| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
The Journal of Neurophysiology Vol. 78 No. 5 November 1997, pp. 2503-2517
Copyright ©1997 by the American Physiological Society
Krieger Mind/Brain Institute, Johns Hopkins University, Baltimore, Maryland 21218
 |
ABSTRACT |
|---|
 Abstract
 Introduction
Methods
Results
Discussion
References
|
|---|
Blake, David T., Kenneth O. Johnson, and Steven S. Hsiao. Monkey cutaneous SAI and RA responses to raised and depressed scanned patterns: effects of width, height, orientation, and a raised surround. J. Neurophysiol. 78: 2503-2517, 1997. The aim of this study was to examine the slowly adapting type I (SAI) and rapidly adapting (RA) primary afferent representation of raised and depressed surface features. Isolated, raised, and depressed squares and small raised squares with a circular surround were scanned across the receptive fields of SAI and RA mechanoreceptive afferents innervating the distal fingerpads of the rhesus monkey. Pattern height ranged from 
620 to +620 µm and width ranged from 0.2 to 7.0 mm. The surround radii ranged from 3.0 to 7.0 mm. Previous combined psychophysical and neurophysiological studies have provided evidence that SAI afferent responses are responsible for the perception of spatial form and texture and that RA afferents are responsible for the detection of stimuli that produce minute skin motion (flutter, slip, microgeometric surface features). Our results strengthen these hypotheses. Response properties shared by both SAI and RA afferent types were that both responded only to the edges of the larger raised and depressed patterns, both responded to falling edges half as vigorously as to rising edges, both responded to rising and falling edges with impulse rates that were proportional to the sine of the angle between the edge and the scanning direction, and both had suppressed responses to a small raised surface feature when a raised surround was closer than 6 mm. Response differences consistent with the hypothesis that SAI afferents are specialized for the representation of form were that SAI responses were confined to areas around the features that evoked them in areas that were 40-50% smaller than the comparable RA response areas, SAI responses were more than four times more sensitive to stimulus height than were RA afferents over the range from 280 to 620 µm, and SAI (but not RA) afferents responded 20-50% more vigorously to corners than to edges. Response differences consistent with the hypothesis that RA afferents are specialized for the detection of minute surfaces features were that only RA afferents responded to very small surface depressions, depressed squares 0.8 mm wide, that were detectable by palpation. Mechanisms underlying the many differences in SAI and RA response properties are discussed.
When a finger is scanned across a surface, the surface features evoke a neural image that moves across the afferent population response in register with the features. Studies in monkeys (Gardner and Palmer 1990 Stimuli
The stimuli were fabricated from sheets of photosensitive plastic that are water-soluble until exposed to UV light (Toyoba Printight plastics, EF-series). A stimulus pattern was produced by laying a photographic negative of the pattern over the plastic sheet and exposing it to UV light. The portion of the surface layer not exposed to UV light was scrubbed off lightly in water. The height of the remaining, raised pattern was determined by the thickness of the water-soluble layer, which was 280, 370, 470, and 620 µm in this study. The sides of the patterns fell away to the background at 60° relative to the plane of the surface. Three sets of stimulus patterns were used. The first set, constructed at all four heights, consisted of seven raised squares, 0.2, 0.8, 1.5, 2.5, 4.0, 5.5, and 7.0 mm wide, which were separated by 30 mm, center to center, in the scanning direction (see Fig. 3). The second set consisted of seven depressed squares, 620 µm deep, with the same widths as the raised squares (see Fig. 10). The third set consisted of seven circular wells, 620 µm deep, 3.0-7.0 mm in radius and separated by 30 mm center to center in the scanning direction (see Fig. 13). In the center of each well was a raised square, 620 µm high and 0.4 mm wide. After exposure and washing, the patterns were trimmed to strips 20 mm wide and 240 mm long and fixed to a drum, 240 mm in circumference (Johnson and Phillips 1988
Neurophysiological experiments
Experiments were performed on barbiturate anesthetized rhesus monkeys (Macaca mulata) weighing 4-5 kg (Mountcastle et al. 1972 Analysis
The action potential times were converted to spatial coordinates by interpolating between the times of drum position signals (200 per revolution) to obtain the horizontal, X, coordinate and assigning each action potential a vertical, Y, position based on the axial position of the rotating drum at the time of the action potential. The resulting rasters (e.g., Fig. 2) are referred to as spatial event plots (SEPs). Apart from a constant, unknown error due to action potential conduction delay between the receptors and the recording site, the action potentials were located with a precision of 8 µm (Johnson and Phillips 1988
Response area and impulse rate
The method of computing the areal extent of a response to a particular pattern is a modified version of the 10% rule used by Johnson and Lamb (1981)
Neurophysiological data were collected from 12 SAI and 14 RA afferents in the median or ulnar nerves of three rhesus monkeys. All afferent fibers had receptive fields on the glabrous surface of one of the four distal finger pads. Twelve SAI and 12 RA afferents were studied using raised squares with and without circular surrounds. Seven SAI and 10 RA afferents were studied using depressed squares.
Responses to raised square surfaces
The spatial structures of the responses of the two afferent types are qualitatively similar as can be seen in Figs. 2 and 3. Both afferent types responded vigorously to the rising and falling edges of the raised stimuli. Although Fig. 2 shows an example where the SAI afferent responded to the falling edge rather weakly, it will be shown later that there was no significant difference in the ratio of mean responses to rising and falling edges between SAI and RA afferents (e.g., see also Figs. 3, 4, and 8). Both afferent types responded much less vigorously to the edges running parallel to the scanning direction, and neither responded to the flat, elevated surface. Figure 3 illustrates particularly well the interaction between the afferent fiber's receptive field and the pattern. The SAI illustrated in Fig. 3 is the same as the one illustrated in Fig. 1. The asymmetry of the receptive field accounts in part for the difference in responses at the top and bottom of the rising edge.
Effects of pattern height and width
Although SAI impulse rates were very sensitive to changes in pattern height, the RA rates were not. The qualitative effects of changes in pattern height can be seen in Fig. 2, the quantitative effects in Fig. 5. SAI impulse rates and areas were affected similarly by changes in pattern height; increasing the pattern height from 280 to 620 µm caused an average 44% increase in mean SAI impulse rate (range 36-49%) and a 38% increase in response area (range 24-55%). In contrast, RA impulse rate was affected only slightly by the same change in pattern height (10% mean increase, range 1-13%), whereas response area was affected strongly (53% mean increase, range 30-78%).
Height and width have independent effects on mean impulse rate
Analysis of the effects of pattern height and width on mean impulse rate revealed that changes in pattern height and width had completely independent effects; that is, the SAI and RA mean impulse rates, illustrated in Figs. 5 and 6, can be separated into multiplicative height and width effects with no detectable interaction. The mean rates of both SAI and RA afferents can be expressed as the product of three terms
Responses to rising and falling edges
The differences between responses to leading and falling edges were nearly identical in the SAI and RA afferents. Impulse rates at the falling edges were 75% as high as the rates at the rising edges, and the response areas at falling edges averaged 70% of the response areas at the rising edges for both SAIs and RAs (see Fig. 3). Since the total impulse count equals the product of the mean rate and the response area, a reduction in total impulse count to 52% (75% times 70%) of the count at the leading edge is predicted. In fact, the trailing impulse counts for SAI and RA afferents averaged 53 and 54% of the leading counts, respectively. Individual action potential counts at the rising and falling edges of the raised square, 7.0 mm wide, are shown in Fig. 8.
Responses to corners
SAI afferents responded more vigorously to the corners than to the central sections of the rising and falling edges, producing impulse rates near the corners that were 20-50% greater than at the middle sections of the rising and falling edges. The RA afferent responses were less sensitive to corners; their impulse rates near the corners were, on average, within 10% of the rates at the middle sections of the rising and falling edges. Profiles of the mean SAI and RA afferent impulse rates in a cross-sectional view of firing rates parallel to the rising and falling edges of the 7.0 mm squares, 280 and 620 µm high, are shown in Fig. 9. There was a 10% elevation in mean RA firing rates at the leading corners of the 620 µm high square (Fig. 9, top right) but also a small drop in mean firing rates at the trailing corners. The mean response to the leading edge of the 280 µm square (Fig. 9, bottom right) was flat to the corners, whereas the trailing responses were depressed at the corners. The RA corner responses at intermediate heights were similar to these.
Receptive field areas
Receptive field areas have been measured frequently with punctate stimuli (Johnson and Lamb 1981 Responses to depressed squares
The second stimulus set, illustrated along with the responses of typical SAI and RA afferents in Fig. 10, consisted of depressed squares that varied in width. The responses of SAI and RA afferents to the wider depressed squares were similar to their responses to the wider raised squares (cf. Figs. 3 and 10, also Figs. 4 and 12). The afferents responded primarily to the leading and trailing edges, producing population responses that outlined the depressed patterns. However, the quantitative details were quite different. The responses evoked by the depressed patterns were less intense, and the more intense response occurred at the trailing rather than the leading edge. Another major difference in the responses of SAIs to the two pattern types lay in their responses to the corners. In the responses to depressed squares, the discharge rates at the corners were decreased relative to the rates at the leading and trailing edges. Also, both SAI and RA afferents responded more vigorously to the edges that run parallel with the scanning direction than to the similar edges in the raised patterns. Other differences, evident in Figs. 3 and 10, are that RA afferents and to a lesser degree SAI afferents tended to fire in the middle of the depressed patterns, which increased the pattern width required for complete separation of the leading and trailing responses.
Effects of a raised surround
The third stimulus set, illustrated in Fig. 13, comprised 0.4 mm wide squares, 620 µm high, surrounded by plateaus, also 620 µm high, at distances of 3.0-7.0 mm from the center of the raised square. Typical SAI and RA responses are displayed in Fig. 13. The responses to the circular depressions were like the responses to square depressions; both afferent types responded to the trailing, rising edge more vigorously than the leading, falling edge, and the RAs tended to respond in the center of the depression whereas the SAIs did not. Also, both SAI and RA afferents, but particularly SAI afferents, provided a clear neural image of the entire depressed pattern, including the edges parallel to the scanning direction.
Effects of edge orientation
The responses evoked by the leading and trailing edges of the raised surrounds illustrated in Fig. 13 clearly are related to the orientations of the edges relative to the scanning direction. This effect was analyzed by removing the response to the central, small square and counting action potentials in 30° segments around the response to the rising and falling edges. The result is shown in Fig. 15 together with a rectified sine wave with a constant offset. The result suggests that both SAI and RA responses are proportional to the sine of the angle between the edge and the scanning direction. The main lack of fit, which applies more to the RA than the SAI responses, is a reduction of the measured responses at 90 and 270° relative to the predicted responses. The reason appears to be that during the few scans where the small raised squares pass over the center of the receptive field, the RA response is depressed. Otherwise, the responses to the trailing, rising edge (90°) would have been larger than to the leading, falling edge (270°) as in the responses to the depressed squares illustrated in Fig. 10.
The responses of SAI and RA afferents to raised and depressed patterns were studied. Many of the details of the responses of the SAI and RA afferents to raised and depressed patterns were similar. They show that both afferent types responded strongly to the rising and falling edges of both raised and depressed patterns and neither responded to the flat central portions of the scanned, raised patterns. Both responded to edges running parallel to the scanning direction much less vigorously than to edges that are orthogonal to the scanning direction. The responses to rising and falling edges were approximated closely by the sine of their orientation relative to the scanning direction. The result of these response properties is an intense neural image in both the SAI and RA population responses that tracks the stimulus pattern and highlights spatial details.
Rising and falling edges
The study reported here complements one by LaMotte and Srinivasan (1987a Corner responses
A striking finding is that SAI afferents are even more sensitive to corners than edges. Previous studies (Goodwin et al. 1995 Effects of pattern height on impulse rate and spread of activity
The rapid growth in RA population discharge rate with increasing pattern height was due almost entirely to changes in RA response area; mean RA impulse rate was almost constant for pattern heights ranging from 280 to 620 µm. The rapid growth of SAI population discharge with pattern height was, in contrast, more dependent on the growth in SAI impulse rates than in response area.
Depressed patterns
SAI and RA responses to depressed patterns were, in some respects, like their responses to raised patterns. Both afferent types responded vigorously to the rising and falling edges of the wider depressed patterns and both, but especially the SAI afferents, represented the boundaries clearly. In some cases, the overall impulses rates were very similar to the overall rates evoked by the raised squares. For example, the raised square, 7 mm wide and 280 µm high, evoked SAI and RA responses whose total rates and response areas were almost identical to those evoked by the depressed square, 7 mm wide and 620 µm deep.
Circular surround
The SAI and RA responses to a small raised square with a circular surround were consistent with their responses to the raised and depressed squares. When the surroundwas >6 mm distant, the SAI and RA responses to the small raised square were similar to their responses to the same isolated raised square. At the smaller surround radii where skin conformation to the depression rather than the raised central element became the determining factor, the RA afferents were more responsive than the SAIs. The distance where the surround first began to affect the response to the central element, ~6.0 mm, is similar to the center-to-center distance where responses to single dots first began to decline in a study of the responses of human primary afferents to an array of equally spaced dots (Phillips et al. 1992
INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References
; Johnson and Lamb 1981) and humans (Phillips et al. 1990, 1992) show that neither slowly adapting type II (SAII) nor Pacinian afferents provide a signal that could account for the human ability to discriminate Braille characters (Loomis 1981; Stevens et al. 1996) or raised letters (Vega-Bermudez et al. 1991). It is clear that surface feature discrimination depends on one or both of the neural images conveyed by the SAI or rapidly adapting (RA) afferent populations. An intriguing hypothesis that, if true, would have implications for the CNS mechanisms underlying the processing of information from the SAI and RA afferents is that the SAI system is specialized for form processing and the RA system for motion processing (Johnson and Hsiao 1992). Investigations concerned with the relative roles of the two afferent fiber types in surface feature perception depend on quantitative hypothesis testing. There have been many quantitative studies of the neural responses to textured surfaces (i.e., repetitive or quasirepetitive stimuli) but fewer quantitative studies of the SAI and RA responses to isolated surface form (Goodwin et al. 1995; Johnson and Lamb 1981; LaMotte and Srinivasan 1987a,b; LaMotte and Whitehouse 1986; LaMotte et al. 1996).
). In a study similar to the one reported here, LaMotte and Srinivasan (1987a,b) scanned rising and falling edges over the receptive fields of SAI and RA afferents and varied scanning velocity and edge slope while they held the height, width, and force constant. In this study, we vary the height and width of square patterns while holding the scanning velocity, edge slope, and force constant. We also study the effect of a raised surround on the responses to a raised pattern. The squares range from 0.2 to 7.0 mm in width and from 620 to +620 µm in height. Both SAI and RA afferents respond vigorously to leading and trailing edges of both raised and depressed patterns. SAI afferent discharge rates are sensitive to pattern height, width, and corners whereas RA discharge rates are relatively insensitive to all three pattern components. Both afferent types are sensitive to the orientation of an edge; each responds with a discharge rate proportional to the sine of the angle between the scanning direction and the edge.
METHODS
Abstract
Introduction
Methods
Results
Discussion
References
).
View larger version (19K):
[in a new window]
FIG. 3.
SAI and RA responses to raised squares of varying width. Patterns were raised 470 µm above the background and were separated from one another in the scanning direction by 30 mm center to center. A: stimuli. B: SEPs for a typical SAI and RA afferent and the mean rate profile over all afferents of a single type (SAI or RA) for the sweep where the center of each square passed over the afferent's point of maximum sensitivity.
View larger version (16K):
[in a new window]
FIG. 10.
SAI and RA responses to depressed squares of varying width. A: white areas represent depressed regions (620 µm deep). B: SEPs for a typical SAI and RA afferent and the mean rate profile over all afferents of a single type (SAI or RA) for the sweep where the center of each square passed over the afferent's point of maximum sensitivity.
View larger version (30K):
[in a new window]
FIG. 13.
SAI and RA responses to a raised square with a raised circular surround. Squares, 0.4 mm wide and 620 µm high, were surrounded by raised plateaus, also 620 µm high. A: black areas represent raised parts of the pattern. B: SEPs for typical SAI and RA afferents and the mean rate profile over all afferents of each type for the sweep where the center of each square passed over the afferent's point of maximum sensitivity.
). Single cutaneous mechanoreceptive fibers were dissected from the median or ulnar nerves using standard methods (Mountcastle et al. 1972). Afferents were classified as SAI, RA, or Pacinian (PC) on the basis of responses to indentation and vibration with a point probe (Talbot et al. 1968). Only SAI and RA afferent fibers with receptive fields on the distal glabrous pads of digits two to five were studied. After mapping the receptive field with von Frey monofilaments, the finger was positioned so that the point of maximum sensitivity was located at the center of the region of contact between the skin and the stimulus. Neurons were not studied when the finger could not be positioned appropriately because, for example, the receptive field was on the side of the finger or too close to the nail. The patterns always were scanned from proximal to distal. The stimulus pattern was presented to the skin (see below) with a contact pressure and scanning velocity chosen to match the motion and subsequent skin distortions that typically occur when human subjects palpate a patterned surface (Johnson and Lamb 1981; Vega-Bermudez et al. 1991).
). The contact force was controlled by a servo-controlled torque motor whose shaft was the fulcrum of a balance beam with the drum and drive motor at one end and counter weights on the other. The contact surface between the drum and the monkey's finger was an oval, 7 mm wide by 9 mm long or more. After each full rotation of the drum containing the stimulus pattern, the drum was shifted 0.2 mm in the axial direction (at right angles to the direction of rotation). This sequence was repeated 
75 times, producing a total transverse of 15.0 mm in the direction orthogonal to rotation. The occurrence times of action potentials and drum position signals were recorded with a precision of 0.1 ms.
). Then the SEPs were converted into two-dimensional firing rate arrays of 0.2 × 0.2 mm bins with a two-dimensional, adaptive Parzen estimator (Twombly et al. 1996), which replaces each impulse with a two-dimensional Gaussian function with unit volume. The ratio and orientation of the Gaussian major and minor axes are determined by the covariance of the locations of all impulses within 1.0 mm of the target impulse in the SEP. The spread (standard deviation of the major axis) is scaled so that it is inversely proportional to the square root of the number of impulses within this 1.0 mm radius. Thus an impulse in a region of high firing rate is represented by a tall, narrow Gaussian, whereas an isolated impulse is represented by the widest allowable Gaussian function, which, in this application, was a circular distribution with a standard deviation equal to 0.63 mm in all directions. Then the Gaussian volume overlying each 0.2 × 0.2 mm bin was calculated to generate a two-dimensional array of firing rates. This binning method was used because extensive analyses with simulated spike trains show that this method generates impulse rate estimates with lower standard errors than conventional bucket binning or fractional interval binning. These two-dimensional arrays of firing rate were used to compute response areas and mean firing rates.
View larger version (28K):
[in a new window]
FIG. 2.
Typical SAI and rapidly adapting (RA) responses to the narrowest and widest raised squares at each of the 4 pattern heights used in the study. A: narrowest pattern, which is barely visible, was 0.2 × 0.2 mm, the widest was 7.0 × 7.0 mm. B: SEPs for typical SAI and RA afferents (display conventions as in Fig. 1).
) and solved using multiplicative or additive regression. All regression analyses and tests of significance were done with SPSS for Windows, Version 7.0.
and Phillips et al. (1992). The original method included all histogram bins where impulse rates evoked by a particular pattern exceeded a threshold defined as 10% of the peak impulse rate evoked by that pattern. In this study, we modified the method of estimating this threshold because the peak rate in 0.2 × 0.2 mm bins can be quite variable. The new method yields the same threshold on average but is less variable. A new estimate of the peak rate was obtained by searching and finding the circular region, 1 mm in diameter, within each response region that contained the maximum number of impulses and, therefore, the maximum mean impulse rate. The coefficient of variation of the new peak rate measure was <0.1 when applied to repeated scans of the same pattern. Because this rate was, on average, 40% of the peak rate measured in single bins (i.e., by the old method) the threshold for measuring response area was set to 1/4 of this value. These thresholds were, on average, 10% of the peak rate within the small, 0.2 × 0.2 mm bins, but were less variable than the thresholds based on the original method.
) for each afferent fiber type.
View larger version (12K):
[in a new window]
FIG. 1.
Typical slowly adapting type I (SAI) response to a single raised square, 0.47 mm high and 0.2 mm wide, scanned repeatedly over its receptive field. Methods of stimulus presentation and the display conventions are the same in all figures in this paper. Stimuli were scanned from proximal to distal on the distal pad of 1 of the digits at 30 g force and 40 mm/s. Patterns were shifted 0.2 mm orthogonal to the scanning direction between scans. A: spatial event plot (SEP). Action potential data at the tops of SEPs came earlier in the stimulus sequence than those lower in the raster. Similarly, the scanning direction in the displays is from left to right; thus ticks at the left of the plot represent action potentials evoked by the leading edge of the stimulus and ticks to the right, the trailing edge. B: instantaneous impulse rate in 0.2 × 0.2 mm bins. Gray level is proportional to firing rate; the peak rate, represented by black squares, was ~400 impulses/s (ips) in this example. C: response area is measured by counting 0.2 × 0.2 mm bins where the impulse rate exceeds 10% of the peak rate (see METHODS).
RESULTS
Abstract
Introduction
Methods
Results
Discussion
References
View larger version (47K):
[in a new window]
FIG. 4.
Averaged responses of all SAI and RA afferents to squares 0.2 and 5.5 mm wide and 470 µm high. Darker regions of the gray-scale represent regions of increased impulse density. Before averaging, the individual discharge patterns were aligned so that their peak impulse rates fell over the leading edge of the stimulus pattern. Location of the peak rate for each afferent was identified as the rectangular area with a length of 1 mm and width equal to the pattern width that contained the largest spike count. Peak mean SAI impulse rates were 260 and 190 ips for the squares 0.2 and 5.5 mm wide. Comparable RA rates were 190 and 145 ips. Vertical stripes in the 3-D profiles resulted from the regularity of firing in both afferent types.
View larger version (17K):
[in a new window]
FIG. 8.
Leading and trailing responses to raised squares 7.0 mm wide in 12 SAI and 12 RA afferents. Abscissa and ordinate in each graph represent the total number of action potentials evoked by the rising and falling edges summed over all scans. +, sample means.
). Averaged responses of all SAI and RA afferents to a narrow and wide square are illustrated in Fig. 4.
View larger version (20K):
[in a new window]
FIG. 5.
Mean response area, mean impulse rate, and population impulse rate versus pattern height for raised squares 0.2, 4.0, and 7.0 mm wide. Other widths produced responses intermediate between those shown here. Response areas were measured as in Fig. 1 (see METHODS). Mean response rate for each afferent was measured as the total number of impulses within the response area divided by the total scanning time within that area. Normalized mean impulse rates for individual afferent responses to the raised squares, 4.0 mm wide, are shown to illustrate the variation between afferents. Normalization for each afferent fiber was achieved by dividing the mean response rate evoked by each square by the fiber's mean rate for all four 4.0 mm wide raised squares. Population rate was calculated as the product of the mean response area, mean impulse rate, and innervation density (1.2 afferents/mm2 for SAI, 1.5 for RA; (Darian-Smith and Kenins 1980 ).
View larger version (25K):
[in a new window]
FIG. 6.
Mean response area, mean impulse rate, and population impulse rate versus square width. Format is the same as in Fig. 5.
one for the rate at one combination of height and width (the reference stimulus), one for the effect of changing height, and one for the effect of changing width:
The smallest raised square, 0.28 mm high and 0.2 mm wide, that evoked mean impulses rates of 155.2 and 113.4 ips in SAI and RA afferents, respectively, was adopted as the reference stimulus. The independent height and width effects are listed in Table 1 and displayed in Fig. 7.
View this table:
TABLE 1.
Effects of height and width on impulse rate
View larger version (13K):
[in a new window]
FIG. 7.
Height and width effects on mean impulse rate. Pattern height and width had independent, multiplicative effects on the mean impulse rate (see text). Points in the left graph represent the normalized mean impulse rates evoked by raised squares, 0.2 mm wide (±SE). Points in the right graph represent the effect of pattern width on the mean impulse rate. Mean impulse rate evoked by any combination of pattern width and height is given by the product of the height and width effects. The dashed lines represent the model values.
View larger version (29K):
[in a new window]
FIG. 9.
Leading and trailing peak edge responses to raised squares 7.0 mm wide. Profiles illustrated here were obtained by selecting a vertical slice 1 mm wide around the peak leading and trailing rates in each SEP and averaging the rates at each vertical location. Top: responses shown to squares 620 µm in height. Bottom: responses shown to squares 280 µm in height. Responses to the squares 370 and 470 µm high were intermediate between those illustrated here and were omitted for clarity.
View this table:
TABLE 2.
SAI and RA functional receptive field diameters at the rising edge
; Phillips et al. 1992; Talbot et al. 1968) but not with complex stimuli of the kind used here. A complication associated with the analysis of cutaneous mechanoreceptors is that receptive field size depends on the stimulus and its intensity. The data presented so far show that RA receptive fields are larger than SAI fields, that both depend on stimulus height, and that both are smaller at falling than at leading edges (by 30%). Because SAI and RA impulse rates evoked by the edges parallel to the scanning direction were <10% of the peak rate, the response area was determined exclusively by the responses to the leading and trailing edges. Consequently, we modeled the response areas as convolutions of a circular receptive field with the leading and trailing edges (see APPENDIX). The field diameter was adjusted to fit the data at each height and was reduced by 30% at the trailing edge to account for the observation that the spread of activity at the trailing edge was 30% less than at the leading edge. Table 2 shows the receptive field diameters and areas that account most effectively for the observed response areas. The modeling results show that response area is predicted accurately by such an analysis, that both SAI and RA receptive field areas depend strongly on pattern height, and that RA receptive field areas are about twice as large as SAI field areas. Analysis of the responses to the smaller raised squares showed that when the response areas produced by the leading and trailing edges overlap, there is only partial (50%) summation of the two response areas, which accounts for the nonlinearity at the lower left of the curves relating response area to pattern width in Fig. 6.
View larger version (48K):
[in a new window]
FIG. 12.
Averaged responses of all SAI and RA afferents to the depressed, 7.0 mm wide square. Display format is the same as in Fig. 4.
View larger version (18K):
[in a new window]
FIG. 11.
SAI and RA mean response area, impulse rate, and population impulse rate evoked by depressed squares of varying width. Format is the same as in Fig. 5.
View larger version (25K):
[in a new window]
FIG. 14.
Impulse rate vs. distance to a raised circular surround (see Fig. 13). Abscissa in each graph represents the radius of the raised surround. Top: ordinates represent the mean impulse rate measured in a circle, 2 mm in diameter, centered over the response to the raised square. Bottom: ordinates represent the response rates of individual neurons normalized with respect to the mean response to all the patterns with a surround.
View larger version (20K):
[in a new window]
FIG. 15.
Effect of edge orientation. Neural responses to the perimeter of the largest (7.0 mm radius) raised surround illustrated in Fig. 13 were divided into 30° segments for the analysis of impulse rates. Ninety degrees on the abscissa of each graph represents a response zone 30 ± 15° wide at the center of the trailing, rising edge; 270° represents a response zone 30° wide at the center of the leading, falling edge. Other angles are measured counterclockwise from those points. Ordinate represents the mean, normalized, total discharge in each 30° segment, which was computed as follows: first, the total number of impulses in each 30° segment were counted after excluding impulses closer than 3.5 mm to the center of the circular pattern, which were presumed to be related to the small raised square in the center of the pattern. Then the counts for each neuron were normalized to a mean of 1.0, since the object of the analysis was the form of the relationship between edge orientation and firing rate rather than the absolute rate. The solid line represents the mean of these normalized curves (±1 SE). The dashed line represents a constant plus the absolute value of the sine of the angle between the edge and the scanning direction, all scaled to fit the data.
DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References
). The most striking difference was the effect of pattern height. The primary effect of increasing pattern height on the SAI neural image was to intensify it by increasing the firing rates of individual afferents. The effect on the RA neural image mainly was to increase the spread of neural activity, thus diminishing its acuity; changing pattern height had a minimal effect on the firing rates of individual RA afferents over the range of pattern heights used in this study. A second difference, consistent with the greater SAI sensitivity to spatial detail, was their sensitivity to raised corners. As a result, the SAI neural image highlights corners even more than edges.
), one for each plane passing through the point. Once the plane is defined, curvature at a point is defined by the circle best fitting the line of intersection between the surface and the plane (more precisely, it is the circle with the same rate of change of direction at that point). The center of curvature is the center of the circle, and curvature is defined as the reciprocal of the radius. We will refer to curvature in the plane of the skin as horizontal curvature and curvature in planes orthogonal to the skin as vertical curvature. We will refer to stimulus curvature whose center lies toward the interior of the stimulus as stimulus convexity and skin curvature whose center lies beneath the skin as skin convexity. Also, when a surface contacts the skin, two profiles of surface curvature are importantthe profile of the stimulus where it contacts the skin, which we will call the contact profile, and the skin profile where it leaves the stimulus, which we will call the free skin profile. In regions of contact, the skin profile and the contact profile are, of course, the same. In regions where the skin is free of the stimulus, the stimulus profile is irrelevant but not the skin profile. The total skin profile is the contact profile plus the free profile.
,b), who scanned rising and falling edges over the receptive fields of SAI and RA afferents. Both studies controlled force rather than displacement and used similar contact forces (20 and 30 g). LaMotte and Srinivasan varied scanning velocity and edge slope, whereas we varied pattern height and width. Where our stimulus conditions and theirs overlap, the results are nearly identical. The only significant difference is that they observed low SAI firing rates on completely flat surfaces, and we did not. This difference is likely explained by the longer interstimululs interval in their study relative to ours (2 vs. 0.175 s).
; Pubols 1980; Talbot et al. 1968). Conversely, sensitivity to edges accounts well for the robust SAI responses but not the RA responses (LaMotte and Srinivasan 1987b; Phillips and Johnson 1981a). However, one study that used vibratory stimuli found that RAs respond more vigorously near an edge than under a flat surface but that the ratio was less than for SAI afferents (Johansson et al. 1982). All these facts suggest that the RA responses to rising and falling edges are due mainly to the dynamic vertical displacement produced by edges rather than the local curvature, whereas the SAI responses are due mainly to the skin curvature produced by edges.
). That must have been a factor, however, if that was the main cause the impulse rates evoked by the trailing, rising edges of the depressed patterns would have been lower than those evoked by the leading, falling edges, but they were not (cf. Figs. 4 and 12). Moreover, although LaMotte and Srinivasan (1987a) allowed 2 s between stimulus presentations, their results and ours are nearly identical.
; Pubols 1980; Talbot et al. 1968) and of the sensitivity of SAI responses to edges relative to flat surfaces (Phillips and Johnson 1981a) have noted similar wide variation in response properties.
; Johansson et al. 1982; LaMotte et al. 1996; Phillips and Johnson 1981b; Srinivasan and LaMotte 1987; Vierck 1979) have shown that SAI afferents are sensitive to convex, vertical stimulus curvature. This study shows that SAI afferents are also sensitive to horizontal stimulus curvature. Taken together, the results suggest that SAI afferents are sensitive to stimulus curvature in all directions.
20 times more sensitive to dynamic vertical indentation at an edge or region of vertical stimulus curvature than at the center of a flat stimulus, whereas RA afferents are relatively insensitive to displacement near an edge (Goodwin et al. 1995; Phillips and Johnson 1981a; Srinivasan and LaMotte 1987; Vierck 1979). The difference can be accounted for by differences in transduction mechanisms. Two studies (Phillips and Johnson 1981b; Srinivasan and LaMotte 1991) have shown that SAI impulse rates are correlated highly with strain components due to shear stresses that occur at regions of high curvature. Those studies also have shown that RA impulse rates are correlated with a different strain component that is sensitive to local vertical deformation but is less sensitive to the stresses produced by curvature.
) and strain energy density (Srinivasan and LaMotte 1991) and that this accounts for the changes in SAI impulse rates at exterior and interior corners.
) have shown that RA afferents respond to ramps of constant velocity with impulse rates that are sensitive to velocity but independent of the final depth of indentation. Thus if the skin rises and falls at a constant angle around a raised pattern with steep sides, the expected effect is a constant discharge rate that is independent of pattern height. That is consistent with LaMotte and Srinivasan's data (1987b), which showed that the RA burst rate at the rising edge of their stimulus was predicted closely by the product of the edge slope and scanning velocity until the edge slope exceeded 25-30°. Steeper edge slopes produced only small increases in firing rates, suggesting that the skin was not conforming to those surfaces but was rising free of the surface at a constant angle near 25-30°.
). Consequently, the prediction of the hypothesis that the skin rose and fell at a constant angle would be that SAI mean impulse rates would contain two componentsone constant and one proportional to pattern height. In fact, that description fits the relationship between mean SAI impulse rate and pattern height well (see Fig. 5).
; Phillips and Johnson 1981a; Srinivasan 1989) but would exhibit some curvature that would translate to increasing deformation velocity with increasing pattern height. In fact, the RA response rate did rise modestly with increasing pattern height.
; Srinivasan et al. 1990).
), which resist rapid movements of the entire assembly, the assembly would produce whatever force was required to drive the skin down and around the advancing raised pattern. The force driving the skin into the depressed pattern is, by contrast, only the skins reaction force to the flat surface, which was ~0.6 g/cm2 (30 g force distributed over 50-60 square mm of skin). These differences account for the generally lower impulse rates evoked by the depressed patterns, but they also account for the higher impulse rates along the depressed edges running parallel to the scanning direction.
). The reason is that once sufficient force is exerted to force the skin around the raised object and onto the flat background, additional force causes little additional deformation around the raised object. Additional force clearly causes increased stress (pressure) but not strain (deformation) because, in incompressible tissues, strain depends on stress gradients not stress per se. Thus the responses of mechanoreceptive afferents are sensitive to the height of a raised, rigid object but are much less sensitive to the application force (Johnson and Lamb 1981). A similar consideration of the interaction between the skin and a depressed pattern suggests a complete reversal of the roles of force and displacement. When the pattern is depressed, the only force driving the skin into the depression is the pressure in the subcutaneous tissues, which is proportional to the application force. Thus the skin deformation and the neural response evoked by a depressed pattern is almost certainly sensitive to the force of application but not the depth of the pattern unless that depth is slight and the skin is supported by the bottom of the depression.
).
| |
ACKNOWLEDGEMENTS |
|---|
We thank S. Patterson, J. Lane, and the reviewers who made many helpful suggestions.
The work reported in this paper was supported by National Institute of Neurological Disorders and Stroke Grants NS-18787 and NS-34086 and the W. M. Keck Foundation.
| |
APPENDIX |
|---|
The data (see Figs. 4 and 6) suggest that SAI and RA afferents have receptive fields with functional radii that depend on the stimulus height and edge orientation. The functional receptive field radius is defined here as the maximum distance at which a particular stimulus evokes a response. The data show that the total activity at the trailing edge is less than at the leading edge and that it evokes activity over a distance that is 70% of the distance at the leading edge. So, we assumed that the functional radius at the trailing edge was 70% of the function radius at the leading edge. We ignored the responses to the edges running parallel to the scanning direction because they contributed <1% of the total discharge for SAI and RA responses to the squares 5.5 and 7.0 mm wide. These assumptions lead to the response model illustrated in Fig. A1. For squares with widths >4.0 mm where there is no overlap between the leading and trailing response areas, this leads to the prediction that total area = leading area + trailing area; leading area = 2·r·w + 
·r2; and trailing area = 2·(0.7·r)·w + ·(0.7·r)2 where w represents the stimulus pattern width (0.2-7.0 mm) and r represents the functional radius of the receptive field, which varies with stimulus height. This results in a linear relationship between pattern width and total response area for any single stimulus height, which fits the observed response areas (Fig. 6) closely for pattern widths of 4.0 mm. However, at smaller pattern widths, this formulation counts the overlapping leading and trailing response areas twice; this results in an overestimate of the total response area. If the overlapping area is counted only once (complete occlusion), the formulation underestimates the observed response areas. The over- and underestimates produced by assumptions of no occlusion and complete occlusion were almost exactly the same, suggesting that there was ~50% occlusion. So the model accounting for response area was formulated as total area = leading area + trailing area k·overlapping area where k represents the degree of occlusion. The two variables, r and k, were adjusted by nonlinear regression (SPSS) to fit the relationship between response area and pattern width at each pattern height. Since the occlusion parameter was not significantly different from 0.5 in any of the regressions (at 4 pattern heights for each of the afferent types), the occlusion parameter was fixed at 0.5. Since the corners evoke different response rates than the leading and trailing edges, at least in the SAI afferents, they are clearly different stimuli and may require different functional radii. However, when the response radius associated with the corners was treated as a separate variable, it did not improve the fit significantly and was never significantly different from the functional radius for the leading and trailing edges. So the model was reduced to one with a single variable at each pattern height: the functional radius of the receptive field.
|
The estimated receptive field diameters (twice the estimated radii), their standard errors, the correlation between predicted and observed areas, and the standard deviations of the differences between predicted and observed areas are listed in Table 2. The overall standard errors of prediction for this model with just one estimated parameter at each height rose insignificantly from 0.427 to 0.448 mm2 for the SAI response areas and from 1.084 to 1.107 mm2 for the RA response areas relative to the predictions with separate parameters for corner radii and occlusion.
| |
FOOTNOTES |
|---|
Present address of D. T. Blake: Keck Center for Integrative Neuroscience, University of California, San Francisco, 513 Parnassus Ave. S-877, San Francisco, CA 94143-0732.
Address for reprint requests: K. Johnson, 338 Krieger Hall, Johns Hopkins University, 3400 N. Charles St., Baltimore, MD, 21218 Email: kenneth.johnson{at}jhu.edu
Received 19 December 1996; accepted in final form 24 July 1997.
| |
REFERENCES |
|---|
|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Jehoel, P. T. Sowden, S. Ungar, and A. Sterr Tactile Elevation Perception in Blind and Sighted Participants and Its Implications for Tactile Map Creation Human Factors: The Journal of the Human Factors and Ergonomics Society, April 1, 2009; 51(2): 208 - 223. [Abstract] [PDF]
|
|
|
|
 |
|
 L. J. Berryman, J. M. Yau, and S. S. Hsiao Representation of Object Size in the Somatosensory System J Neurophysiol, July 1, 2006; 96(1): 27 - 39. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. C. Leiser and K. A. Moxon Relationship Between Physiological Response Type (RA and SA) and Vibrissal Receptive Field of Neurons Within the Rat Trigeminal Ganglion J Neurophysiol, May 1, 2006; 95(5): 3129 - 3145. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. T. Blake and M. M. Merzenich Changes of AI Receptive Fields With Sound Density J Neurophysiol, December 1, 2002; 88(6): 3409 - 3420. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. M. Friedman, P. S. Khalsa, K. W. Greenquist, and R. H. LaMotte Neural Coding of the Location and Direction of a Moving Object by a Spatially Distributed Population of Mechanoreceptors J. Neurosci., November 1, 2002; 22(21): 9556 - 9566. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. O. Johnson, S. S. Hsiao, and T. Yoshioka Book Review: Neural Coding and the Basic Law of Psychophysics Neuroscientist, April 1, 2002; 8(2): 111 - 121. [Abstract] [PDF]
|
|
|
|
 |
|
 P A McNulty and V G Macefield Modulation of ongoing EMG by different classes of low-threshold mechanoreceptors in the human hand J. Physiol., December 15, 2001; 537(3): 1021 - 1032. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. E. Wheat and A. W. Goodwin Tactile Discrimination of Edge Shape: Limits on Spatial Resolution Imposed by Parameters of the Peripheral Neural Population J. Neurosci., October 1, 2001; 21(19): 7751 - 7763. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Pare, R. Elde, J. E. Mazurkiewicz, A. M. Smith, and F. L. Rice The Meissner Corpuscle Revised: A Multiafferented Mechanoreceptor with Nociceptor Immunochemical Properties J. Neurosci., September 15, 2001; 21(18): 7236 - 7246. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. W. Bisley, A. W. Goodwin, and H. E. Wheat Slowly Adapting Type I Afferents From the Sides and End of the Finger Respond to Stimuli on the Center of the Fingerpad J Neurophysiol, July 1, 2000; 84(1): 57 - 64. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Vega-Bermudez and K. O. Johnson SA1 and RA Receptive Fields, Response Variability, and Population Responses Mapped with a Probe Array J Neurophysiol, June 1, 1999; 81(6): 2701 - 2710. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Vega-Bermudez and K. O. Johnson Surround Suppression in the Responses of Primate SA1 and RA Mechanoreceptive Afferents Mapped with a Probe Array J Neurophysiol, June 1, 1999; 81(6): 2711 - 2719. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. DiCarlo and K. O. Johnson Velocity Invariance of Receptive Field Structure in Somatosensory Cortical Area 3b of the Alert Monkey J. Neurosci., January 1, 1999; 19(1): 401 - 419. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. H. Lamotte, R. M. Friedman, C. Lu, P. S. Khalsa, and M. A. Srinivasan Raised Object on a Planar Surface Stroked Across the Fingerpad: Responses of Cutaneous Mechanoreceptors to Shape and Orientation J Neurophysiol, November 1, 1998; 80(5): 2446 - 2466. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. DiCarlo, K. O. Johnson, and S. S. Hsiao Structure of Receptive Fields in Area 3b of Primary Somatosensory Cortex in the Alert Monkey J. Neurosci., April 1, 1998; 18(7): 2626 - 2645. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
