Stein (1983)Achronic headholder minimizing facial obstructions. Brain Res. Bull. 10:859-860. McIlwain, J.T. (1975) Visual receptive fields and their images in superior colliculus of the cat. J. Neurophysiol. 38319-230. Meredith, M.A., and B.E. Stein (1990) Visuotopic component of the multisensory map in the deep layers of the cat superior colliculus. J. Neurosci. 10:3727-3742. Meredith, M.A., H.R. Clemo, and B.E. Stein (1990) Somatotopic component of the multisensory map in the deep layers of the cat superior colliculus. Soc. Neurosci. Abstr. 16:223. Meredith, M.A., H.R. Clemo, and B.E. Stein (1991) The somatotopic

H.R. CLEM0 AND B.E. STEIN component of the multisensory map in the deep laminae of the cat superior colliculus. J. Comp. Neural. 312353-370. Middlebrooks, J.C., and E.I. Knudsen (1987) Changes in external ear position modify the spatial tuning of auditory units in the cat’s superior colliculus. J. Neurophysiol. 57;672-687. Mountcastle, V.B. (1957) Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J. Neurophysiol. 20:408-434. Mountcastle, V.B., and T.P.S. Powell (1959) Neural mechanisms subserving cutaneous sensibility, with special reference to the role of afferent inhibition in sensory perception and discrimination. Bull. Johns Hopkins Hosp. 105:201-252. Nagata, T., and L. Kruger (1979) Tactile neurons of the superior colliculus of the cat: Input and physiological properties. Brain Res. 174:19-37. Ogasawara, K. (1981) Trigeminotectal projections in cats and the pathway of extraocular muscle proprioception. Neuro-optharnology 1:219-230. Ogasawara, K., and K. Kawamura (1982) Cells of origin and terminations of the trigeminotectal projections in the cat as demonstrated with the horseradish peroxidase and autoradiographic methods. Okajimas Folia Anat. Jpn. 58.247-264. Ogasawara, K., J.G. McHaEie, and B.E. Stein (1984) Two visual corticotectal systems in cat. J. Neurophysiol. 52:1226-1245. Poggio, G.F., and V.B. Mountcastle (1963) The functional properties of ventrobasal thalamic neurons studied in unanesthetized monkeys. J. Neurophysiol. 26;775-806. Stein, B.E., B. Maglahaes-Castro, and L. Kruger (1976) Relationship between visual and tactile representation in cat superior colliculus. J. Neurophysiol. 39;401419. Stein, B.E., R.F. Spencer, and S.B. Edwards (1983) Corticotectal and corticothalamic efferent projections of SIV somatosensory cortex in cat. J. Neurophysiol. 50t896-909. Sur, M. (1980) Receptive fields of neurons in areas 3b and 1of somatosensory cortex in monkeys. Brain Res. 198:465-471. Towe, A.L., and T.T. Kennedy (1961) Response of cortical neurons to variation of stimulus intensity and locus. Exp. Neurol. 3:570-587. Tsumoto, T., and S. Nakamura (1974) Inhibitory organization of the thalamic ventrobasal neurons with different peripheral representations. Exp. Brain Res. 21:195-210.