1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37

Title: The firing characteristics of foot sole cutaneous mechanoreceptor afferents in
response to vibration stimuli
Authors: Nicholas D.J. Strzalkowski1, R. Ayesha Ali2, Leah R. Bent3
Affiliations: 1University of Calgary, Department of Clinical Neuroscience, Calgary AB,
Canada, 2University of Guelph, Department of Mathematics and Statistics, Guelph ON,
Canada, 3University of Guelph, Department of Human Health and Nutritional Science,
Guelph ON, Canada
Running head: Vibration response of foot sole cutaneous afferents
Corresponding author:
Dr. Leah R. Bent
Associate Professor
Department of Human Health and Nutritional Science
Guelph, Ontario, N1G 2W1
E-mail: lbent@uoguelph.ca
Phone: 519 824 4129 ext. 56442

38

Abstract

39

Single unit microneurography was used to record the firing characteristics of the

40

four classes of foot sole cutaneous afferents (fast and slowly adapting type I and II; FAI,

41

FAII, SAI, SAII) in response to sinusoidal vibratory stimuli. Frequency (3-250Hz) and

42

amplitude (0.001-2mm) combinations were applied to afferent receptive fields through a

43

6mm diameter probe. The impulses per cycle, defined as the number of action potentials

44

evoked per vibration sine wave, were measured over one second of vibration at each

45

frequency-amplitude combination tested. Afferent entrainment threshold (lowest amplitude

46

at which an afferent could entrain 1:1 to the vibration frequency) and afferent firing

47

threshold (minimum amplitude for which impulses per cycle was greater than zero) were

48

then obtained for each frequency. Increases in vibration frequency are generally associated

49

with decreases in expected impulses per cycle (p < 0.001), but each foot sole afferent class

50

appears uniquely tuned to vibration stimuli. FAII afferents tended to have the lowest

51

entrainment and firing thresholds (p < 0.001 for both); however, these afferents seem to be

52

sensitive across frequency. In contrast to FAII afferents, SAI and SAII afferents tended to

53

demonstrate optimal entrainment to frequencies below 20Hz, and FAI afferents faithfully

54

encoded frequencies between 8-60Hz. Contrary to the selective activation of distinct

55

afferent classes in the hand, application of class specific frequencies in the foot sole is

56

confounded due to the high sensitivity of FAII afferents. These findings may aid in the

57

development of sensorimotor control models or the design of balance enhancement

58

interventions.

59

Key words: cutaneous afferents, mechanoreceptor, foot sole, vibration, microneurography

60

2

61

New & Noteworthy

62

Our work provides a mechanistic look at the capacity of foot sole cutaneous afferents to

63

respond to vibration of varying frequency and amplitude. We found that foot sole afferent

64

classes are uniquely tuned to vibration stimuli; however, unlike in the hand, they cannot

65

be independently activated by class specific frequencies. Viewing the foot sole as a

66

sensory structure, the present findings may aid in the refinement of sensorimotor control

67

models and design of balance enhancement interventions.

68

Introduction

69

Skin feedback from the feet and ankles plays an important role in standing balance

70

and the control of gait. It is well established that cutaneous afferents from the foot sole and

71

dorsum can modulate lower (Fallon et al., 2005) and upper limb (Bent and Lowrey, 2013)

72

motor neuron excitability, evoke posturally relevant reflexes (Zehr and Stein, 1999), in

73

addition to providing proprioceptive (Collins, 2005; Aimonetti et al., 2007; Mildren and

74

Bent, 2016) and exteroceptive (Kavounoudias et al., 1998) feedback about body orientation

75

and interactions with the environment. This understanding has led to a growing interest in

76

improving postural control through the enhancement of foot sole cutaneous feedback

77

(Priplata et al., 2005; Perry et al., 2008; Zehr et al., 2014; Lipsitz et al., 2015). Facilitatory

78

shoe insoles, which employ suprathreshold (Novak and Novak, 2006) and subthreshold

79

(Priplata et al., 2005; Galica et al., 2009; Lipsitz et al., 2015) vibrations, have been shown

80

to improve balance and gait parameters in older adults and clinical populations. The

81

benefits of these subthreshold insole vibrations are believed to manifest themselves by

82

lowering the activation threshold of cutaneous receptors, thus making natural inputs

83

suprathreshold and able to generate viable and appropriate balance responses. Vibration has

3

84

been shown to induce sensations of whole body lean in restrained subjects (Roll et al.,

85

2002) and evoke postural sway away from stimulated sites during quiet stance

86

(Kavounoudias et al., 1998; 1999; Roll et al., 2002), thereby suggesting that the central

87

nervous system (CNS) uses cutaneous feedback from the soles of the feet to deduce body

88

spatial position. The sensitivity of foot sole cutaneous afferents and their ability to provide

89

vibration feedback has direct consequences on standing balance.

90

To comprehend the contributions of foot sole cutaneous feedback in postural

91

control, it is imperative to have an accurate understanding of the firing properties of the

92

four primary cutaneous afferent classes in response to vibration (i.e., when and how do they

93

respond). As an experimental tool, vibration provides a controllable stimulus that can be

94

used to investigate these properties. Impulses per vibratory cycle (ImpCycle) represents the

95

number of discharges evoked in response to vibration divided by the stimulus frequency,

96

and indicates the ability of the afferent to entrain to the stimulus. An ImpCycle response of

97

1:1 signifies entrainment, where the afferent discharges once per probe indentation.

98

Similarly, an ImpCycle response of 0.5:1 indicates that the afferent is firing on average

99

once every other indentation. As such, the ImpCycle response provides a normalized

100

afferent response, which highlights the capacity of the different afferent classes to encode

101

vibration stimuli across frequencies.

102

Previous work has investigated cutaneous afferent firing characteristics in the leg

103

and dorsum of the foot (Vedel and Roll, 1982; Ribot-Ciscar et al., 1989; Trulsson, 2001),

104

and specific afferent class tuning curves have been developed for afferents innervating the

105

glabrous skin on the hand (Johansson et al., 1982). Johansson et al. (1982) identified the

106

specific frequency ranges over which each cutaneous afferent class (Fast and slowly

4

107

adapting type I and II; FAI, FAII, SAI, SAII) is tuned to respond (has ability to entrain) in

108

the hand and found that FA afferents were tuned to high frequencies. Specifically, FAI

109

afferents were most easily activated between 8-64Hz, and FAII afferents between 64-

110

400Hz. In contrast, SA afferents were tuned to low frequencies; SAIs between 2-32Hz, and

111

SAIIs below 8Hz. The presence of unique cutaneous afferent class vibration tuning in the

112

hand indicates that the CNS has access to a range of feedback that can be used to shape

113

tactile experience and reflex responses.

114

The hands and feet are used for different functional roles and it is reasonable that

115

differences exist in the response properties of the underlying mechanoreceptor afferents.

116

Foot sole cutaneous afferents have been shown to have higher thresholds in response to

117

light touch compared to afferents in the hand and arm (Johansson and Vallbo, 1979a;

118

Vallbo et al., 1995; Kennedy and Inglis, 2002; Strzalkowski et al., 2015a). Further, there is

119

a distinct proximal to distal increase in type I receptors in the hand (Johansson and Vallbo,

120

1979b), while a more even distribution has been proposed in the foot sole (Kennedy &

121

Inglis 2002). Although these findings have provided valuable insights into the general

122

firing and receptive field characteristics of cutaneous afferents, the firing properties of the

123

specific afferents classes innervating the skin of the foot sole in response to vibration

124

remains unknown.

125

The cutaneous mechanoreceptor classes that innervate the glabrous skin on foot sole

126

are thought to be the same as those in the hand (Kennedy and Inglis, 2002). We

127

hypothesized that foot sole afferent class vibration tuning curves would be similar to those

128

in the hand (Johansson et al., 1982), but with higher thresholds as previously demonstrated

129

with light touch (Strzalkowski et al., 2015a). In the hand, the availability of distinct

5

130

afferent class feedback may be necessary to provide the sensory resolution associated with

131

texture perception and fine motor control (Yau et al., 2016). The tactile stimuli associated

132

with postural sway and gait may also require distinct afferent responses that cannot be fully

133

represented by a single vibration stimuli. In order to evaluate the contribution of cutaneous

134

input in postural control, it is essential to have an understanding of the capability of

135

individual foot sole cutaneous afferents to contribute to postural control. As such,

136

understanding the capacity of foot sole afferents to respond to select vibration is an

137

important step in creating a model of tactile perception and motor control.

138

The objectives of the present study are to: (i) investigate the vibration response

139

characteristics of cutaneous afferents in the glabrous skin of the foot; and (ii) expand upon

140

previous studies conducted on the glabrous skin of the hand. Accordingly, single unit

141

microneurography was used to record the firing responses of cutaneous afferents, i.e.,

142

ImpCycle. These data provide a measure of afferent vibration tuning, considered to be the

143

frequency, or range of frequencies where individual afferents can easily entrain (at a low

144

amplitude) and when entrainment is insensitive to amplitude changes.

145

Methods

146

Ethical Approval

147

Fifty-nine recording sessions were performed on 21 healthy subjects (12 males, 9

148

females; mean age 24 with range 20-27 years). None of the participants had any known

149

neurological or musculoskeletal disorders. Following an explanation of the protocol, each

150

subject gave written informed consent to participate in the experiment. The protocol was

151

approved by the University of Guelph research ethics board and complied with the

152

declaration of Helsinki.

6

153

Experimental setup

154

Subjects lay prone on an adjustable treatment table. Both legs were straight with the

155

right, test leg supported with a Versa Form positioning pillow at the level of the ankle. All

156

recordings were performed in the right tibial nerve. The path of the tibial nerve was located

157

at the level of the popliteal fossa using transdermal electrical stimulation (1-ms square

158

wave pulse, 1Hz 0-10mA, SIU-C Grass stimulus isolation unit and a S88X Grass

159

stimulator; Grass Instruments Astro-Med, West Warwick, RI). Stimuli were applied

160

through a handheld probe with a reference surface electrode (Ag/AgCl) attached to the

161

patella of the right knee. The tibial nerve location was established by observable muscle

162

twitches in the plantar flexor muscles, paired with subject reported sensations of parasthesia

163

in the foot and leg (representative of the tibial innervation zone). The location for insertion

164

was established as the location with the largest response at the lowest current. A low

165

impedance reference electrode (uninsulated, tungsten, 200µm diameter; FHC Inc. Bowdoin,

166

ME, USA) was inserted through the skin ~2cm medial to the recording site at a depth of

167

~5mm. The recording electrode (insulated 10MΩ, tungsten, 200µm diameter, 1-2 µm

168

recording tip, 55mm length; FHC Inc.) was then inserted at the predetermined tibial nerve

169

recording location. Using audio feedback of the neural activity, the tibial nerve was located

170

and penetrated using small manipulations of the recording electrode. Mechanical activation

171

(light tapping, stroking and stretching) of the foot sole skin was then applied to help guide

172

fine manipulations of the electrode to isolate single afferents. Neural recordings were

173

amplified (gain 104, bandwidth 300Hz-10kHz, model ISO-180; World Precision

174

Instruments, Sarasota, FL), digitally sampled (40kHz), and stored for analysis (CED 1401

175

and Spike2 version 6; Cambridge Electronic Design).

7

176
177

Cutaneous mechanoreceptor classification
Single afferents were classified as fast adapting (FAI or FAII) or slowly adapting

178

(SAI or SAII) based on previously described criteria (Johansson, 1978; Kennedy and Inglis,

179

2002). FA afferents are sensitive to dynamic events and adapt quickly to sustained

180

indentations. In contrast, SA afferents respond throughout sustained skin indentation, and

181

their firing rates are proportional to the magnitude of deformation. FAI and SAI afferents

182

typically have small receptive fields with multiple hotspots (locations of highest sensitivity)

183

and distinct borders, while FAII and SAII afferents have large receptive fields, a single

184

hotspot and less well-defined borders. To improve classification accuracy, additional tests

185

were performed to identify units, such as manual skin stretch of SAII afferent receptive

186

fields, blowing across the receptive field of FAII afferents, and calculations of the

187

instantaneous frequency (SAII regular, SAI irregular). Semmes-Weinstein monofilaments

188

(North Coast Medical Inc, Gilroy, California) capable of applying ~0.078-2941mN of

189

force, were applied by hand to calculate receptive field size, monofilament firing threshold,

190

and hotspot location (site of maximum sensitivity). Receptive fields were then determined

191

with a monofilament 4-5 times afferent firing threshold and drawn on the skin with a fine

192

tip pen. Only units whose receptive fields fell within the plantar surface of the foot sole

193

were included in this study.

194

Vibration protocol

195

Sinusoidal vibrations were delivered through a 6mm diameter probe driven by a

196

vibration exciter (Mini-shaker type 4810, Power amplifier type 2718, Bruel & Kjaer,

197

Naerum, Denmark) secured on an adjustable arm (143BKT, LinoManfrotto, Markham,

198

Canada). The probe was positioned perpendicular to the receptive field hotspot, and 2mm

8

199

of pre-indentation was applied with a manual displacement gauge. Force was measured

200

with a force transducer (load cell model 31, Honeywell, MN, USA) placed in series with

201

the probe. Force feedback was used to monitor the probe position and to ensure consistent

202

contact with the foot sole at that position. Probe acceleration (ΔV/sec) was recorded with

203

an accelerometer (sampled at 2kHz; 4507 B 002; Bruel and Kjaer, Germany), and used in a

204

closed-loop system to control stimulus frequency and amplitude (VR8500 Vibration

205

Controller, VibrationVIEW v. 7.1.4; Vibration Research corporation, Jenison MI).

206

A single frequency was tested at a time and delivered through a vibration ramp

207

consisting of multiple 2-second vibration bursts of increasing amplitudes. To limit

208

habituation, there was a four-second pause given between each vibration burst within each

209

ramp. A broad range of control stimulus frequencies (3, 5, 8, 10, 20, 30, 60, 100, 150 and

210

250Hz) and of amplitudes (0.001, 0.0025, 0.005, 0.0075, 0.01, 0.025, 0.05, 0.075, 0.1, 0.25,

211

0.5, 0.75, 1, 1.25, 1.5, 1.75 and 2mm) were tested. Note that the order of frequencies tested

212

was pseudo-randomly selected, to limit any influence of order while maximizing the range

213

of frequencies tested for each afferent given the unpredictability of recording stability.

214

Further, amplitudes were not randomized within each ramp to ensure consistent stimuli

215

across recordings and to facilitate real time firing threshold identification. Not every

216

frequency-amplitude combination was possible due to an acceleration feedback

217

requirement of the closed loop system necessary for the accurate control of peak-to-peak

218

probe displacement. Consequently, some low frequency-small amplitude and high

219

frequency-large amplitude combinations were not testable. For example, we could not

220

vibrate below 3Hz (acceleration signal was too weak), and only at low-subthreshold

221

amplitudes at 400Hz. We were unable to evoke responses at 400Hz at the highest amplitude

9

222

(2.5um) and 400Hz vibration ramps were not tested. Depending on the stability/quality of

223

the neural recording, 1 to 10 vibration ramps (representing 1 to 10 different frequencies)

224

were delivered to each afferent and 7 to 15 amplitudes (vibration bursts) were tested within

225

a vibration ramp at any delivered frequency.

226

Analysis of neural recordings

227

Recorded afferent signals were analyzed using Spike2 (version 6; Cambridge

228

Electronics Design). Spike morphology was used to generate a template for the visual

229

classification of single units. Figure 1 depicts an example of a 30Hz vibration ramp and

230

FAII afferent firing response. The top panel presents the instantaneous afferent firing

231

frequency response (Hz), the 30Hz acceleration profile of the relative timing and amplitude

232

of each vibration ramp (ΔV/sec), the peak-to-peak amplitude of each vibration burst (mm),

233

and the raw neurogram of the single unit recording (V). Recordings in which a single

234

afferent could not be confidently identified without interference from multiple units or

235

excessive signal noise were excluded from further analysis.

236

Afferent firing characteristics were calculated from a representative one-second of

237

each two-second vibratory burst. The one-second period was selected at the end of each

238

vibration burst just before the vibration amplitude decreased; see Figures 1A) and 1B). This

239

period was selected to standardize the analysis period, and to limit the influence of an

240

amplitude overshoot present at the beginning of some vibration bursts. The ImpCycle

241

response was calculated from this one-second period.

242

Subsequently, afferent firing threshold and entrainment threshold were identified.

243

For this analysis, entrainment threshold was defined as the lowest amplitude for which

244

impulses per cycle was greater than 0.9 for a given frequency. Greater than 0.9 was chosen

10

245

as entrainment threshold to account for instances where the afferent could clearly entrain

246

but failed to discharge on an indentation due to inherent spiking variability. Tables and

247

figures are labelled as 1:1 instead of >0.9:1 for ease of interpretation.

248

Statistical analyses

249

Afferent served as the observational unit in all statistical analyses. Descriptive

250

statistics of the monofilament threshold, receptive field area, ImpCycle, entrainment

251

threshold, and firing threshold were used to provide an overview of the data. At each

252

frequency-amplitude combination, the mean impulses per cycle were plotted (Figure 3,

253

MATLAB 6.1, The MathWorks Inc., Natick MA, and Figures 4-7, GraphPad Prism version

254

5.0c for Mac OS X, San Diego CA). The percentage of afferents firing (% firing) and

255

percentage firing at or greater than 1:1 (% ≥1:1) were also tabulated to highlight afferent

256

firing behaviour both within and across vibration ramps at select frequency-amplitude

257

combinations.

258

Regression analyses of ImpCycle, entrainment threshold, and afferent firing

259

threshold were performed. In order to accommodate the correlation of multiple

260

observations within an afferent, the data were modelled using generalized estimating

261

equations (GEE) (Zeger and Liang, 1986); exchangeable correlation structure and robust

262

standard errors via the gee function from the gee package in R (R Core Team, 2017).

263

There are several correlation structures one could specify, with typical ones being

264

independent (the responses at each frequency-amplitude combination are independent of

265

each other), exchangeable (equal correlation for all observations across the frequency-

266

amplitude plane), and auto-regressive (correlations exist but are stronger for nearby

11

267

frequency-amplitude combinations and decrease for combinations that are further away).

268

While the exchangeable correlation structure is likely inaccurate, (Zeger and Liang, 1986)

269

show that misspecification of correlation structure does not affect consistency of coefficient

270

estimates. Further, the use of robust standard errors helps mitigate any potential influence

271

on standard errors.

272

Both frequency and amplitude were transformed to the natural logarithm scale to

273

improve model fit. Afferent class was treated as a factor with FAI afferents forming the

274

reference group. Accordingly, coefficients associated with every other afferent class in the

275

regression of ImpCycle can be interpreted as the expected difference in ImpCycle between

276

that afferent class and FAI afferents. Since amplitude was logged, in the regressions for

277

(log) entrainment and for (log) firing threshold, the interpretation of coefficients is no

278

longer in terms of the mean response (i.e., entrainment or firing threshold), but in terms of

279

the median response.

280

Results

281

Out of 111 recordings in which an afferent was confidently identified, 52 individual

282

cutaneous afferents were recorded from 16 healthy subjects over 39 recordings sessions.

283

Some of the identified afferents were reported in a previous study (Strzalkowski et al.,

284

2015a). For each subject, the number of successful recording sessions ranged from 1 to 6,

285

and at most 3 cutaneous afferents were identified in any single recording session. The total

286

number of afferents identified on any subject ranged from 1 to 9.

287

The recorded afferents all had receptive fields in the plantar surface of the foot sole,

288

and the recordings were stable enough to permit at least one complete vibration ramp to be

289

applied. On average, five different frequency ramps, out of a possible ten, were tested on

12

290

each of the fifty-two recorded afferents. Afferents were not observed to habituate to the 2-

291

second bursts or across multiple vibration ramps. The afferents tested included 19 FAI

292

(37%), 9 FAII (17%), 14 SAI (27%) and 10 SAII (19%) afferents.

293

The monofilament firing thresholds tend to be larger for slowly adapting afferents

294

compared to fast adapting afferents, with both larger means and variation (Table 1). The

295

receptive fields for type II afferents (FAII and SAII) tended to be larger, with larger

296

variation, compared to the type I afferents (FAI and SAI). The location and distribution of

297

the cutaneous afferent receptive fields are presented in Figure 2.

298

Descriptive Statistics

299

Table 2 presents descriptive statistics for the impulses per cycle, entrainment

300

threshold, and firing threshold by afferent class. All afferent classes were tested across the

301

respective tested frequency ranges, though SA afferents responded over a restricted range

302

of frequencies (FAs: 3-250Hz, SAI: 3-150Hz, SAII: 3-100Hz). On average, ImpCycle

303

responses for FAI (mean: 0.63), SAI (mean: 0.55), and SAII (mean: 0.33) were slightly

304

lower than that of FAII afferents (mean: 1.07). Entrainment and afferent firing threshold

305

tended to be highest for SAII afferents (means: 1.11mm and 0.79 mm, respectively) and

306

lowest for FAII afferents (means: 0.33mm and 0.17mm, respectively).

307

Figure 3A) shows the mean ImpCycle for each afferent as a heat map over the

308

frequency-amplitude plane, with higher ImpCycle responses corresponding to hotter

309

colours. Despite the restricted frequency range for SA afferents, there is little that

310

differentiates the SAI afferent responses from that of the FAI afferents. For reference,

311

Figure 3B) provides the analogous plot for the hand, adapted from (Johansson et al., 1982).

312

Impulses per cycle

13

313

Figures 4-7 show the tuning curves for each afferent class, respectively. The largest

314

FAI afferent ImpCycle responses (>2:1) were observed at low frequencies (3-8Hz), and

315

responses greater than 1:1 were observed at frequencies up to 60Hz (Figure 4). Similar to

316

the other afferent classes, the ImpCycle responses varied greatly across individual FAI

317

afferents (Figure 4B,C). FAII afferents were found to be the most sensitive class to

318

vibration stimuli, demonstrating the highest ImpCycle responses across frequencies

319

compared to the other classes (Figure 5). Average ImpCycle responses >3:1 were observed

320

up to 10Hz, and >1:1 up to 150Hz at the available vibration amplitudes. Individual FAII

321

afferent responses varied within the population, indicating a continuum of coverage

322

(varying levels of sensitivity) class for a particular frequency (Figure 5B, C). Two different

323

FAII afferent responses were observed: a high amplitude (>0.5mm) response where the

324

impulses per cycle decreased with increasing stimulus frequency; and, a low amplitude

325

response (≤0.5mm), where the impulses per cycle were similar across frequencies. These

326

trends support a high capacity for FAII afferents to respond across frequencies.

327

SAI afferents tended to respond across the range of frequencies tested and at

328

elevated thresholds compared to FA afferents (Table 2, Figure 6). SAI afferents

329

demonstrate their largest ImpCycle responses at low frequencies, but do not appear to be

330

tuned to a specific range like FAI afferents, as a consistent ImpCycle response is not seen

331

across stimulation frequencies at a given vibration frequency. SAII afferents were the most

332

insensitive to vibration. They were associated with the lowest ImpCycle responses across

333

most frequencies and a near absence of firing at amplitudes below 0.25mm (Figure 7).

334

Average SAII afferent ImpCycle responses >1:1 were only achieved at 3Hz and 5Hz;

335

however, average responses >0.5:1 were found up to 60Hz (Figure 7).

14

336

Table 3 presents the regression of the ImpCycle on frequency and amplitude. These

337

results validate the trends seen in Figure 3. At a given frequency-amplitude combination,

338

the ImpCycle tended to be larger for FAII afferents but smaller for SAII afferents (p <

339

0.001 for both afferents) compared to FAI afferents; however, there was no statistically

340

significant difference between SAI and FAI afferents (Table 3, p = 0.274). ImpCycle

341

response was also found to depend on the interaction of frequency and amplitude (p <

342

0.001). In particular, for a given frequency, as amplitude was ramped up, ImpCycle also

343

increased; however, the amount of increase depended on the frequency, with the effect of

344

amplitude dampened at higher frequencies. For a given amplitude, as frequency was

345

increased, ImpCycle tended to decrease, but the amount of decrease depended on the

346

amplitude.

347

Entrainment threshold

348

FAI afferent firing remained robust at higher frequencies, where 77% and 43% of

349

the FAI afferents achieved 1:1 firing at 100Hz and 150Hz at the largest available

350

amplitudes of 0.5mm and 0.25mm respectively (Table 4). These ImpCycle responses reflect

351

the fast adapting nature of FAI afferents and their ability to entrain to high frequencies.

352

FAI had the most consistent ImpCycle responses between 8-60Hz, as the ability of FAI

353

afferents to entrain did not deviate greatly between 8Hz and 60Hz at low vibration

354

amplitudes (Figure 4). As such, 8-60Hz may specify a frequency range over which FAI

355

afferents are tuned.

356

FAII afferents had the lowest entrainment thresholds compared to the other afferent

357

classes. Entrainment was reached in all FAII afferents at amplitudes of 0.25mm or 0.5mm

358

(except 100Hz 75% at 0.5mm and 250Hz 20% at 0.075mm) (Table 4). Note that at 250Hz

15

359

one afferent (of five tested) did not respond, and one entrained to the stimulus. Further,

360

entrainment threshold at 10Hz could not be isolated for FAII afferents because >1:1 firing

361

was evoked on average at the smallest applied amplitude (0.05mm). As a population, FAII

362

afferents responded and entrained at all the frequencies tested (except at 250Hz where an

363

average afferent entrainment threshold was not observed at the largest amplitude of

364

0.075mm).

365

SAI afferent entrainment thresholds increased with increasing frequency up until

366

30Hz, above which an average population response of 1:1 was not observed (Figure 6,

367

Table 4). At 3Hz SAI entrainment threshold was 0.1mm, which increased to around

368

1.25mm at 30Hz. An individual afferent (1 of 5) was found to entrain up to 100Hz;

369

however, entrainment was not observed in frequencies above 30Hz in the majority of SAI

370

afferents (Table 4). These data suggest that as a population, SAI afferents preferentially

371

encode for low frequencies below 30Hz, but some afferents can respond and entrain at

372

higher frequencies.

373

SAII afferents had a limited capacity to entrain to the vibratory stimuli, and

374

entrainment was only observed at amplitudes greater than 1mm over the 3-10Hz frequency

375

range (Table 4). At 10Hz, all five of the SAII afferents tested fired at 1.5mm and 40% had

376

reached entrainment threshold. Above 10Hz (and tested up to 30Hz), 1:1 firing was not

377

achieved even at 2mm, so 10Hz represents the upper frequency at which SAII afferents

378

were shown to consistently respond.

379

Overall, entrainment threshold was found to decrease with increasing frequency for

380

both FAI and FAII afferents (Figure 8). From the regression of entrainment threshold on

381

frequency, the ratio of median entrainment threshold of FAII afferents relative to that of

16

382

FAI afferents measured at the same frequency was estimated to be exp(-1.168)=0.311

383

(Table 5). In other words, the median entrainment threshold of FAII afferents tended to be

384

just over 30% of the median entrainment threshold of FAI afferent for a given frequency (p

385

< 0.001). Further, when comparing afferents of the same class but for which the frequency

386

of the first group of afferents is, say, double that of the second group, then the ratio of

387

median entrainment threshold for the two groups was estimated to be 2–0.151= 0.901. So,

388

the doubling of frequency was associated with a 10% decrease in median entrainment

389

threshold (p = 0.069). The trends for SA afferents were inconclusive, likely due to the fact

390

that they were tested at fewer frequencies compared to FA afferents.

391

Firing threshold

392

The trends in firing thresholds across frequencies for each afferent class were

393

similar to that of entrainment, but with responses at higher frequencies as well (Figure 9).

394

From the GEE analysis, the ratio of median firing threshold of FAII afferents relative to

395

that of FAI afferents measured at the same frequency was estimated to be exp(–1.454) =

396

0.233 (Table 6). In other words, the median firing threshold of FAII afferents was, again,

397

much lower and tended to be just under 25% of the median firing threshold of FAI afferents

398

for a given frequency (p < 0.001), i.e. more sensitive. Further, when comparing afferents of

399

the same class but for which the frequency of the first group of afferents is, say, double that

400

of the second group, then the ratio of median firing threshold for the two groups was

401

estimated to be 2–0.245= 0.844 (Table 6). So, the doubling of frequency was associated with

402

a just over 15% decrease in median firing threshold (p = 0.008).

403

Discussion

17

404

The present study was conducted to (i) identify and characterize the firing

405

characteristics of foot sole cutaneous afferents in response to vibratory stimuli, and (ii)

406

provide a reference to analogous studies performed in the hand. We have identified class

407

specific vibration tuning frequencies for foot sole cutaneous afferents, demonstrated that

408

there are (overlapping) ranges of individual afferent responses associated with each class,

409

and shown how these responses differ from afferent responses in the hand.

410

Vibration tuning of foot sole cutaneous afferents

411

We vibrated the foot sole across a range of frequencies and amplitudes and found

412

that FAII afferents demonstrated the most robust firing response compared to the other

413

classes of cutaneous afferents. FAII afferents fired and entrained at relatively low vibration

414

amplitudes, and had the highest ImpCycle responses across nearly all frequency-amplitude

415

combinations. FAII afferents appear to be tuned across our range of vibration input;

416

reaching entrainment threshold at similar amplitudes (0.05-0.25mm) across frequencies (5-

417

150Hz). The relatively low firing threshold of FAII afferents suggests that FAIIs may be

418

the only class of cutaneous afferents that can be isolated (respond without contamination

419

from other afferents) with foot sole vibration. The functional significance (i.e. tactile

420

perception, postural control) of FAII feedback cannot be determined from the present data;

421

however it is clear that FAII afferent feedback will be present in most or all tactile

422

responses.

423

In contrast to the ubiquitous firing of foot sole FAII afferents, overlapping

424

entrainment frequency ranges were present for FAI, SAI, and SAII afferent classes. FAI

425

afferents were found to have consistent ImpCycle responses between 8-60Hz, indicating a

426

vibration range over which they most faithfully encode the frequency. As a population, SAI

18

427

and SAII afferents most readily entrained to frequencies below 20Hz, where SAI afferents

428

tended to have lower entrainment thresholds, and larger ImpCycle responses, compared to

429

SAII afferents.

430

Comparison between body regions

431

The foot sole vibration tuning curves established in the present study are in partial

432

agreement with data from the hand. Johansson et al. (1982) published optimum entrainment

433

frequency ranges for hand cutaneous afferents and reported that SA afferents were most

434

easily entrained at low frequencies (SAI 2-32hz, SAII <8Hz), while FA afferents were

435

tuned to higher frequencies (FAI 8-64Hz, FAII >64Hz). In the foot sole, FA afferents are

436

similarly found to respond and entrain to higher frequencies compared to SA afferents,

437

however distinct tuning ranges are less apparent.

438

Interestingly, the foot sole and hand data demonstrate differences in the number of

439

spikes evoked across frequencies when exposed to similar amplitudes. Hand afferents are

440

found to be more sensitive, discharging an increased number of spikes at a given frequency

441

amplitude combination. At 4Hz, 1mm peak-to-peak vibration evoked an average firing rate

442

of 4:1 in FAI afferents, and 5:1 in FAII afferents in the hand (Johansson et al., 1982). In

443

contrast, 5Hz-1mm vibration evoked 2:1 and 3:1 firing in FAI and FAII foot sole afferents

444

respectively. Similarly, the firing rates of hand SA units were typically shown to be double

445

that of foot sole SA afferents. It appears that cutaneous afferents in the hand have lower

446

firing thresholds, entrain more easily, and discharge more spikes at a given amplitude

447

compared to the same afferent classes in the foot sole. This disparity between the foot sole

448

and hands may be due to distinct mechanoreceptor adaptations related to sensory function

449

and skin mechanical property differences (Kekoni et al., 1989; Strzalkowski et al., 2015b).

19

450

The increased afferent firing in the hands relative to the feet may represent the requirement

451

for texture and velocity perception necessary for handling objects. Conversely, the postural

452

significance of foot sole feedback may not necessitate high individual afferent firing and

453

rather rely on population characteristics across the foot sole.

454

Vibration responses of cutaneous afferents in the lower limb have been previously

455

investigated; however, these data were limited to recordings from the lateral fibular nerve,

456

with afferent receptive fields in the leg and foot dorsum (Vedel and Roll, 1982; Ribot-

457

Ciscar et al., 1989). These early studies combined cutaneous afferents into FA and SA

458

groups, and applied fewer frequency-amplitude combinations (10-300Hz, 0.2 and 0.5mm

459

amplitude, 1mm diameter probe) than the present study. Despite these methodological

460

differences, FA afferents were found capable of entraining to higher frequencies compared

461

to SA afferents, which is in agreement with the present findings. Interestingly, firing rates

462

greater than 1:1 were not observed in leg and foot dorsum afferents at the largest 0.5mm

463

amplitude (Ribot-Ciscar et al., 1989). In contrast, 0.5mm vibration amplitude evoked firing

464

rates greater than 1:1 in all foot sole afferent classes in the current work at class specific

465

frequencies (Table 4). The elevated thresholds of afferents innervating the leg and foot

466

dorsum compared to the plantar surface of the foot sole may reflect the functional

467

significance of foot sole cutaneous feedback in controlling standing balance compared to

468

feedback from other lower limb regions.

469

Functional implications of foot sole vibration

470

It is well established that cutaneous feedback from the soles of the feet is important

471

in the control of upright stance and gait (Hayashi et al., 1988; Kavounoudias et al., 1998;

472

Meyer et al., 2004; Kars et al., 2009). The application of subthreshold foot sole vibration,

20

473

which is thought to increase the availability of foot sole feedback, has been shown to

474

improve standing balance and gait, as evidenced by reductions in postural sway (Priplata et

475

al., 2005; Lipsitz et al., 2015) and measures of gait variability (Galica et al., 2009; Lipsitz et

476

al., 2015). In addition, suprathreshold foot sole vibration has been shown to modulate

477

postural sway, where the magnitude and velocity of sway increases with higher frequencies

478

(Kavounoudias et al., 1999). The present data demonstrate that the vibration frequencies

479

and amplitudes (20, 60, 100 Hz and 0.2-0.5mm) employed by Kavououdias et al. (1999)

480

can cause robust firing across afferent classes. However, what we have also shown is that it

481

may not be possible to establish how each afferent class response contributes specifically to

482

the observed postural responses. Importantly, these previous studies support the concept

483

that altering foot sole cutaneous feedback via subthreshold and suprathreshold vibration

484

does change afferent firing, and can ultimately modulate standing balance. However, since

485

it is unknown how the cutaneous afferent firing characteristics change under different

486

loading conditions, the present data should not be taken to represent afferent firing in

487

different contexts.

488

The frequencies transmitted through the foot sole in association with gait and

489

standing balance are not well understood. Regardless, our results demonstrate that foot sole

490

cutaneous afferents respond and contribute tactile feedback in response to large amplitude

491

vibrations. Further, foot sole cutaneous afferents are silent at rest but show robust

492

responses, especially FA afferents, to dynamic mechanical perturbations of the skin. In fact,

493

the heterogeneous vibration tuning we observed across classes suggests that the CNS may

494

be primed for class specific firing rates. For example, we hypothesize that a high frequency

495

burst from FA afferents will produce a different, perhaps muted, response in higher order

21

496

neurons than that from a similar SA afferent input. We also conjecture that standing

497

balance likely evokes robust firing across all foot sole afferent classes, and the relative

498

contributions of each class in modulating motor output may depend on the postural context.

499

Future work is still needed to investigate higher order class specific responses and their

500

behaviour under different postures.

501

The functional significance and central weighting of feedback from different

502

afferent classes is not clear. It is reasonable that low frequency response patterns from SA

503

afferents would evoke a different, perhaps larger, postural response than a similar low

504

frequency firing pattern from a population of FA afferents. The afferent signal is important

505

in the modulation of muscle reflexes and postural control likely arises from a combination

506

of afferent input, where external stimuli has been shown to evoke a range of firing across

507

classes (Fallon et al., 2005). We found that afferent firing rate increased with high vibration

508

amplitude; however, since the data were collected with the subject prone and the foot sole

509

unloaded (2mm probe pre-indentation), these results cannot be extrapolated to an upright

510

loaded posture. Future studies are needed to identify the firing characteristics of cutaneous

511

afferents under prolonged loaded conditions and in older adult and/or diseased populations

512

to further explore the functional significance of distinct afferent class vibration tuning.

513

Summary and conclusions

514

The present experiment provides an analysis of the vibration response

515

characteristics of cutaneous afferents in the glabrous skin of the human foot sole. Tactile

516

feedback from the feet plays an important role in the control of standing balance and gait,

517

and the present findings expand upon how the foot sole is viewed as a sensory structure.

518

We have demonstrated that cutaneous afferents in the foot sole display class specific tuning

22

519

to vibratory stimuli, and that FAII afferents have the lowest firing and entrainment

520

thresholds across frequencies. Optimal entrainment frequencies for FAI, SAI, and SAII

521

afferents were found to overlap below 20Hz. Foot sole FAI afferents were tuned to

522

faithfully encode frequencies between 8Hz and 60Hz, where similar ImpCycle responses

523

were found at low vibration amplitudes. Afferent class vibration entrainment ranges

524

provide a clear indication of the capacity of cutaneous afferents to faithfully encode a given

525

vibration stimulus. Vibrations associated with natural stimuli are expected to evoke

526

complex patterns of afferent firing that combine to inform perceptual experience and motor

527

control. Ultimately this work provides a mechanistic look at the capacity of foot sole

528

cutaneous afferents to respond to vibration, which may aid in the development of

529

sensorimotor control models and design of balance enhancement interventions.

530
531

Grants: This project was funded by a NSERC Discovery Grant (400723) to Leah Bent and

532

a NSERC PGSD to Nicholas Strzalkowski.

533

Conflict of interest: The authors declare that there are no conflicts of interest, financial or

534

otherwise.

535

Author contributions: N.D.J.S and L.R.B conceived and designed the research; N.D.J.S

536

and L.R.B performed the experiments; N.D.J.S, R.A.A and L.R.B interpreted results of

537

experiments; N.D.J.S and R.A.A analyzed data and prepared figures; N.D.J.S drafted

538

manuscript; N.D.J.S, R.A.A and L.R.B edited and revised manuscript; N.D.J.S, R.A.A and

539

L.R.B approved final version of manuscript.

23

540

Acknowledgments: The authors would also like to thank Robyn Mildren for her

541

contributions towards data collection, and Derek Zwambag for his help in preparing Figure

542

3.

24

543

References

544
545
546

Aimonetti J-MJ, Hospod VV, Roll J-PJ, Ribot-Ciscar EE. Cutaneous afferents provide a
neuronal population vector that encodes the orientation of human ankle movements. J
Physiol (Lond) 580: 649–658, 2007.

547
548
549

Bent LR, Lowrey CR. Single low-threshold afferents innervating the skin of the human
foot modulate ongoing muscle activity in the upper limbs. J Neurophysiol 109: 1614–1625,
2013.

550
551

Collins DF. Cutaneous Receptors Contribute to Kinesthesia at the Index Finger, Elbow,
and Knee. J Neurophysiol 94: 1699–1706, 2005.

552
553
554

Fallon JB, Bent LR, McNulty PA, Macefield VG. Evidence for strong synaptic coupling
between single tactile afferents from the sole of the foot and motoneurons supplying leg
muscles. J Neurophysiol 94: 3795–3804, 2005.

555
556
557

Galica AM, Kang HG, Priplata AA, D’Andrea SE, Starobinets OV, Sorond FA,
Cupples LA, Lipsitz LA. Subsensory vibrations to the feet reduce gait variability in
elderly fallers. Gait & Posture 30: 383–387, 2009.

558
559
560

Hayashi R, Miyake A, Watanabe S. The functional role of sensory inputs from the foot:
stabilizing human standing posture during voluntary and vibration-induced body sway.
Neuroscience Research 5: 203–213, 1988.

561
562
563

Johansson RS, Landstrom U, Lundstrom R. Responses of mechanoreceptive afferent
units in the glabrous skin of the human hand to sinusoidal skin displacements. Brain Res
244: 17–25, 1982.

564
565

Johansson RS, Vallbo AB. Detection of tactile stimuli. Thresholds of afferent units related
to psychophysical thresholds in the human hand. J Physiol (Lond) 297: 405–422, 1979a.

566
567
568

Johansson RS, Vallbo AB. Tactile sensibility in the human hand: relative and absolute
densities of four types of mechanoreceptive units in glabrous skin. J Physiol (Lond) 286:
283–300, 1979b.

569
570

Johansson RS. Tactile sensibility in the human hand: receptive field characteristics of
mechanoreceptive units in the glabrous skin area. J Physiol (Lond) 281: 101, 1978.

571
572
573

Kars HJJ, Hijmans JM, Geertzen JHB, Zijlstra W. The Effect of Reduced
Somatosensation on Standing Balance: A Systematic Review. Journal of Diabetes Science
and Technology 3: 931–943, 2009.

574
575

Kavounoudias A, Roll R, Roll JP. The plantar sole is a “dynamometric map” for human
balance control. Neuroreport 9: 3247–3252, 1998.

25

576
577

Kavounoudias A, Roll R, Roll JP. Specific whole-body shifts induced by frequencymodulated vibrations of human plantar soles. Neurosci Lett 266: 181–184, 1999.

578
579
580

Kekoni J, Hämäläinen H, Rautio J, Tukeva T. Mechanical sensibility of the sole of the
foot determined with vibratory stimuli of varying frequency. Exp Brain Res 78: 419–424,
1989.

581
582

Kennedy PM, Inglis JT. Distribution and behaviour of glabrous cutaneous receptors in the
human foot sole. J Physiol (Lond) 538: 995–1002, 2002.

583
584
585

Lipsitz LA, Lough M, Niemi J, Travison T, Howlett H, Manor B. A Shoe Insole
Delivering Subsensory Vibratory Noise Improves Balance and Gait in Healthy Elderly
People. Archives of Physical Medicine and Rehabilitation 96: 432–439, 2015.

586
587

Meyer PF, Oddsson LIE, De Luca CJ. The role of plantar cutaneous sensation in
unperturbed stance. Exp Brain Res 156: 505–512, 2004.

588
589

Mildren RL, Bent LR. Vibrotactile stimulation of fast-adapting cutaneous afferents from
the foot modulates proprioception at the ankle joint. J Appl Physiol 120: 855–864, 2016.

590
591

Novak P, Novak V. Effect of step-synchronized vibration stimulation of soles on gait in
Parkinson's disease: a pilot study. J NeuroEngineering Rehabil 3: 9, 2006.

592
593

Perry SD, Radtke A, McIlroy WE, Fernie GR, Maki BE. Efficacy and effectiveness of a
balance-enhancing insole. J Gerontol A Biol Sci Med Sci 63: 595–602, 2008.

594
595
596

Priplata AA, Patritti BL, Niemi JB, Hughes R, Gravelle DC, Lipsitz LA, Veves A,
Stein J, Bonato P, Collins JJ. Noise-enhanced balance control in patients with diabetes
and patients with stroke. Ann Neurol 59: 4–12, 2005.

597
598

Ribot-Ciscar E, Vedel JP, Roll JP. Vibration sensitivity of slowly and rapidly adapting
cutaneous mechanoreceptors in the human foot and leg. Neurosci Lett 104: 130–135, 1989.

599
600

Roll RR, Kavounoudias AA, Roll J-PJ. Cutaneous afferents from human plantar sole
contribute to body posture awareness. Neuroreport 13: 1957–1961, 2002.

601
602

Strzalkowski NDJ, Mildren RL, Bent LR. Thresholds of cutaneous afferents related to
perceptual threshold across the human foot sole. J Neurophysiol 114: 2144–2151, 2015a.

603
604
605

Strzalkowski NDJ, Triano JJ, Lam CK, Templeton CA, Bent LR. Thresholds of skin
sensitivity are partially influenced by mechanical properties of the skin on the foot sole.
Physiological Reports 3: e12425–e12425, 2015b.

606
607

Trulsson M. Mechanoreceptive afferents in the human sural nerve. Experimental Brain
Research 137: 111–116, 2001.

26

608
609
610

Vallbo AB, Olausson H, Wessberg J, Kakuda N. Receptive field characteristics of tactile
units with myelinated afferents in hairy skin of human subjects. J Physiol (Lond) 483 ( Pt
3): 783–795, 1995.

611
612

Vedel JP, Roll JP. Response to pressure and vibration of slowly adapting cutaneous
mechanoreceptors in the human foot. Neurosci Lett 34: 289–294, 1982.

613
614

Yau JM, Kim SS, Thakur PH, Bensmaia SJ. Feeling form: the neural basis of haptic
shape perception. J Neurophysiol 115: 631–642, 2016.

615
616

Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes.
Biometrics 42: 121–130, 1986.

617
618
619

Zehr EP, Nakajima T, Barss T, Klarner T, Miklosovic S, Mezzarane RA, Nurse M,
Komiyama T. Cutaneous stimulation of discrete regions of the sole during locomotion
produces “sensory steering” of the foot. BMC Sports Sci Med Rehabil 6: 33, 2014.

620
621

Zehr EP, Stein RB. What functions do reflexes serve during human locomotion? Progress
in neurobiology 58: 185–205, 1999.

622

27

623

Table captions

624

Table 1: Characteristics of cutaneous afferents identified and tested in the foot sole

625
626

Table 2: Descriptive statistics for impulses per cycle, entrainment threshold and firing
threshold, post stratified by afferent class

627
628

Table 3: Multivariate regression of impulses per cycle with exchangeable correlation
structure

629
630
631

Table 4: Cutaneous afferent class firing characteristics, percent firing (% firing) and percent
firing at or greater than 1:1 (% ≥1:1) as well as the sample number (n) across select
frequency-amplitude combinations

632
633

Table 5: Regression of entrainment threshold on frequency and afferent class with
exchangeable correlation structure

634
635

Table 6: Regression of firing threshold on frequency and afferent class with exchangeable
correlation structure

636

Figure captions

637
638
639
640
641
642
643
644
645
646
647

Figure 1: A 30Hz vibration ramp (amplitude 0.005-1.25mm) with the firing response of a
representative FAII afferent. The top panel presents from top to bottom: the instantaneous
afferent firing frequency response (Hz), the 30Hz acceleration profile of the vibration ramp
(ΔV/sec), the peak-to-peak amplitude of each vibration burst (mm), and the raw neurogram
of the single unit recording (V). Within the afferent firing response panel the gray dashed
and dotted lines indicate 60Hz and 30Hz responses respectively. With increasing vibration
amplitude (left to right) the FAII afferent firing response increased. Low stimulus
amplitudes (0.005-0.0075mm) did not evoke an afferent response. A) Highlights a
0.025mm vibration burst, and B) highlights a 0.5mm vibration burst. The two small panels
present A) 1:1 at 0.025mm and B) 2:1 at 0.5mm ImpCycle responses over one second of
vibration.

648
649
650

Figure 2: Afferent class; Fast Adapting Type I (FAI: 19) and Type II (FAII: 9), Slowly
Adapting Type I (SAI: 14) and Type II (SAII: 10) receptive field locations. Circles
represent receptive field (RF) and are drawn to represent difference in RF size.

651
652
653
654
655
656

Figure 3: The average impulses per cycles of FAI, FAII, SAI, and SAII afferents at each
vibration frequency-amplitude combination in the foot A) and hand B). The foot data is
from the present study while the hand data has been adapted from Johansson et al., 1982.
The legend on the right indicates the magnitude of afferent responses, ranging from <0.5:1
impulses per cycle (dark blues) to >6:1 impulses per cycle (dark reds). Open circles
indicate vibration stimuli that were delivered but did not evoke an afferent firing response.

657

28

658
659
660
661

Figure 4: Average FAI impulses per cycle responses across vibration frequency and
amplitude combinations. 1:1 firing rate is highlighted by a grey line (A). Individual afferent
ImpCycle responses of select FAI afferents across vibration frequency at 1mm (B) and
0.25mm (C) amplitudes. The average of each group is represented by the black stars.

662
663
664
665

Figure 5: Average FAII impulses per cycle responses across vibration frequency and
amplitude combinations. 1:1 firing rate is highlighted by a grey line (A). Individual afferent
ImpCycle responses of select FAII afferents across vibration frequency at 1mm (B) and
0.25mm (C) amplitudes. The average of each group is represented by the black stars.

666
667
668
669

Figure 6: Average SAI impulses per cycle responses across vibration frequency and
amplitude combinations. 1:1 firing rate is highlighted by a grey line (A). Individual afferent
ImpCycle responses of select SAI afferents across vibration frequency at 1mm (B) and
0.25mm (C) amplitudes. The average of each group is represented by the black stars.

670
671
672
673

Figure 7: Average SAII impulses per cycle responses across vibration frequency and
amplitude combinations. 1:1 firing rate is highlighted by a grey line (A). Individual afferent
ImpCycle responses of select SAII afferents across vibration frequency at 1mm (B) and
0.25mm (C) amplitudes. The average of each group is represented by the black stars.

674
675
676
677

Figure 8: Profile plots of entrainment threshold versus frequency, by afferent class.
Entrainment is defined as a ratio of at least 0.9:1, meaning the afferent fires for each sine
wave. Individual afferent data are presented; each dashed line represents a different
afferent. Black stars are the average firing thresholds at each stimulation frequency.

678
679
680

Figure 9: Profile plots of firing threshold versus frequency, by afferent class. Each dashed
line represents a different afferent. Black stars are the average firing thresholds at each
stimulation frequency.

681

29