Beyond the Bottom of the Foot

Topographic Organization of the Foot Dorsum in Walking

Abstract

Introduction 

Sensory feedback from the foot dorsum during walking has only been studied globally by whole nerve stimulation. Stimulating the main nerve innervating the dorsal surface produces a functional stumble corrective response that is phase-dependently modulated. We speculated that effects evoked by activation of discrete skin regions on the foot dorsum would be topographically organized, as with the foot sole.

Methods 

Nonnoxious electrical stimulation was delivered to five discrete locations on the dorsal surface of the foot during treadmill walking. Muscle activity from muscles acting at the ankle, knee, hip, and shoulder were recorded along with ankle, knee, and hip kinematics and kinetic information from forces under the foot. All data were sorted on the basis of stimulus occurrence in 12 step cycle phases, before being averaged together within a phase for subsequent analysis.

Results 

Results reveal dynamic changes in reflex amplitudes and kinematics that are site specific and phase dependent. Most responses from discrete sites on the foot dorsum were seen in the swing phase suggesting function to conform foot trajectory to maintain stability of the moving limb. In general, responses from lateral stimulation differed from medial stimulation, and effects were largest from stimulation at the distal end of the foot at the metatarsals; that is, in anatomical locations where actual impact with an object in the environment is most likely during swing. Responses to stimulation extend to include muscles at the hip and shoulder.

Conclusions 

We reveal that afferent feedback from specific cutaneous locations on the foot dorsum influences stance and swing phase corrective responses. This emphasizes the critical importance of feedback from the entire foot surface in locomotor control and has application for rehabilitation after neurological injury and in footwear development.

Walking requires complex neurological interaction between supraspinal and spinal regulatory mechanisms. In addition to central nervous system processing, afferent feedback regulates rhythmic locomotor activity in accordance with environmental demands. Feedback from muscle and mechanoreceptors, as well as cutaneous and subcutaneous tactile mechanoreceptors, dynamically interacts to sculpt locomotor output (^14,24,35). Although all types of feedback (and their related reflexes) are important, exteroceptive information from cutaneous receptors permeating the skin is particularly important in the fine regulation of walking (³⁵).

In particular, skin from the feet is important for regulating locomotion where information about the environment is obtained. Receptors in the foot sole are required to perceive the ground during locomotion and indicate destabilization of posture producing responses to counteract uneven terrain through ankle joint stabilization (^20,21,34). Furthermore, somatosensory feedback from the foot sole acts in a functional and location-specific manner to maintain stability and balance. In standing, forefoot stimulation caused inhibition in the plantar flexors soleus and medial gastrocnemius (MG), but excitation in dorsiflexor tibialis anterior (TA) and the opposite occurred after heel stimulation (²⁰). In walking, the importance of regional information, obtained by stimulation at five discrete areas on the foot sole, also revealed site-specific topographical organization of cutaneous reflexes (³⁴). Activation under the foot sole produced a kind of tuned “sensory steering” response directed at guiding the foot away from or around a perceived obstacle (³⁴).

The entire foot acts as a sensor antenna, and dorsal skin also plays a role in regulating walking. During the swing phase, rapid corrections of the foot trajectory compensate for a perturbation that obstructs the forward progression of the limb if we hit an unexpected obstacle with the top of the foot. The rapid correction needed to prevent a trip has been termed a “stumbling corrective reaction” (⁹) and has been observed in decerebrate and intact cats as well as human infants and adults (^{3,4,7,8,17,22,25–27,32}). In human adults, the stumble corrective reaction consists of increased knee flexion and a decrease in ankle dorsiflexion during the swing phase allowing for smooth forward progression as the leg is lifted over an obstacle to preserve balance during walking. This response is observed with both mechanical and electrical stimulations to the foot dorsum (^8,25) and is finely regulated by the nervous system in a phase-dependent manner such that the same touch during the stance phase of walking, likely not hindering walking, elicits a smaller response. The response is also amplified and spreads to the arms when the arms are crossed, producing instability, compared with the arms swinging freely (¹²). Thus, responses from dorsal skin stimulation also contribute to the regulation of walking, yet they have a functionally different role from plantar skin stimulation. Topographically organized responses may also be present in this area of the skin which may be functionally different from responses from stimulation to the foot sole. Sensory regulation of human locomotion has typically focused on the bottom of the foot, the surface interacting directly with the ground for propulsion. Currently, there are few corresponding data on the fractionalized function of sensory input from the foot dorsum during walking. Cutaneous responses are dictated by the location of stimulation on the plantar foot surface, but it remains unknown if the same fine tuning of tactile location-specific responses are also found from the foot dorsum. The purpose of the current study was to examine cutaneous reflexes evoked by stimulation at discrete foot dorsum regions during locomotion. In addition, we simultaneously looked to determine the neuromechanical effects of cutaneous inputs from discrete sites during walking. The main hypotheses tested in this study were that 1) stimulation at discrete skin locations on the foot dorsum would evoke topographically discrete cutaneous reflexes during walking and 2) cutaneous reflexes would have mechanical correlates detected as changes in biomechanics of the stimulated leg.

METHODS

Participants

Fifteen neurologically intact volunteers (7 men and 8 women) participated in the study. Participants were 22.8 ± 2.9 yr old, with an average height of 170.8 ± 10.3 cm and weight of 68.2 ± 14.1 kg. Informed, written consent was obtained from all participants before the experiment under a protocol approved by the University of Victoria Human Research Ethics Committee and performed in accordance with the Declaration of Helsinki.

Experimental protocol

To improve procedural standardization and consistency, all participants were fitted with the same make and model of running shoe (Nike Free 4.0 Flyknit) and the same sock (Nike Elite Dri-Fit Sock). Participants walked on a motorized treadmill (Woodway USA, Waukesha, WI) at a self-selected, comfortable walking pace (0.8 ± 0.13 m·s⁻¹) that remained constant throughout the experiment. During each trial, participants walked continuously while electrical stimulation was delivered to one of the five stimulation sites for five separate trials. Background data were collected simultaneously for each trial. The order of presentation for each of the five trials was randomized.

Cutaneous stimulation

Stimulating electrodes (Grass, 10-mm Gold Cup-Disk Electrodes) were filled with electrode conductive gel (Spectra 360 Electrode Gel; Parker Laboratories, Inc., Orange, NJ) and taped onto the five different sites on the dorsal surface of each participant’s right foot with the cathode positioned distally. The five sets of electrode pairs were located at the following: 1) distal end of the first metatarsal (red), 2) distal end of the fourth metatarsal (orange), 3) middle medial skin surface (yellow), 4) middle lateral skin surface (green), and 5) ankle crease (blue). These locations can be seen in each figure legend. Electrodes and lead wires were adhered to the surface of the skin with flexible tape (BSN Medical, Hypafix Tape), and the sock and shoe were then placed over the electrodes. Stimulation was provided by a Digitimer Constant Current High Voltage Stimulator (Model DS7AH) with trains of 5 × 1.0-ms pulses at 300 Hz. During each trial, a total of 160 randomly timed stimulations (1–3 s) were delivered throughout the step cycle.

Immediately before each trial, perceptual threshold (PT) was determined for each stimulation site. PT was defined as the stimulus intensity found to evoke a detectable tactile sensation at the lowest intensity possible. Participants remained standing while stimulation intensities were gradually decreased by the researchers until the participants could barely discern the stimulus (identified as PT). The stimulation intensity delivered to the foot dorsum was set to approximately 3 PT. The stimulation intensity of 3 PT was chosen to evoke a nonnoxious cutaneous sensation during each trial by activating cutaneous afferents immediately under the electrodes and to provide the same relative activation at all stimulation sites. Participants described this as a fluttering, buzzing, or prickly sensation that was strong but not painful.

Electromyography

Once the skin was cleaned with alcohol wipes, disposable surface electromyography (EMG) electrodes were placed on the skin over muscles in the upper and lower leg and shoulder. All EMG recordings were ipsilateral to the site of stimulation (right side). Muscles included TA, MG, peroneus longus (PL), vastus lateralis (VL), biceps femoris (BF), gracilis (GR), and gluteus medius (GM) and from the upper limb muscle posterior deltoid (PD). These muscles were chosen on the basis of previous observations of site-specific effects from discrete foot sole skin stimulation (³⁴). Ground electrodes were placed over electrically neutral tissue at the knee cap and medial olecranon of the elbow. EMG signals were amplified at 5000 times and filtered from 100 to 300 Hz (Grass P511; Astro-Med Grass Inc.).

Biomechanics

Under-foot pressure was recorded with force sensing resistors (FSR) that were firmly attached to the insole of the participant’s right shoe at the heel, head of the first metatarsal, and head of the fifth metatarsal. Validity of FSR has been previously shown (³⁷), and this measure has been previously useful in determining changes in under-foot kinetics (³⁴). Throughout the experiment, force signals were preamplified and recorded for offline analysis. Angular positions of the hip (flexion/extension), knee (flexion/extension), and ankle (inversion/eversion and plantar/dorsiflexion) were measured using electrogoniometers (Biometrics Ltd., Gwent, UK) positioned over the sagittal planes of the ankle, hip, and knee. These devices were calibrated and output in degrees was recorded. As with similar studies (^12,19,32,33), all data were sampled at 1000 Hz with a 12-bit A/D converter connected to a computer running custom-written LabVIEW (National Instruments Corp., Austin, TX) acquisition software and stored for offline analysis.

Data analysis

All data were analyzed with custom-written software (MATLAB; MathWorks, Inc., Natick, MA). Background EMG data were full-wave rectified and low-pass filtered at a cutoff frequency of 100 Hz using a fourth-order Butterworth filter, kinematic data were low-pass filtered at a cutoff frequency of 6 Hz using a fourth-order Butterworth filter, and mechanical data were low-pass filtered at a cutoff frequency of 20 Hz using a fourth-order Butterworth filter. FSR signals from the foot sole were summed and used to establish step cycle parameters; heel contact (HC), periods of stance, toe-off, and swing on the basis of methods previously described (³²). For each step in each condition, EMG and biomechanical data (FSR and kinematic data) were normalized to 100% of the gait cycle and averaged together to create a representative step. For comparisons between participants, EMG data of each muscle from each dorsal stimulation site condition were normalized to the peak average value of that muscle’s activation pattern for that trial for each participant.

To examine cutaneous reflexes, the step cycle was divided into 12 separate equally timed phases, beginning with phase 1 (early stance) at HC, through to phase 7 (swing transition) and phase 8 (early swing), and ending with phase 12 (stance transition) at the subsequent HC (³⁴). Responses for all data for each stimulus condition occurring in the same phase of the step cycle (n = ~10–20 responses per phase) were aligned to stimulus delivery and averaged together. Averages from the same phase of walking during unstimulated cycles (“control” EMG, n = ~50–60 per phase) were then subtracted from each of the corresponding 12 averages after stimulation yielding subtracted traces of reflex EMG and stimulation-induced changes in kinematics. For evaluation of the effect of stimulation, the subtracted data traces were analyzed in all instances.

Cutaneous reflexes were quantified as the average cumulative reflex occurring 150 ms after stimulation (ACRE150). This quantification method may not accurately capture the cases where after stimulation, the EMG is initially inhibited and then excited, but it is useful in determining if overall excitation or inhibition is changed between conditions (³⁰). As with previous studies (^15,31,32), this measure involved calculating a subtracted reflex (see previous discussion) and then cumulatively summing the signal during a poststimulus period of 150 ms and dividing by the time interval of integration to measure an overall reflex effect. Net reflex values were normalized to peak control EMG amplitude. Changes in kinematic and FSR data due to stimulation were analyzed within a 140- to 220-ms window after stimulus (^16,32). Responses were considered significant if they exceeded a 2-SD band of the mean value of the subtracted prestimulus level of the ongoing kinematic or mechanical parameter at each phase. Background activity between sites (pooled across phases) was also compared to inspect for a possible scaling effect on reflex activity.

Statistics

All statistical analyses were completed using SPSS version 18 (Chicago, IL). In all instances, analysis was conducted on averaged data from each part of the step cycle for each participant. Each variable was analyzed separately, as was each phase of walking, to determine if the site of stimulation at a specific location on the foot dorsum and during a particular part of the step cycle had a discernible effect.

To test our hypothesis that stimulation at discrete locations on the foot dorsum would evoke topographically distinct cutaneous reflexes, the initial approach for all data was to conduct omnibus 12 (phase) × 5 (dorsal sites of stimulation) repeated-measures (rm)ANOVA. To address our hypothesis of site dependency of dorsal skin site stimulation throughout walking, using an approach applied in prior work (^18,19), each phase of walking was individually investigated. This was achieved with a phase-specific 1 (phase) × 5 (dorsal sites of stimulation) rmANOVA. This method was also chosen as a follow-up to significant main effects for site, instead of interpreting results from the omnibus rmANOVA, to combat the need for correction from many statistical comparisons. Note that all significant differences indicated on the figures showed a significant 1 × 5 ANOVA main effect. Corrections for multiple tests were handled with Fisher LSD post hoc tests, and results were used to determine site-specific differences.

As a control test, linear regression analysis using Pearson correlation coefficient r was used to determine relationships between reflex amplitudes and background activity levels for each muscle of interest across stimulation sites. A test was considered significant if r was greater than the critical value of r at 0.497.

All statistical tests were two-tailed, and significance was accepted at P < 0.05. Thus, all data described in the following text or shown in the accompanying figures and described as “significant” or indicated with an asterisk were determined as a main effect or interaction from the omnibus and from the phase-specific rmANOVA.

RESULTS

Background data

Figure 1 shows the grand average background data (without stimulation) across all participants across the entire step cycle. From the omnibus rmANOVA, background data for all muscles, kinematics, and kinetics showed no significant main effects of site and no significant interactions (statistical details can be found beneath each variable in Fig. 1). There were also no differences observed from the phase-specific rmANOVA between foot dorsum skin site conditions within each phase. Therefore, differences from stimulation in EMG or biomechanics between dorsal foot sites are independent from changes in background data.

FIGURE 1: Background EMG, kinematic, and kinetic data for variables of interest averaged across all participants for stimulation at five dorsal skin sites shown as a percentage of the walking cycle. The outcomes from interaction (Int) terms and main effects (ME) for site and phase in the omnibus rmANOVA are listed in the panel for all muscles. The cartoon foot diagram containing five pairs of colored dots refers to dorsal foot skin stimulation sites.

Cutaneous reflex modulation

Quantified data for each phase are shown for PL, MG, TA, GR, and PD in Figure 2. In terms of the magnitude of effects, for the PL, MG, TA, GR, and PD muscles, responses to stimulation were 2.13% (±0.50%), 1.72% (±0.33%), 1.70% (±0.42%), 2.33% (±0.53%), and 3.02% (±0.84%) of background activity, respectively. These magnitudes were calculated by determining the ratio between reflex amplitude, measured as the ACRE150 response, averaged across all phases and sites, divided by background activity. There were no differences between the magnitudes of effects across stimulation sites for any of the muscles. From the omnibus rmANOVA, main effects of foot dorsum stimulation site emerged for several of the muscles tested including PL (F_(4,56) = 2.314, P = 0.012), MG (F_(4,56) = 2.405, P = 0.007), TA (F_(4,56) = 2.214, P = 0.042), GR (F_(4,56) = 3.225, P = 0.011), and PD (F_(4,56) = 4.535, P = 0.003). Given that there were no statistical effects of site for VL, BF, and GM, these muscles will not be discussed further and the focus will remain on muscles with significant site-specific main effects. We expected changes in ankle muscle responses given the location of the stimulation sites; however, changes in muscles beyond the ankle are of interest. In several instances, for PL (F_(11,154) = 2.418, P = 0.029), TA (F_(11,154) = 2.146, P = 0.038), and PD (F_(11,154) = 2.040, P = 0.041) muscles, significant main effects of phase are noted by statistical analysis with the omnibus rmANOVA.

FIGURE 2: Average quantified net (ACRE₁₅₀) cutaneous reflexes across all 12 phases of the step cycle for muscles of interest. Data are percentages (±SEM) normalized to maximum background EMG measured across all phases of walking. Negative values indicate overall suppression, and positive values indicate overall facilitation of muscle activity. *Statistical differences, at P < 0.05, between stimulation conditions within a phase. The cartoon foot diagram containing five pairs of colored dots refers to dorsal foot skin stimulation sites.

Main effects of stimulation site were further investigated at each phase, and all statistical details can be found in Table 1. For the muscles that act about the ankle, PL (ankle evertor and plantar flexor), MG (plantar flexor and ankle evertor), and TA (dorsiflexor and ankle invertor), significant location-dependent effects were found mainly at the swing to stance transition and in early swing from the phase-specific rmANOVA. In PL, at the swing transition (phase 7: F_(4,56) = 3.808, P = 0.008) and early swing (phase 8: F_(4,56) = 2.939, P = 0.028), significant effects of site were found. With post hoc testing, it was found that responses were predominately facilitation except at the middle medial skin surface for phase 7 (sites 3–1, P = 0.022; sites 3–2, P = 0.011; sites 3–4, P = 0.014; sites 3–5, P = 0.000) and at the distal end of the fourth metatarsal for phase 8 (sites 2–1, P = 0.008; sites 2–5, P = 0.023; sites 2–4, P = 0.045). In the MG, at the swing transition (phase 7), an effect of site was found (F_(4,56) = 2.599, P = 0.046) where the post hoc test showed that all stimulation sites displayed facilitation except for the ankle crease site (P = 0.035). At early swing (phase 8), an effect of site was also found (F_(4,56) = 4.067, P = 0.006) where stimulation at the lateral skin surface showed increased activation compared with other stimulation sites (sites 4–1, P = 0.001; sites 4–2 P = 0.002; sites 4–5, P = 0.005). For the TA, there was general inhibition evident for swing (phases 8–12), and therefore, there were few phases with a significant effect. At phase 9, however, there was a significant effect of site (F_(4,56) = 2.818, P = 0.033), where post hoc testing revealed that inhibition was largest for the distal end of the first metatarsal skin site compared with all other dorsal skin sites (sites 1–2, P = 0.000; sites 1–3, P = 0.028; sites 1–4, P = 0.002; sites 1–5, P = 0.008).

TABLE 1: Statistical summary of phase-specific analysis for significant effects of stimulation site and list of significant post hoc contrasts.

Stimulation to dorsal skin sites extends beyond muscles at the ankle to also affect the GR and PD muscles. Site dependence was evident for GR, a hip adductor and flexor and a knee flexor, at the end of the swing phase (phase 10: F_(4,56) = 4.593, P = 0.003) and in the stance transition (phase 12: F_(4,56) = 2.759, P = 0.036). At phase 10, medial foot dorsum stimulation, specifically at the first metatarsal and at the middle medial skin surface, caused GR inhibition, whereas lateral dorsal skin site stimulation shows facilitation. Post hoc analysis also revealed that stimulation at the lateral metatarsal showed increased facilitation compared with all other stimulation sites (sites 2–1, P = 0.008; sites 2–3, P = 0.000; sites 2–4, P = 0.023; sites 2–5, P = 0.001). At the stance transition (phase 12), post hoc analysis revealed that stimulation at the first metatarsal produces GR inhibition compared with stimulation at the fourth metatarsal (sites 1–2, P = 0.006) and at the middle medial dorsal skin surface (sites 1–3, P = 0.016). PD muscle, a muscle contributing to shoulder abduction and extension, also showed significant and noteworthy location-dependent modulation at the transitions of walking: at the swing transition (F_(4,56) = 2.735, P = 0.038) and at the stance transition (F_(4,56) = 2.573, P = 0.049). At end stance before the swing phase transition (phase 6), stimulation at the first metatarsal causes increased facilitation compared with other stimulation sites (sites 1–4, P = 0.027; sites 1–5, P = 0.049). At the swing to stance transition (phase 12), inhibition was evoked by stimulation at the first metatarsal compared with stimulation at other sites (sites 1–4, P = 0.031).

Biomechanics

To address our other main hypothesis, reflex-induced changes in biomechanics (kinematics and kinetics) were also investigated for significant location-specific modulation. We expected changes in ankle kinematics and kinetics given the location of the stimulation sites. Changes in ankle kinematics would be expected to be most prominent in the swing phase when the foot is moving freely, and changes in under-foot kinetics would be expected to be most prominent in the stance phase when the foot is in contact with the ground.

Data for ankle plantar/dorsiflexion and ankle inversion/eversion, and for heel, medial toe, and lateral toe forces are plotted in Figure 3. In terms of the magnitude of effects in biomechanical data, for ankle plantar/dorsiflexion, ankle inversion/eversion, heel FSR, medial toe FSR, and lateral toe FSR, responses to stimulation were 1.35% (±0.26%), 3.46% (±1.44%), 0.86% (±0.18%), 1.28% (±0.27%), and 1.48% (±0.31%) of background data, respectively. There were no differences between the magnitudes of effects across stimulation sites for any of the biomechanical data. Omnibus rmANOVA revealed a significant effect of site for ankle inversion/eversion (F_(4,56) = 5.916, P = 0.000) and ankle plantar/dorsiflexion (F_(4,56) = 2.152, P = 0.037). There were no significant main effects in the knee or hip data; therefore, these data are not plotted or discussed further. For FSR data, omnibus rmANOVA revealed a significant effect of site for the heel FSR (F_(4,56) = 2.110, P = 0.038), a main effect of site (F_(4,56) = 2.082, P = 0.042) and phase (F_(11,154) = 2.948, P = 0.007) for the medial FSR, and a main effect of site (F_(4,56) = 2.075, P = 0.047) for the lateral FSR.

FIGURE 3: Stimulation-induced average changes in ankle joint kinematics for plantar flexion/dorsiflexion (P/D) (dorsiflexion = up) and ankle inversion/eversion (I/V) (inversion = up), and in forces under the foot detected by FSR at the heel, medial foot, and lateral foot. Data are percentages (±SEM) normalized to maximum range of motion across all phases of walking. *Statistical differences, at P < 0.05, between stimulation conditions within a phase. The cartoon foot diagram containing five pairs of colored dots refers to dorsal foot skin stimulation sites.

For ankle kinematics, most location-dependent effects were evident in the swing phase. However, for ankle plantar/dorsiflexion, there was a significant effect of site from the phase-specific rmANOVA at the end of stance (phase 6: F_(4,56) = 2.755, P = 0.037). Post hoc analysis revealed that stimulation at the ankle crease caused dorsiflexion, whereas at site 2 (P = 0.008) and site 4 (P = 0.019), sites on the lateral margin of the foot, plantar flexion (by reduced dorsiflexion) was observed. In the swing phase, reduced dorsiflexion is evident for most phases, and therefore, few effects of site from the phase-specific rmANOVA are found. For ankle inversion/eversion, however, stimulation to the dorsal foot surface predominately caused significant effects of stimulation site during the swing phase. There were significant effects of site at phase 7 (F_(4,56) = 3.071, P = 0.023), phase 8 (F_(4,56) = 6.282, P = 0.000), phase 9 (F_(4,56) = 6.216, P = 0.000), phase 10 (F_(4,56) = 6.137, P = 0.000), phase 11 (F_(4,56) = 2.650, P = 0.043), and phase 12 (F_(4,56) = 2.566, P = 0.048). In general, at the swing transition and in early swing, stimulation to the medial dorsal skin surface (at the first metatarsal and in the middle of the foot) reduced inversion, whereas stimulation to the lateral part of the foot (at the fourth metatarsal and in the middle of the foot) increased inversion. For example, post hoc analysis revealed that at phase 8, stimulation at the middle lateral foot increased inversion compared with stimulation at the middle medial foot (sites 3–4, P = 0.000). In another example, at phase 10, stimulation at the first metatarsal caused ankle eversion compared with stimulation at the ankle (sites 1–5, P = 0.000). This trend was seen for phases 8–11 where, in general, stimulation to the medial side of the foot dorsum caused eversion and stimulation to the lateral foot dorsum caused inversion. At the swing to stance transition, all stimulation sites reduced inversion, and at the distal end of the foot, the reduction was largest.

For kinetic data, location-dependent modulation was noteworthy during the stance phase. For the heel FSR, there was a significant effect of site during early stance (phase 3: F_(4,56) = 2.737, P = 0.038). Post hoc analysis revealed that medial stimulation at the first metatarsal (site 1) and medial skin surface (site 3) tended to produce an increase in heel force compared with the decrease in heel force observed with lateral metatarsal (site 2) stimulation and ankle crease (site 5) stimulation (sites 1–2, P = 0.002; sites 3–5, P = 0.004). The medial FSR showed site dependence at early (phase 2: F_(4,56) = 2.967, P = 0.027) and mid to late stance (phase 4; F_(4,56) = 2.787, P = 0.035; phase 5: F_(4,56) = 2.952, P = 0.028; and phase 6: F_(4,56) = 2.759, P = 0.036). Post hoc analysis revealed that at early stance, stimulation at the first metatarsal increased medial FSR pressure compared with ankle crease stimulation (P = 0.026). In mid stance (phase 4), a general reduction in medial FSR force was observed with stimulation at the ankle crease producing a greater reduction in force compared with stimulation at the medial surface (sites 3–5, P = 0.036). At end stance (phases 5 and 6), stimulation at the lateral dorsal skin surface tended to increase medal FSR force compared with stimulation at the medial skin surface and ankle crease. For example, at phase 6, stimulation at the medial skin (site 3) decreased medial toe force compared with stimulation to the lateral skin (site 4) that increased medial toe force (sites 3–4, P = 0.007). For the lateral toe FSR, site dependence of dorsal skin site stimulation was observed in mid stance (phase 3: F_(4,56) = 2.695, P = 0.039) where post hoc analysis revealed that stimulation to the fourth metatarsal produced increased lateral FSR force compared with stimulation at the lateral skin site (P = 0.040). For the swing phase, because the foot is not in contact with the ground, effects of stimulation were not interpreted. However, it can be seen that FSR values do not reduce to zero during the swing phase and this may be due to constant pressure provided by the sock-like shoe design.

Control test with background data and response amplitude

Reflex amplitude is typically uncoupled from rhythmic background EMG amplitude; thus, investigating background EMG levels between conditions allows for comparison of reflex amplitudes that cannot be explained by simple gain scaling with motoneuronal pool excitability. To examine the extent to which reflex amplitudes were related to background activity, Pearson correlation coefficient was calculated between response data (pooled across phases) and background data. Results for all variables, across all dorsal skin stimulation sites, are shown in Table 2. Examination of Table 2 reveals that significant correlation was rarely seen across conditions. Of a total possible of 50 (10 variables, 5 stimulation sites), there were only two significant correlations across dorsal skin stimulation site trials.

TABLE 2: Summary of linear regression analysis between background data and response amplitudes for all variables across dorsal skin stimulation sites.

DISCUSSION

In this article, we examined the site dependency of gait adaptations in response to nonnoxious cutaneous stimulation at five discrete sites on the dorsal foot surface. We evaluated reflex and neuromechanical effects to explore the topographical organization of cutaneous afferents innervating the foot dorsum. Our results indicate that cutaneous inputs from discrete regions on the foot dorsum evoke location-specific reflexes in muscles acting at the ankle that produced changes in ankle kinematics and kinetics. This indicates that the functional control of tactile sensory feedback from all parts of the skin surface of the foot is regionally organized. As with stimulation of whole cutaneous nerves innervating the foot dorsum from the superficial peroneal nerve, changes were found in muscles proximal from the ankle. The results of this study provide important information that increase our understanding on the physiology and biodynamics of how afferent feedback from specific cutaneous locations on the foot dorsum influences the mechanisms involved in locomotor adjustments.

In general, in the stance phase, topographic effects are prominent in forces under the foot, and in the swing phase, topographic effects are prominent in ankle muscles and ankle kinematics. It was found that medial and lateral stimulation evoke increased medial and lateral forces in the stance phase and cause ankle eversion and inversion in the swing phase, respectively. Functionally, these responses serve to tilt, or steer, the foot away from the perturbation.

This information may have potential use in the design of rehabilitation strategies for those with impaired gait, such as in those arising after neurological damage. With a better understanding of how each receptive site on the dorsal foot contributes to locomotion, researchers may be able to harness the intrinsic effects of cutaneous reflexes. Site-specific stimulation, at a targeted phase of walking, may be applied to aid in enhancing functional modulation of muscle activity. This study provides a better understanding of the behavioral relevancy and potential rehabilitative use of cutaneous input from specific regions on the dorsal foot during locomotion.

Topographical organization in each phase of locomotion

Interpretations of the integrated neuromechanical responses from stimulation at discrete foot dorsum regions are organized within the functional phases of walking: stance, stance to swing transition, swing, and swing to stance transition. Figure 4 includes a diagram of the functional effects of dorsal skin surface stimulation during the stance and swing phases of walking. In some cases, the effects of stimulation are proposed, because responses from localized stimulation were not always substantial enough to be able to quantify their functionality.

FIGURE 4: Functional diagram of effects of foot dorsum skin surface stimulation. A and B, Functional effects in the stance and swing phases, respectively. The foot outline on the left of each panel shows the overall functional effect of stimulation. The foot outline on the right of each panel shows the location-specific functional effects of stimulation. For stance (panel A), the overall functional effect is downward motion of the foot from increased ankle extensor muscle activity, increased plantarflexion, and increased pressure at all FSR. The location-specific functional effects are that lateral stimulation increases downward lateral FSR pressure (downward arrow) and medial stimulation increases downward medial FSR pressure (downward arrow). For the swing phase (panel B), the overall function effect is downward foot motion from TA inhibition and ankle extensor facilitation, and an increase in ankle plantarflexion. The location-specific functional effects revealed an increase in ankle inversion with lateral stimulation (orange and green arrows) and an increase in ankle eversion with medial stimulation (red arrow).

Stance phase

Generally, during the stance phase, responses to superficial peroneal nerve stimulation are small or absent in ankle muscles, with no response evoked in TA muscle (^7,27,32). For the responses that are present, a general extensor response is observed (^1,4,5,32,36). In the data presented here, from recordings of ankle extensor muscles, similar results were found where an extensor response in ankle muscles PL and MG was observed. It was also observed that stimulation mainly caused inversion of the ankle except during early stance where stimulation to the medial side of the foot reduced inversion. This could potentially serve to quickly place the foot on the ground for safe weight acceptance. At the end of the stance phase, before the transition into the swing phase, stimulation on the lateral side of the foot decreased dorsiflexion. This is consistent with previous observations of plantar flexion facilitation generally evoked with superficial peroneal nerve stimulation at the end of the stance phase (²⁷).

Stimulation at discrete dorsal skin foot sites showed site dependency for force produced at the heel, medial FSR and lateral FSR. In general, stimulation to the lateral side of the foot increased lateral force and stimulation to the medial side of the foot increased medial force (see Fig. 4). This can be seen, for example, at phase 3 in the heel and lateral toe FSR. This is opposite to the effect seen with stimulation to the plantar surface of the foot where stimulation to the lateral foot margin tended to decrease lateral FSR force and stimulation to the medial foot margin decreased medial FSR force (³⁴). The difference in force patterns observed between plantar and dorsal surface skin stimulation was not surprising given the functional differences expected with stimulation to these regions. With lateral stimulation, for example, the decrease in lateral force observed with plantar stimulation is needed to remove weight from the stimulus, and conversely, the increase in lateral force observed with stimulation to the dorsal surface is needed to tilt the foot away from the stimulus.

In previous experiments, superficial peroneal nerve stimulation also significantly increases activity in the BF and semitendinosus producing knee flexion, which could be incorporated to facilitate flexor activity at the end of the stance phase, aiding the transition into the swing phase (⁶). We therefore speculate a similar effect with these data; however, this was not seen because there were no significant reflex effects evoked in muscles around the knee and no effects on knee kinematics. Perhaps, this occurred because stimulation was not substantial enough to require a whole leg correction. Prior work suggested an intensity-dependent regulation of cutaneous effects during walking (³⁶).

We also found interlimb effects, evaluated with responses obtained from the PD muscle acting at the shoulder, where stimulation to the foot dorsum facilitated muscle activity. Functionally, this would cause abduction of the arms to maintain stability to compensate for the potential destabilization from dorsal foot stimulation. At the end of the stance phase, a location-specific effect was observed where stimulation to the first metatarsal further increased PD facilitation for balance maintenance before the unstable transition into swing. In another study, a large facilitatory reflex in the PD muscle in the shoulder was also found, likely produced to stabilize the upper body as the leg leaves the ground (¹¹).

Swing transition

The dominant location-specific effects of stimulation applied during the swing transition were changes in PL and MG muscle activity. As seen in past experiments, at the stance to swing transition, stimulation of the superficial peroneal nerve in the foot increased MG facilitation, which then produces plantar flexion and forward propulsion into the swing phase (¹¹). Here the general response in MG is also facilitation. However, when stimulation is applied specifically to the ankle, facilitation is reduced. Functionally, this may reduce plantar flexion to avoid driving the foot into the destabilizing perturbation. Also showing a location-specific effect, activity in PL muscle is facilitated except when stimulation is applied to middle medial foot dorsum. Functionally, it is speculated that contact on the medial aspect of the foot may require reduced plantar flexion and eversion, by a decrease in activity in PL muscle, to slow the initiation of the swinging leg and reduce push-off force.

Swing phase and stance transition

Stimulation to the foot dorsum had the largest effect on muscle activity and ankle kinematics during the swing phase. During this phase, the swinging foot is particularly vulnerable to encountering an unexpected obstacle that may cause a fall, thus necessitating flexible control (³⁵). Location-specific effects were seen for the TA muscle. Here, TA muscle activity shows predominately inhibition after foot dorsum stimulation, and when stimulation is applied to the first metatarsal, inhibition is largest compared with all other stimulation sites. In past experiments, stimulation to the superficial peroneal nerve, innervating the foot dorsum, also causes inhibition of TA muscle activity, as a defining feature of a stumbling corrective response (^13,27,29,32).

Correlated with this reduction in TA activity, ankle dorsiflexion is generally reduced with superficial peroneal nerve stimulation (^27,32). This is similar to results here because dorsal skin stimulation also generally caused reduced ankle dorsiflexion. However, at the end of the swing phase, stimulation at the fourth metatarsal increased dorsiflexion, revealing topographical organization. The general reduction in dorsiflexion, seen at most phases in swing for most sites, may not be safe at this phase of the gait cycle with stimulation on this part of the foot, as the heel prepares to strike the ground. Contact at this location of the foot at this phase could increase the risk of ankle inversion causing ankle sprain.

For the plantar flexor muscles, PL and MG, dorsal foot stimulation generally increased activity. However, in early swing, PL muscle activity was reduced with stimulation at the fourth metatarsal. Functionally, this may be helpful in reducing plantar flexion and push-off force to avoid pushing the swinging foot into a perturbation.

For ankle inversion/eversion, there were many significant effects during the swing phase. Stimulation to the first metatarsal caused reduced inversion at all points in the swing phase, especially compared with sites on the lateral side of the foot. Conversely, stimulation to the lateral side of the foot increased ankle inversion when compared with medial stimulation sites. Functionally, this response serves to tilt the foot away from the perturbation where lateral stimulation inverts the foot and medial stimulation everts the foot (see Fig. 4).

Swing phase stimulation effects were also seen in GR muscle. In early swing, general inhibition is seen, which functionally may decrease hip flexion to slow the swinging leg as it enters the swing phase to avoid a perturbation. Increased hip extension after superficial peroneal nerve stimulation has been previously documented by a facilitation of the BF and semitendinosus (⁶). In mid swing, stimulation to the lateral side of the foot increases GR activity to perhaps promote hip and knee flexion, whereas stimulation to the medial side of the foot reduced GR activity to decrease hip and knee flexion. Location-specific effects were also seen for the GR at the stance transition, where stimulation to the first metatarsal inhibited activity compared with facilitation seen from stimulation at the other sites. Functionally, this could translate into reduced hip and knee flexion, which would allow the foot to get to the ground quickly to avoid a potentially dangerous perturbation at the toe to maintain balance.

As a means of inspecting interlimb effects, the PD muscle was investigated. It was found that stimulation to the first metatarsal also had a significant inhibitory effect compared with other stimulation sites where PD muscle is facilitated. At this point in the gait cycle, it could be dangerous to lift the arms, causing destabilization, as the body transitions into stance.

Topographic organization of responses from discrete activation of the foot dorsum during walking

Location-specific effects are a feature of cutaneous reflexes during locomotion in humans (²⁷). Increased resolution of effects can be seen here where stimulation to discrete foot dorsum skin sites reveal topographic organization of responses in muscle activity and in ankle kinematics and kinetics. The swing phase is considerably more unstable compared with the stance phase of walking, when the foot is safely on the ground without much potential for adjustment; therefore, responses to dorsal foot stimulation were expected to be more prominent during this time. Indeed, there were increased responses to stimulation from discrete foot dorsum skin site stimulation in the swing phase compared with the stance phase. Swing phase disturbances to the dorsal surface of the foot resulted in a response to withdraw the foot away from the perturbation. In general, responses from lateral stimulation differ from medial stimulation and effects seem most prominent from stimulation at the distal end of the foot at the first and fourth metatarsals. Given the flexible nature of the swing phase, responses to stimulation go beyond the muscles that control the ankle to include muscles at the hip and at the shoulder. This was expected given the strong interlimb reflexes evoked in muscles across the body by cutaneous stimulation of the superficial peroneal nerve (^11,30).

Our control test, investigating relationships between reflex amplitudes and background activity levels, allows for confirmation that reflex effects cannot be explained by simple gain scaling of motoneuronal pool excitability. We found only two significant correlations (out of 50 tests) in our analysis; therefore, changes in background EMG, kinematic, or kinetic activity cannot be implicated as a significant source of modulation for cutaneous reflexes.

Neuronal networks underlying corrective responses from foot dorsum stimulation

The neuronal networks underlying the stumbling corrective and preventive reactions from foot dorsum stimulation are expected to lie within the spinal cord. Cutaneous afferents are especially important for evoking a corrective stumble reaction because the normal response is abolished by anesthesia of the skin of the foot (^10,26). In a fictive cat model, where supraspinal and afferent input is removed, excitatory and inhibitory postsynaptic potentials, reminiscent of the stumble correction response, are recorded after stimulation of the superficial peroneal nerve (²³). Human infants also demonstrate well-organized responses to touch of the foot dorsum during the swing phase. These responses appear before the full maturation of descending pathways from supraspinal centers (^17,28). In adults, observations of reflex effects, as seen here, causing brief kinematic changes also support the notion that stumbling corrective reflexes are evoked mainly by activation of cutaneous afferents in the superficial peroneal nerve and integrated within the spinal cord.

With the results presented here, it can be seen that activation of these presumably spinal pathways continues with activation of local populations of afferents, as opposed to whole nerve trunks (as in superficial peroneal nerve stimulation). The number of afferents activated and subtypes recruited could differ at each site, and it is not fully known how each type of afferent influences muscle activity, although it has been shown that individual classes of cutaneous afferents may serve different functional roles (²). The distance between electrodes may also be a limitation where the electrodes may not have been placed far enough apart to warrant different reflex effects. It is also possible that activation of these afferents might signal different skin stretch or compression requiring interpretation of the central nervous system.

CONCLUSIONS

The results of this study further support suggestions that cutaneous nerves of the foot dorsum produce location-specific responses that are highly organized and produce topographic reflex effects. Both site and phase dependences were observed in the muscular, kinematic, and kinetic responses to discrete foot dorsum stimulation, and these responses serve to maintain ongoing locomotion. This information is of importance to increase our understanding of how afferent feedback from specific cutaneous locations on the foot dorsum influences the mechanisms involved in stance and swing phase corrective responses. Our data show that it is time to move beyond the bottom of the foot and consider the entire foot surface as a true sensory antenna. This has rehabilitative application during locomotor recovery after a neurological injury and can help inform the development of more effective footwear and assistive devices.

The authors would like to acknowledge the critical assistance of all the research participants.

This research was supported by funding to E. P. Z. from the Natural Sciences and Engineering Research Council of Canada and NIKE Inc. A portion of the funding for this project was obtained from a research contract-for-hire from NIKE Inc. One author (E. P. Z.) has worked in the capacity as consultant for NIKE Inc., and B.M. and N.F. are employees of NIKE Inc. The authors further certify and declare that none of these competing interests had any impact on the analysis, interpretation of results, or conclusions derived within the article.

Results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation, and results do not constitute endorsement by the American College of Sports Medicine.

E. P. Z., T. K., N. F., and B. M. conceived the experiment and contributed to the experimental design. Data collection, analysis, and creation of figures were completed by T. K., G. P., Y. S., T. B., and C. K. The article was prepared by T. K., and the final draft of the article was edited by G. P., T. B., E. P. Z., N. F., and B. M.