Share this article on:

Mechanical inputs related to perception of lower extremity impact loading severity


Medicine & Science in Sports & Exercise: January 1998 - Volume 30 - Issue 1 - p 136-143
Applied Sciences: Biodynamics

Purpose: Human responses to repetitive locomotor loadings are likely dependent upon the perceived severity of the impact. Few researchers have attempted to identify the mechanical variables upon which perception of impact severity is based. This study examined the relationship of selected impact loading variables to the perception of impact severity by employing an established psychophysical test procedure.

Methods: A human pendulum apparatus was used to administer and measure impact loadings similar to those encountered during running. Nineteen subjects experienced over 100 right foot impacts which comprised nine different impact conditions presented in a random manner. The conditions represented combinations of three impact velocities and three interface materials covering a force platform.

Results: Group mean subjective ratings of impact severity were highly related to all measured biomechanical descriptors of impact severity. The variables of impact force rate of loading (FRA) and peak shank acceleration had correlation coefficients of 0.99 with perceived severity. When all individual results were combined to determine the relationship of impact loading variables to perception, correlations were generally 0.7 or above with FRA alone explaining 64% of the perceptual rating variability.

Conclusions: These results indicate that impact perception was highly associated with the mechanical input variables commonly measured and that midsole materials such as those typically found in athletic footwear do not remove our ability to perceive the severity of impact loads.

Liverpool John Moores University, UK; and Nike Sport Research Laboratory, Nike Inc., Beaverton, OR

Submitted for publication August 1995.

Accepted for publication October 1997.

Despite the extensive research interest in the reduction of impact-related injuries, it is still unclear as to which mechanical input factors are used by the body to sense the magnitude of impact. The perceived severity of foot-ground impact is likely to be an important factor in the adaptive response of the body to protect against chronic overloading and high impact shocks during locomotion (24). Responses of the body to reduce these loads rely to a large extent on the ability to recognize or perceive their existence (1,5,25). Nevertheless, few researchers have attempted to identify the mechanical variables upon which this perception is based. An inter-disciplinary approach accounting for biomechanical and cognitive characteristics is clearly needed to understand further the response of the body to repetitive impact loading.

Two studies have examined the relationship between various impact loading variables and the perception of either shoe cushioning or impact severity during running in footwear with midsoles of differing densities(14,20). Among the biomechanical variables measured, impact force rate of loading and impact force median power frequency demonstrated moderate-to-high correlation with the perception ratings in both studies. Peak impact force and peak shank shock displayed poor relationships to perception in those studies. Whittle et al. (33) related the perceived comfort of walking on carpets to the cushioning properties of the carpets as determined from mechanical impact testing. These authors found that comfort was best related to the response of a carpet to transient forces and argued that pressure distribution across the sole of the foot might be important for the perception of comfort during walking. The importance of pressure was confirmed by Chen et al. (7) who demonstrated that pressure distribution between the plantar surface of the foot and shoe were related to the perceived comfort of athletic shoes during walking. These authors found that peak pressure and pressure-time-integral were more sensitive to the change of the comfort conditions than maximal force and force time integral. In contrast, Jordan and Bartlett(15) reported that differences in perceived plantar comfort between three types of footwear were not related to any in-shoe plantar pressure distribution parameters (including peak pressure and pressure-time-integral). Changes in temporal load distribution under the foot have been observed during running in response to footwear that had modifications in midsole hardness and in which cushioning was perceived very differently (14).

The phase of impact during which high loads are experienced is very brief(less than 50 ms), and the loading characteristics must be determined by the initial conditions such as impact velocity, limb posture, preprogrammed muscular activity, and the material properties of the contacting elements(1,2). The lack of a clear relationship between shoe hardness and initial impact loading has been attributed to imprecise recording of the impact event (27) and, more recently, to kinematic and neuromuscular adjustments of the locomotor system to the expected severity of the impact (10,11). Adaptive lower limb kinematic adjustments have been demonstrated when changing from shod to barefoot locomotion (11) and during locomotion in footwear of different midsole densities(8,11,14). Assuming that perception of impact severity can be affected by posture of the foot and lower leg at impact, then it becomes difficult to determine whether the reported perceptual changes were the result of changes in lower limb posture or mechanical inputs or a combination of both when the foot contacted the ground. Isolation and manipulation of mechanical inputs to the body are difficult because of the confounding kinematic adaptation. This adaptation may partly explain why the understanding of impact mechanics during locomotion has not progressed as rapidly as can be expected from over 20 years of research attention from biomechanists. To identify the mechanical basis for the perception of impact severity an alternative controlled in vivo approach is needed to systematically manipulate mechanical inputs to the body. Once the important factors have been established then experimentation can move back to locomotion and examine the adaptation or response to perceived loading.

In vivo techniques to modify initial impact conditions in a controlled and systematic manner have used instrumented missiles that impact the foot in either a drop (16) or pendular manner(6). However, the loads did not have the same temporal characteristics as those typically experienced during locomotion. In addition, they failed to reach the same magnitude without pain to the subject. A recent human pendulum approach uses the subject lying supine on a bed as the missile that impacts a wall-mounted force platform. Along with easy control and manipulation of initial impact conditions, the human pendulum imparts loads that are similar to those produced during locomotion(17). This technique can be used to systematically change and quantify the mechanical inputs to the body in vivo.

The relationship between psychological magnitude and stimulus intensity can be inferred from the observable judgments of the subject. Yet subjective judgments can be easily biased. Perception is a dynamic process that depends not only on the current stimulus but on the expected stimuli and on the recent and not so recent stimuli that have occurred (4). Bias may be created by conditions within the experiment or may be a product of the subject's past history. With that knowledge, psychophysical procedures have been developed so that individuals can assess their impressions and communicate them to the experimenter with as few biasing cues, suggestions, and constraints as possible. The magnitude estimation approach(12,28-31) has been the most common procedure employed by psychophysicists over the last 40 years. Stevens(28) suggested that, if properly used, the method of magnitude estimation could provide a simple, direct means of determining a scale of subjective magnitude. This psychophysical technique was used in combination with the human pendulum apparatus to explore the relationship of mechanical inputs from lower extremity impacts to the perception of impact severity. Specifically, this study examined whether perception of impact severity could be associated with the commonly measured biomechanical variables describing impact loading.

Back to Top | Article Outline


Nineteen male subjects volunteered to participate in the present study. No subject had a history of severe musculoskeletal injury. Each subject declared himself to be free of lower extremity and low back discomfort or injury. The subjects' mean mass, height, and age were 75.9 ± 7.7 kg, 1.77 ± 0.08 m and 24.4 ± 3.3 yr, respectively. Before any experimental testing in the laboratory, subjects were required to complete an informed consent form.

Human pendulum apparatus. Impacts to the right foot of each subject were imparted and quantified using the recently introduced human pendulum approach (17). In brief, the system consisted of a lightweight bed that was suspended by four chains from a high ceiling and acted as a swinging pendulum. Each subject laid supine on the bed with the right lower limb extending over the edge of the bed closest to the impacting wall (Fig. 1). The position of each subject on the bed was adjusted such that the right foot barely touched a Kistler force platform(Model #9281B11) or layers of foam that covered it. The left leg was flexed with the foot resting against the bedding PVC frame. Different impact velocities were generated by pulling the bed set distances away from the platform. The right knee of each subject was held in 20° of flexion before impact by padded straps at the ends of two thin steel cables suspended from above. The straps were adjustable in height and secured to the proximal and distal ends of the subjects' shank (above the ankle and below the knee). This dual suspension system maintained a perpendicular orientation of the shank to the force platform as well as fixing its position relative to the pendulum. The 20° of knee flexion was selected as the initial knee angle because it represented the typical knee angle at foot contact during running(22,34). Each subject was instructed to maintain a slightly dorsiflexed (less than 5°) ankle that ensured a heel first contact. After impact with the wall-mounted force platform, they were also instructed to actively resist the forward motion of the pendulum but not to subsequently push off the wall. Impact conditions were modified by changing the impact velocity and the material characteristics of 2.5-cm thick layers of EVA foam covering the force platform. The human pendulum allowed a tight control of impact velocity and knee angle at impact (20°). It also permitted different impact conditions to be administered in a random manner.

Measurement of psychological magnitude. Measurements of psychological magnitude have sources of potential bias that must be minimized in the experimental design. Stevens (28,29) pioneered the first work in the ratio scaling of psychological magnitudes and originated the magnitude estimation procedure. This study incorporates some of his recommendations that were deemed important for reducing bias. Subjects were completely free to assign any numerical value (excluding zero or negative numbers) to a particular perceived severity of the impact. A specific standard or “anchor” impact condition had a numerical value assigned by the experimenter, and the subject used this as a reference point in assigning numerical values to the other test impacts subsequently presented. Thus, each impact had a magnitude assigned to it by the subject. The assumption in this method is that numerical judgments are directly proportional to sensory magnitude. Work on the additivity of measurements seems to indicate that, at least for data averaged over several subjects, the assumption is correct(4). The standard impact condition was referred to as a number, 10, that was easily multiplied and divided. It was in the middle of the range of conditions administered so that it would not impress as being either extremely soft or hard. This selection enables subjects to more easily“take hold of” the standard (28). The range of impact conditions was not large and they were presented in a mixed order. All impact testing was performed in silence to facilitate the subjects' concentration of perceived severity.

Experimental procedure. Before pendulum testing, subjects were familiarized with the pendulum impacts. They were given 20 practice impacts at the standard reference condition and also presented with impacts at the extremes of the range of impact conditions. This provided them with an overview of the impacts to expect during data collection. Subjects then underwent nine sets of 12 right foot impacts (108 total) during a single visit to the laboratory. Each set comprised three impacts at the constant“standard” condition followed by nine different “test” impact conditions. The test conditions represented all combinations of three impact velocities (0.9, 1.05, and 1.2 m·s-1) and three interface materials (65, 55, and 40 on Asker C) covering the wall-mounted force platform. The three impact velocities were measured before subject testing with a Celesco velocity transducer (DV-301-0075). The mixed order of impact condition presentation was determined using a Latin square design. This ensured that each of the nine impact conditions was administered nine times to each subject for averaging purposes. The standard condition was always comprised of the 1.05 m·s-1 velocity and 55 Asker C material. Each subject was given the same explicit instructions for the rating of perception. They were told that the standard was given a rating of 10 and to assign any numerical value (>zero) to the perceived severity of impact using the sfandard as a reference. They were also informed that impact conditions would vary randomly. They were asked to focus on the perceived severity of each impact. Within each set there were approximately 6 s between impacts during which subjects informed the experimenter of their estimate of impact severity. Subjects had 3 min rest between sets while impact data were processed.

Measurements. External impact loadings that resulted from the contact between the subject's foot and the force platform interfaces were measured with the wall-mounted force platform. Pressure on the plantar surface of the heel was measured using eight discrete sensors (H.A.L.M, Inc., Germany) encapsulated in silicon rubber and attached firmly to the bare heel. The encapsulation of sensors in this manner has been shown to minimize the distortion of pressure caused by point loading artifacts(18). The shock transmitted to the shank was captured with a uniaxial accelerometer (Entran EGA-100D). This transducer was fixed to a rectangular piece of balsa wood (20 × 70 × 6 mm) which was then mounted onto the skin overlying the antero-medial aspect of the tibia 15 cm proximal to the medial malleolus. The balsa wood was affixed to the shank with acrylic glue and medical tape (32). The accelerometer sensitive axis was visually oriented parallel to the shaft of the tibia. Head shock was monitored using a biaxial accelerometer (Entran EGA2-C-50D) mounted on a bite bar that was firmly gripped between the teeth. In this position the sensitive axes of the transducer were approximately along the axial(cranio-caudal) and antero-posterior directions. The resultant was used to quantify the shock reaching the head. Electromyographic (EMG) activities of gastrocnemius, vastus medialis, rectus femoris, and biceps femoris lower extremity muscles were recorded using silver-silver chloride electrodes, after thorough preparation of the skin. EMG signals were full-wave rectified and integrated for the 200 ms immediately before each impact. These integrated EMG values provided a representation of precontact muscle activity and neuromuscular preparation for the impact (5,19). All analog signals were sampled simultaneously at 1500 Hz and all except the EMG signals were subjected to a fourth order Butterworth low pass filter with a 200 Hz cut-off.

Variables measured. The mechanical input variables measured were time-to-peak impact force (TPKF), and the peaks and transient rates of impact force (PKF and FRA), heel plantar pressure (PKP and PRA), shank acceleration(PKS and SRA), and head acceleration (PKH and HRA). Transient rates were calculated for the period that the signals varied between 10 and 90% of their initial peak values. For each subject, these variables were scaled to the mean value recorded for the standard condition. Perceptual rating values were divided by ten so that they were also scaled to the standard condition in the same manner (i.e., both biomechanical variables and perceptual ratings at the standard condition were scaled to one for each subject). This scaling procedure enabled the relationship between biomechanical variables and perception to be examined after allowing for individual differences in mechanical input variables at each impact condition. Perceptual differences were examined using a two-way repeated ANOVA for correlated means and subsequent multiple comparisons (Newman Keuls) to evaluate impact velocity and interface effects (α = 0.01). The relationship between biomechanical input variables and perceived impact severity was examined using multiple correlation analyses on three different levels. Initially, group mean values for the biomechanical variables were calculated for each of the nine impact conditions. These group mean values were correlated against the group mean perceptual rating for each impact condition. This evaluation of the relationship to perception using nine data points per biomechanical variable was referred to as the GROUP correlation. The next level of correlation was calculated by using all the mean subject-condition values of perceptual ratings and mechanical input variables. The relationship used 171 data points(9 * 19) per biomechanical variable and was referred to as the ALL correlation. Finally, biomechanical variables and perceived impact severity ratings were correlated for each individual (within-subject correlation with nine data points per biomechanical variable). For comparison with the above correlations, the within-subject correlations were averaged across all subjects and termed the INDV correlation.

Back to Top | Article Outline


Over the nine test conditions, peak impact force typically ranged between one and two body weight units (BW) in magnitude. Some subjects experienced peak forces of up to 2.5 BW. The mechanical input variables describing impact loading were modified between 0.25 and 1.65 times their mean values for the human pendulum standard condition (1.05 m·s-1 impact velocity and 55 Asker C interface material). The standard condition generated a mean impact force of 1.53 BW which occurred on average 21.5 ms after contact; there was a corresponding peak shank shock of 11.5 g at 20.6 ms. The transient rate variables demonstrated the greatest change over the range of conditions(>200%). Ratings of perceived impact severity ranged between 1 and 26 for the different impact conditions. On average, the within-subject coefficient of variability in perception ratings for a given impact condition was about 20%.

The ANOVA revealed that there was a significant interaction between velocity and interface material effects on perception ratings. In general, perceived impact severity significantly increased alongside increases in impact velocity and changes from soft to medium or hard interface. Post hoc means comparisons revealed that perception ratings were not significantly different between the 1.05 and 1.2 m·s-1 impact velocities on the soft (less dense) interface. The only other nonsignificant comparison was between medium and hard interface material at the slowest impact velocity (0.9 m·s-1). Perceived impact severity differences among the three impact velocities were more evident as the interface material density increased and there was a tendency for interface effects on impact perception to be larger at the greatest impact velocity (1.2 m·s-1).

For all three levels of multiple correlation analysis every mechanical input variable was significantly correlated with perceived impact severity rating (Table 1). The group mean data for several of the variables (FRA, PKS, SRA, and HRA) demonstrated a correlation coefficient of 0.99 with group mean perception rating. The lowest correlation seen inTable 1 for this group mean data was -0.77 for TPKF. Almost every mechanical input variable displayed a correlation of 0.7 or greater when the mean subject-condition data (ALL) were related to corresponding mean perceived severity ratings (Table 1). The best relation to perception for the subject-condition data was found for impact force rate of loading (FRA), which explained 64% of the variability in perception ratings. Using a stepwise linear regression procedure a combination of FRA, PKH, and PKP only accounted for an additional 2% (66%) in perception variability. This indicates that FRA alone provides the most parsimonious model. In addition to being highly related to perception, the mean subject-condition biomechanical data was also significantly correlated with each other (Table 2). In particular, high correlations were found between peak impact force and peak head acceleration (0.94) and between force rate of loading and heel plantar pressure rate of loading(0.96). The lowest correlation was found between peak force and time to peak force (-0.45).

For each individual, the correlations between perceptual ratings and all input variables were on average 0.86 ± 0.06. The mean correlation between perception and the nine mechanical variables was 0.87 or higher in 13 of the 19 subjects. The specific input variables that were best related to perception at the individual level varied between subjects. However, among the individual responses two trends were noticed. Peak impact force was either moderately or highly correlated with perception of impact severity. In every subject the transients rates of force, shank, and head acceleration were generally the better predictors of perception. This tendency is illustrated inTable 1 where the transient rate variables have the highest mean within-subject correlations of 0.9.

Although all the input variables were significantly related to perceived impact severity, the group mean results demonstrated that specific variables were more closely related to perception. This was exemplified by plotting mean perception ratings against mean values for peak impact force (PKF) and force rate of loading (FRA). For the soft interface condition at every impact velocity, the force rate of loading follows perception ratings much better than peak impact force (Fig. 2). At the greatest velocity, FRA was also closer to the perception rating at the hardest interface condition.

The relationships of input variables to perception generally increased with severity of impact loading conditions. Correlation coefficients for mean subject-condition data of each variable were calculated at each impact velocity (across interface materials). For every input variable the lowest correlations were seen at the slowest impact velocity (Fig. 3). It can also be seen from Figure 3 that, when the correlations were determined at each impact velocity separately, PKF and PKH displayed a poorer relationship with perception.

Back to Top | Article Outline


This study was designed to explore the relationship between commonly measured biomechanical variables used to quantify impact loading and the perception of impact severity. The results indicated that the perceived magnitude of impact loading was highly related to the mechanical inputs that were measured. The human pendulum modality allows the control of initial impact conditions while imparting loads similar to those experienced during locomotion. The various mechanical input variables selected for measurement were modified systematically by changing either impact velocity or impacting interface material. The interfaces consisted of shoe midsole materials and reflected the extremes of the range of midsole density found in athletic footwear. High correlations, particularly at the individual level, were found between all mechanical input variables and perceived severity of impact. The main limitation with the present research is that application of the results to actual locomotion should be done with care because the human pendulum does not reproduce all initial locomotor heel strike conditions. In addition, day-to-day variability in the relationship between perception and impact loading severity was beyond the scope of the present investigation. Further work needs to be done to examine comprehensively the strength of the relationship on a day-to-day and individual basis.

The high correlations demonstrated between mechanical input and perceived severity is only in partial agreement with the results of Robbins et al.(26). These authors also attempted to simulate impact loading during running in a controlled in vivo test environment. They found a strong linear relationship between perceived magnitude and load applied to the lower leg. However, after extrapolation of their group mean data to typical loads that are experienced during running, they postulated that athletic footwear substantially attenuated the perception of loads so that perceived severity is less than actual severity. In contrast, this study found a close relationship between perceived severity and actual severity at a range of impact conditions that were very similar to those experienced during running. This contrasting finding may be partly explained by the test modality used in the study by Robbins et al. (26). Those authors attempted to simulate the impact of locomotion by having a pneumatic cylinder apply loads to the flexed knees (90°) of seated subjects. The cylinder loads were applied for approximately 1 s and then removed. The duration of imparted loads using the human pendulum (220 ms) were identical to those typically experienced during running (21) and the pendulum loads closely simulated the time history during the initial impact phase of running (17). In addition to providing a closer simulation of the magnitude and temporal characteristics of locomotor-like loading, the present study mimicked the angle of the knee at ground contact during locomotion (20°) and isolated the mechanical input stimuli to those generated only at the foot.

The correlations using group mean data were substantially higher than group data correlations obtained when perceptual ratings were related to similar mechanical impact variables during running(14,20). Similar to the present study, one of the variables best correlated to perception of impact severity or cushioning in those studies was force rate of loading (FRA), which demonstrated coefficients of 0.34 (14) and 0.76 (20). In contrast, the correlation between group mean FRA and perception data was 0.99 in this study. The control of initial impact conditions and the isolation of mechanical inputs likely contribute to the higher relations found using the pendulum modality. The three experimental footwear conditions used in the study by Hennig et al. (14) had extreme differences in midsole density. By recording in-shoe plantar pressures, these authors found that the different footwear induced the subjects to modify the loading distribution under their feet and likely their landing kinematics during running. This adaptation of initial conditions may have confounded the relation between footwear cushioning properties and perceived impact severity because the adjustments may serve to move perceived `hardness' to a different or optimal level (9).

Although the close relationship of the mechanical input variables to perception ratings indicates that the severity of impact loads on the body can be perceived, the sensitivity of this perception appears to have limitations. For the less severe impact conditions, small changes in conditions were not readily “sensed” by the subjects. Subjects were not able to differentiate between impacts on the medium and high density interface at the slowest impact velocity or between all impact velocity conditions on the low density interface material. For these particular adjustments in initial impact conditions, the changes in the mechanical input variables were relatively small. The difference in material properties between the medium and high density interface (10 on Asker C scale) was not as large as the difference between low and medium density (15 on Asker C scale). This would explain why changes in impact perception and mechanical input variables were much less between the medium and hard interface. Nevertheless, there were small changes in impact severity that were not perceived differently by the subjects. Another factor that may partly contribute to the limited perception sensitivity is the frequency content of the impact experienced. For both impact force and shank shock, the signal power of frequencies above 20 Hz decreased substantially when the high and medium density interfaces were replaced with the low density interface. There are specific mechanoreceptors(Pacinian corpuscles) that respond preferentially to high frequency mechanical stimuli in excess of about 40 Hz (13). The Pacinian corpuscles are abundant in the fat pad under the heel (3) and may be involved in the sensation of impact transients. It can be hypothesized that because these receptors are tuned to high frequencies, it may be more difficult to perceive the severity of impact conditions that generate much lower signal power above 40 Hz (e.g., all impact conditions on the low density surface).

At the greater impact velocities subjects were able to perceive differences in the severity of impacts on the medium and hard interface. The main reason for this apparent improved sensitivity is that the differences in impact severity between these interfaces at the higher velocities became substantially larger. Subjects had more stimulus to allow them to better differentiate between interfaces. Additionally, it can be speculated that as the impact severity moves toward a level that might be considered damaging, then perhaps different sensory receptors come into play to fine tune our severity perception. Nociceptors in the plantar skin of the foot could have become increasingly involved as impacts became more severe. However, their involvement appears unlikely for the present range of impact severities because none of the subjects reported any pain caused by the impact loads experienced, and this pain is necessary for nociceptor stimulation(23).

Individually, the mechanical input relationships to perception were high and characteristic to each subject. In contrast, Milani et al.(20) found low correlations at the individual level between perception and mechanical input variables during running. As they pointed out, the complex demands of the experimental protocol used in their study may help explain why the individual relationships were much lower. Subjects were required to perceive the impact severity, pressure, and rear foot movement while attempting to hit a force platform mounted in the floor at a specific velocity. Also, the subjects in the study by Milani et al.(20) may have relied upon other stimuli to assess impact severity. Mechanical inputs that were unmeasured (e.g., from the leg that didn't hit the force plate) may have influenced the perception ratings. It has been suggested that the best approach to minimize experimental bias is to simplify the testing situation as much as possible to avoid information overload. The subject's task should be made as simple as possible, and instructions should be very explicit with regard to what is required of the subject during testing (4). The human pendulum apparatus allowed the mechanical inputs to the body to be isolated and quantified, and the subjects were only asked to judge the impact severity. Magnitude estimation results are affected by both sensory and judgmental processes, and both contribute to the total variability in the results from different subjects. Therefore, although group data has been demonstrated to be relatively free of bias, this will not be true for all data of the individual subjects making up the group (4). Some subjects compress the number continuum and others expand it, and other natural inclinations can also be observed. The perceptual ratings of the present study were likely affected by natural inclinations and judgments based on prior experience. Such judgmental differences probably limit comparisons between subjects. In view of this bias inherent in each individual data set, inter-individual differences were not examined. The possibility of predicting impact perception in a given individual without making a direct measure of perception still presents an appealing challenge.

The multiple correlation analysis performed using mean subject-condition data used 171 data points (9 * 19) per biomechanical variable and incorporated some of the variability in perceptual ratings and mechanical inputs between subjects. These relatively robust correlations also demonstrated that the mechanical input variables were highly related to the perception of impact severity as well as each other. The lack of improvement in explained impact perception variability with the combination of several input variables indicated that, in addition to being highly correlated with each other, the mechanical input variables explained a similar portion of the perceptual ratings variability. For similar reasons, a principal component analysis failed to simplify the underlying structure of the variables and their relationships to perception.

The present study employed a mixed order of presentation of impact conditions in an attempt to minimize the level of anticipation or expectation of impact severity. Although care was taken to administer the impacts quickly and randomly, subjects could have been aware of the impact velocity and this in turn may have been a biasing factor on the ratings of perception. Some indication of the level of expectation of impact severity was obtained from the precontact integrated electromyographic (EMG) activities of four lower extremity muscles. The activity of all four examined muscles immediately before impact was not significantly altered by the three mat conditions, but they were all modified significantly by impact velocity. Integrated precontact activity increased with increasing impact velocity. This suggests that the subjects may have been aware of the faster impact velocities and, correspondingly, had some expectation of a more severe impact. This potential biasing influence on the results was removed by re-calculating relationships of input variables with perception ratings at each impact velocity separately(Fig. 3). From this figure it can be seen that, when the level of expectation of the impact severity is minimized, some of the transient rather than peak values of the mechanical input variables (namely, impact force and head acceleration) became more related to the perception of impact severity. The slightly better relationship of the transient variables was also seen when plotting group mean data at the fastest impact velocity. The impact force rate of loading followed perceptual ratings more closely than the peak value (Fig. 2).

Under controlled initial impact conditions that are similar to those experienced during locomotion, the mechanical input variables measured were all related to the perceived severity of impact. The high correlations suggest that, within the range of impact severities examined, the perceived impact severity was closely associated with the mechanical input variables commonly measured. Collaborative work with researchers specializing in sensory neurophysiology may reveal cause and effect relationships between impact loading and perceptual sensations. Once the basis for the perception of impact severity has been established, then it remains to be examined how we adapt or respond to perceived loading. The present findings indicate that midsole materials such as those typically found in athletic footwear do not remove our ability to perceive the severity of impact loads. This sensory faculty of the human body is probably important in the evaluation of comfort and in the setting up of injury avoidance strategies when the cushioning capabilities of the body and/or foot-ground interface cannot sufficiently attenuate the impact. Adaptive responses to perceived loads likely help in protecting the body against chronic overloading and high impact shocks during locomotion. Further information in this area is fundamental to our understanding of how man adapts to his environment.

The authors would like to thank the funding support of NSERC Canada and NIKE Inc. for providing the athletic shoe midsole materials.

Address for correspondence: Mark J. Lake, Ph.D., School of Human Sciences, Mountford Building, Liverpool John Moores University, Liverpool L3 3AF, UK. E-mail:

Figure 1-Schematic of the human pendulum apparatus. Steel up-rights mounted on a trolley allowed rapid adjustment of the pull back distance and pendulum impact velocity

Figure 1-Schematic of the human pendulum apparatus. Steel up-rights mounted on a trolley allowed rapid adjustment of the pull back distance and pendulum impact velocity

Figure 2-Group mean perception ratings plotted with mean values for peak impact force (PKF) and force rate of loading (FRA) at each impact velocity. In contrast to peak force, the force rate of loading variable matches the changes in perceptual ratings for the soft interface condition

Figure 2-Group mean perception ratings plotted with mean values for peak impact force (PKF) and force rate of loading (FRA) at each impact velocity. In contrast to peak force, the force rate of loading variable matches the changes in perceptual ratings for the soft interface condition

Figure 3-Correlation coefficients relating perception of impact severity to mechanical input variables determined at each impact velocity seperately (across interface materials)

Figure 3-Correlation coefficients relating perception of impact severity to mechanical input variables determined at each impact velocity seperately (across interface materials)

Back to Top | Article Outline


1. Bates, B. T., J. Hamill, and P. Devita. The evaluation of strategies used to accommodate additional loads during running. Proc. 5th Biennial Conf. of the Canadian Society of Biomechanics, Ottawa, Canada, 1988, pp. 40- 41.
2. Bobbert, M. F., M. R. Yeadon, and B. M. Nigg. Mechanical analysis of the landing phase in heel-toe running. J. Biomech. 25:223-234, 1992.
3. Bojsen-Møller, F. and U. Jøgensen. The plantar soft tissues: functional anatomy and clinical applications. In:Disorders of the foot and ankle, Vol. 1, M. H. Jahss (Ed.). Philadelphia: W. B. Saunders, 1991, pp. 532-540.
4. Bolanowski, S. J. and G. A. Gescheider. Ratio scaling of Psychological Magnitude, Hillsdale, NJ: Lawrence Erlbaum Associates, 1991, pp. 295-311.
5. Caster, B. L. and B. T. Bates. The assessment of mechanical and neuromuscular response strategies during landing. Med. Sci. Sports Exerc. 27:736-744, 1995.
6. Cavanagh, P. R., G. A. Valiant, and K. W. Misevich. Biological aspects of modeling shoe/foot interaction during running. In:Sport Shoes and Playing Surfaces. C. E. Frederick (Ed.). Champaign, IL: Human Kinetics, 1984, pp. 24-46.
7. Chen, H., B. M. Nigg, and J. de Koning. Relationship between plantar pressure distribution under the foot and insole comfort.Clin. Biomech. 9:335-341, 1994.
8. Clarke, T. E., E. C. Frederick, and L. B. Cooper. The effects of shoe cushioning upon ground reaction forces in running. Int. J. Sports Med. 4:247-251, 1983.
9. Clarke, T. E., E. C. Frederick, and L. B. Cooper. Biomechanical measurement of running shoe properties. In: Biomechanical Aspects of Sports Shoes and Playing Surfaces. B. Nigg and B. Kerr (Eds.). Calgary: University of Calgary Press, 1983, pp. 25-33.
10. Cole, G. K., B. M. Nigg, G. H. Fick, and M. M. Morlock. Internal loading of the foot and ankle during impact in running. J. Appl. Biomech. 11:25-46, 1995.
11. de Koning, J. and B. M. Nigg. Kinematic factors affecting initial peak vertical ground reaction forces in running.Proc. Int. Congress of Biomechanics, Paris, France, 1993, pp. 316-317.
12. Goldstein, B. E. Sensation and Perception(2nd Ed.). Belmont, CA: Wadsworth Publishing Company, 1984, pp. 18-23.
13. Griffin, P. Handbook of Human Vibration. New York: Academic Press, 1990.
14. Hennig, E. M., G. A. Valiant, and Q. Liu. Biomechanical variables and the perception of cushioning for running in various types of footwear. J. Appl. Biomech. 12:143-150, 1996.
15. Jordan, C. and R. Bartlett. The relationship between plantar and dorsal pressure distribution and perception of comfort in causal footwear. Proc. IV Emed Foot Pressure Conference, Ulm, Germany, 1994, p. 23.
16. Kinoshita, H., T. Ogawa, K. Arimoto, K. Kuzuhara, and K. Ikuta. Shock absorbing characteristics of human heel properties.Proc. Int. Congress of Biomechanics, Perth, Australia, 1991, pp. 527-528.
17. Lafortune, M. A. and M. J. Lake. Human pendulum approach to simulate and quantify locomotor impact loading. J. Biomech. 28:1111-1114, 1995.
18. Lake, M. J. and M. A. Lafortune. Heel plantar pressure distortion caused by discrete sensors. Proc. Int. Congress of Biomechanics, Perth, Australia, 1991, pp. 370-372.
19. McNitt-Gray, J., D. M. E. Irvine, C. A. Barbieri, and D. D. Anderson. The effect of landing surface on muscle activity in preparation for landing. Proc. American Society of Biomechanics, Tempe, Arizona, 1991, pp. 230-231.
20. Milani, T. L., Hennig, E. M., and Valiant, G. A. Perception of biomechanical variables for running in footwear of varying midsole stiffness. Proc. 2nd Symp. on the Biomechanics of Functional Footwear, Cologne, Germany, 1995.
21. Miller, D. I. Ground reaction forces in distance running. In: Biomechanics of Distance Running, P.R. Cavanagh (Ed.). Champaign, IL: Human Kinetics, 1990, pp. 203-224.
22. Milliron, M., and P. R. Cavanagh. Sagittal plane kinematics. In: Biomechanics of Distance Running, Ch. 3, P. R. Cavanagh (Ed.). Champaign, IL: Human Kinetics, 1990, pp. 71-72.
23. Perl, E. R., T. Kumazawa, B. Lynn, and P. Kenins. Sensitization of high threshold receptors with unmyelinated (C) afferent fibers. Prog. Brain Res. 43:263-277, 1976.
24. Robbins, S. E. and G. J. Gouw. Athletic footwear: unsafe due to perceptual illusions. Med. Sci. Sports Exerc. 23:217-224, 1991.
25. Robbins, S. E., A. M. Hanna, and G. J. Gouw. Overload protection: avoidance response to heavy plantar surface loading. Med. Sci. Sports Exerc. 20:85-92, 1988a.
26. Robbins, S. E., A. M. Hanna, and L. A. Jones. Sensory attenuation induced by modern athletic footwear. J. Testing Eval. 16:412-416, 1988b.
27. Shorten, M. R. and D. S. Winslow. Spectral analysis of impact shock during running. Int. J. Sport Biomech. 8:288-304, 1992.
28. Stevens, S. S. The direct estimation of sensory magnitudes: loudness. Am. J. Psychol. 69:1-25, 1956.
29. Stevens, S. S. On the psychophysical law.Psychol. Rev. 64:153-181, 1957.
30. Stevens, S. S. Psychophysics and Social Scaling,: Morristown, NJ: General Learning Press, 1972, pp. 1-39.
31. Stevens, S. S. Psychophysics: II. Scaling Methods. Handbook of Experimental Psychology, Ch 1, Morristown, NJ: General Learning Press, 1972.
32. Valiant, G. A., T. A. McMahon, and E. C. Frederick. A new test to evaluate the cushioning properties of athletic shoes. In:Biomechanics X-B. B. Jonsson (Ed.). Champaign, IL: Human Kinetics, 1987, pp. 937-941.
33. Whittle, M. W., T. A. Orofino, and K. Miller. Relationship between mechanical properties of carpet and comfort in walking.Proc. 2nd N. Am. Congress on Biomechanics, pp. 283-284: Chicago, 1992.
34. Williams, K. R. Biomechanics of running. Exerc. Sport Sci. Rev. 13:389-441, 1985.


©1998The American College of Sports Medicine