YOUNG ADULTS PERCEIVE SMALL DISTURBANCES TO THEIR WALKING BALANCE EVEN WHEN DISTRACTED

Abstract

Background:

The ability to perceive disturbances to ongoing locomotion (e.g., slips and trips) may play an important role in walking balance control. However, how well young adults can perceive such disturbances is unknown.

Research Question:

The purpose of this study was to identify the perception threshold in young adults to subtle slip-like locomotor disturbances.

Methods:

Subjects (n = 12) walked on a split-belt treadmill performing a perturbation discrimination task at their preferred walking speed while randomly experiencing locomotor balance disturbances every 8–12 strides. Balance disturbances were imposed through a short-duration decrease in velocity of a single treadmill belt triggered at heel-strike. The treadmill belt returned to the subject’s preferred walking speed during the subsequent swing phase. Locomotor disturbances were given with eight different velocity changes ranging from 0 to 0.4 m/s and were randomized and repeated 5 times. Subjects were prompted to respond when asked if they perceived each disturbance. Using a psychophysical approach, we determined the perception thresholds of slip-like locomotor disturbances (i.e., just noticeable difference). The perturbation discrimination task was repeated with subjects performing a secondary cognitive distraction (counting backward by threes).

Results:

Subjects perceived small locomotor disturbances during both normal walking (dominant: 0.07 ± 0.03 m/s, non-dominant: 0.08 ± 0.03 m/s) and while performing the secondary cognitive task (dominant: 0.08 ± 0.01 m/s, non-dominant: 0.09 ± 0.02 m/s). There was no significant difference between legs (p=0.466), with the addition of the cognitive task (p=0.08), or interaction between leg and task (p=0.994).

Significance:

The ability to perceive subtle slip-like locomotor disturbances was maintained even when performing a cognitively distracting task, suggesting that young adults can perceive very small locomotor disturbances.

1. Introduction

The ability to perceive disturbances to ongoing motion may play an important role in the control of walking balance. Falls caused by external disturbances can be prevented by rapid compensatory action, which has historically been attributed to automatic processes. However, there is growing support that cortically mediated responses also play a critical role. In particular, cortical responses to balance perturbations have been identified even in healthy young adults [1,2]. Although cortical recruitment is highest in sensory and motor areas, pre-frontal areas are also recruited [1,3]. These same pre-frontal areas are also involved in the conscious perception of stimuli [4,5], suggesting a role of conscious perception of body motion in the maintenance of balance. Several studies have identified declined disturbance perception in balance-impaired populations [6,7] supporting its importance for balance maintenance. Despite this support, little is known regarding the ability to perceive disturbances that occur during locomotion. As a first step towards understanding the role of disturbance perception in the maintenance of locomotor balance, the goal of this study was to characterize the psychometric threshold of locomotor disturbances in young adults with no balance impairments.

Many falls occur during locomotion due to slips and trips [8,9], yet the perception of such disturbances has not been studied. To date, most studies investigating the perception of balance disturbances have done so in standing. Such studies have found that young adults can consciously perceive very small disturbances to their standing balance, e.g., slow support-surface rotations (<1°/s [10]) and translations [11],) and small changes in the direction of support-surface translations (<10° difference between perturbation directions [12]). However, the sensitivity of the nervous system changes across behaviors (e.g., sitting, standing, walking). For example, both spinal and corticospinal excitability is decreased in walking compared to standing [13,14], and proprioceptive feedback from muscle spindles is reduced during dynamic versus static conditions [15]. Such decreased excitability and sensitivity may lead to differences in the ability to perceive locomotor disturbances. As such, results from prior standing balance studies may not translate to the ability to perceive disturbances during locomotion. Several recent studies have evaluated the perception of inter-limb asymmetry under split-belt walking conditions, in which the speed of the treadmill belt under each leg moves at a different speed [7,16,17]. However, perceiving inter-limb speed differences within such a paradigm may not necessarily translate to real-world falls due to trips and slips. The former occurs over several strides whereas the latter is more discrete in nature, occurring over less than a single stance phase.

Many of the same brain areas involved in executive function overlap with those involved in the conscious perception of stimuli (e.g., [4,5]). Thus, performing a secondary cognitive task may interfere with the ability to consciously perceive locomotor disturbances. This potential for interference is important to understand because navigating daily life often involves concurrent performance of both motor and cognitive tasks (e.g., walking while talking or walking while texting). The extent to which cognitive distraction affects the ability to consciously perceive balance disturbances is unknown. Several studies have demonstrated that young adults sacrifice cognitive performance to maintain their walking balance when they encounter a locomotor disturbance, e.g., a change in surface rigidity [18], surface elevation and narrowing [19], and stepping over an obstacle [20]. This “posture first” strategy suggests that young adults may maintain their ability to perceive locomotor disturbances even with the addition of a cognitive task.

As a first step towards understanding the role of conscious perception of locomotor disturbances in the maintenance of walking balance, the first goal of this study was to quantify the conscious perception threshold to external locomotor disturbances in young adults. Our secondary goal was to test the hypothesis that the conscious perception threshold to external locomotor disturbances would not change with cognitive distraction. We used a psychophysics approach to determine just-noticeable thresholds of small slip-like locomotor disturbances while walking on a treadmill and compared the thresholds identified with and without performing a secondary distractive cognitive task.

2. Methods

2.1. Subjects

Thirteen young adults (6 Females, 23.0 ± 3.0 years old, Table 1) participated. All subjects were in good health with no self-reported history of recent lower extremity injuries, neurological deficits, or vision or balance problems. Subjects were novel to the experiment paradigm used in this study and provided written informed consent before participating according to the protocol approved by the Institutional Review Board of West Virginia University.

Table 1:

Subject demographics and conscious perception thresholds.

				Normal			Cognitive
Subject	Age	Sex	SSWS (m/s)	Perception Threshold (m/s)			Perception Threshold (m/s)
				Dominant	Non-Dominant	Catch Trials Guessed	Dominant	Non-Dominant	Catch Trials Guessed
S1	23	F	1.25	0.055	0.088	1	0.076	0.112	0
S2	21	M	1.15	0.076	0.076	2	0.063	0.053	0
S3	21	M	1.2	0.067	0.088	0	0.075	0.075	1
S4	20	F	1.15	0.086	0.067	0	0.076	0.068	2
S5	24	F	1.2	n/a	n/a	2	n/a	n/a	4
S6	22	F	1.15	0.063	0.063	1	0.063	0.063	0
S7	30	F	1.05	0.04	0.075	0	0.104	0.086	0
S8	22	M	1	0.055	0.048	0	0.076	0.088	0
S9	23	M	0.95	0.068	0.088	0	0.088	0.125	0
S10	24	M	1.25	0.125	0.104	0	0.1	0.113	0
S11	27	F	1	0.128	0.15	0	0.104	0.108	0
S12	19	M	1.25	0.049	0.075	0	0.088	0.105	0
S13	21	M	1.3	0.076	0.059	0	0.091	0.074	1

2.2. Experimental Protocol

Subjects were placed into a support harness over a split-belt treadmill (Bertec, Columbus, OH) that collected ground reaction forces (GRFs) under each foot at 1000 Hz. Subjects were given 2 minutes to adapt to walking on the treadmill before determining self-selected walking speed (SSWS). To determine SSWS, we modified belt speeds using verbal responses. Participants were first asked “Would you consider yourself to walk fast or slow when you walk comfortably?” If they responded “fast”, the speed was increased to 1.3 m/s and then decreased by 0.05 m/s intervals until they stated this was close to their “comfortable pace” they would walk at for 10 min. In contrast, if they responded “slow”, the speed was increased by 0.05 m/s from 1.0 m/s until it was close to their comfortable speed. Subjects walked at their SSWS for an additional 2–3 minutes to further acclimate to the treadmill (Fig. 1A).

Figure 1: Experimental Setup.

A) Subjects walked for 2–3 minutes at their self-selected walking speed (SSWS) with no perturbations. B) Baseline counting performance (serial subtraction by threes) was performed while seated. Subjects then walked on a split-belt treadmill while experiencing slip-like disturbances in a C) normal condition and a D) cognitively distracted condition in which subjects performed serial subtraction by threes. Subjects were given a new random number after each perturbation. The order of perturbation discrimination conditions (C: normal vs. D: cognitive) differed between subjects. E) To minimize fatigue, each discrimination task was split into four 8–10 minute trials. Each trial contained 20 randomized perturbations (out of 80 total) and a 5 min break was given in-between trials.

Subjects performed the following discrimination task (Fig. 1C). Approximately every 8–12 strides, one supporting leg was decelerated during stance (Fig. 2). The response to such treadmill perturbations induces similar kinematic responses as real-world slips [21]. The deceleration was triggered at heel-strike, defined as the point where the vertical GRF achieved 20% of subject body weight. The perturbed belt was decelerated by a magnitude (dV) of 0, 0.02, 0.05, 0.1, 0.15, 0.2, 0.3, and 0.4 m/s and returned to SSWS during the subsequent swing phase. These eight dV perturbations were randomized and repeated 5 times per leg (80 total perturbations). After each perturbation (including 0 m/s), subjects were asked to respond “Yes” or “No” to whether they felt a disturbance to their balance and responded within 5 steps. This protocol was repeated while performing a cognitive distractive task - counting backwards by threes (Fig. 1D). After recovering from each perturbation, subjects were given a new number from which to count backwards and experienced their next perturbation after 8–12 strides. The average time subjects spent counting per perturbation trial was 9.6 ± 2.1 seconds. The audio of counting was recorded for each subject for later analysis. Subjects were instructed to look forward and not use the handrails for support. They wore noise canceling headphones playing white noise to prevent them from hearing the treadmill changing speeds.

Figure 2: Discrimination Tasks.

A) Subjects walked on a split-belt treadmill at their self-selected walking speed (SSWS) with subtle slip-like disturbances delivered every 8–12 strides. Subjects wore a safety harness and noise cancelling headphones to limit auditory feedback from the treadmill changing speed. B) Eight locomotor disturbance speeds (dV) were randomized and repeated 5 times on each leg. The disturbance started at heel strike and belt velocity was returned to SSWS of during the subsequent swing phase.

The following measures were taken to increase consistency between subjects, prevent fatigue, and diminish learning effects. First, prior to the start of the discrimination task all subjects received standardized instructions and two practice perturbations: “While you walk, you may experience balance challenges of varying difficulty. When these balance challenges occur, try to continue walking normally. Throughout the experiment we will ask you whether you recently felt a challenge to your balance. Please respond either ‘Yes’ or ‘No’. Now we are going to give you an example of both when a balance challenge occurs and when it does not. After I say ‘Response’ please answer with ‘Yes’ or ‘No’.” Second, perturbation order was consistent between subjects within each condition (normal, cognitive). Third, the discrimination task for each condition was split into sub-trials of 8–10 minutes each (20 perturbations per sub-trial) with at least 5 minutes of rest between sub-trials to prevent fatigue (Fig. 1E). Finally, to diminish the effect of learning on the discrimination of balance perturbations between normal and cognitive conditions, task order was reversed in approximately half of the subjects.

2.3. Data analysis

The conscious perception threshold was calculated for each leg and condition independently using the psignifit4 MATLAB toolbox [22]. Briefly, this toolbox fits a psychometric curve to the proportion of perceived disturbances (out of 5 repetitions) for all perturbations (Fig 3A). We used a logistical sigmoid to fit the proportions of perceived disturbances and determined the threshold at 50% of the fit. Because significant guessing (i.e., positive “Yes” responses at 0 m/s perturbation) prevents accurate threshold identification, we removed any subject who had a positive response rate at 0 m/s in more than one-third of trials in either condition. Based on this cutoff, one subject (S5) was removed from further analysis (Table 1).

Figure 3: Psychophysical Perception Threshold of Locomotor Disturbances.

A) Individual box and whisker distributions for each perturbation level. These plots show the number of perceived perturbations for both legs across each condition per subject (n = 13). Outliers are labeled with the subject ID from Table 1. S05 was removed from further analyses due to their high response rate at the 0 m/s catch trials B) Example identification of the conscious perception threshold of locomotor disturbances. The conscious perception threshold was identified as the 50% point on the psychometric curve fit to the proportion of correct responses for each dV. C) There was no significant difference between legs (p = 0.466), with the addition of the cognitive tasks (p = 0.08), or interaction between leg and task (p = 0.994)(n=12, mixed-method ANOVA; blue = normal condition, gray = cognitive condition).

All statistical analyses were performed in the R-based statistical software JAMOVI (Version 1.1.9.0) [23]. A mix-method ANOVA was used to test for main effects on the perception threshold due to condition (normal, cognitive) and interactions between condition and leg (dominant, non-dominant). Leg dominance was labeled as the limb preferred when kicking a ball. Pearson’s correlation analyses were used to determine if there was a relationship between SSWS and perception threshold. To confirm the order of tasks did not affect perception thresholds, the difference in thresholds between conditions (normal minus cognitive) was compared between subjects who performed the normal (n=7) versus cognitive (n=5) condition first using a two-sample independent t-test.

We focused on two metrics–counting accuracy and rate–to examine the effect of dual-tasking on counting performance, calculated as follows:

$${\text{Accuracy}=\text{N}}_{\text{correct}}{/(\text{N}}_{\text{correct}}{+\text{N}}_{\text{incorrect}})$$

$${\text{Rate}=(\text{N}}_{\text{correct}}{+\text{N}}_{\text{incorrect}}{)/\text{T}}_{\text{trial}}$$

where T_trial is defined as the time elapsed between when the experimenter gave the starting number to the subject and when the experimenter asked for the perception response.

We calculated the dual task cost (DTC) for both counting accuracy and rate according to the following formula [24,25]:

$$\text{DTC}=(\text{single task}-\text{dual task})/\text{single task}\times \text{100}$$

The baseline cognitive task was seated counting (Fig. 1B). Immediately prior to experiencing the discrimination task with cognitive distraction, each subject performed two trials of serial subtraction by threes from a random number for 15 seconds while seated. The single task cost was taken as the average of the two seated trials. A separate dual task cost was calculated for each perturbation resulting in 80 DTCs each for accuracy and rate. We then averaged DTCs across all perturbations to get one value each for accuracy and rate per subject.

One subject was excluded from DTC calculations due to missing audio recordings, and 7 trials across all other subjects (< 1% of trials) were excluded due to audio recording errors. A Wilcoxon rank-signed test was used to determine if DTC was affected by experiencing locomotor disturbances. To determine if there was a relationship between DTC and the ability to consciously perceive locomotor disturbances, Pearson’s correlations were performed between the DTC (for both accuracy and rate) and the change in dominant leg threshold across conditions (Cognitive - Normal).

3. Results

Subjects walked at a SSWS of 1.14 ± 0.11 m/s (Table 1). All but one subject was right leg dominant. There was no significant correlation between SSWS and perception threshold during normal walking (dominant leg: p = 1.00, non-dominant leg: p = 0.537) or walking with the secondary cognitive task (dominant: p = 0.393, non-dominant: p = 0.673).

Subjects perceived small locomotor disturbances during normal walking (dominant: 0.07 ± 0.03 m/s, non-dominant: 0.08 ± 0.03 m/s) and while performing the secondary cognitive task (dominant: 0.08 ± 0.01 m/s, non-dominant: 0.09 ± 0.02 m/s) (Fig. 3B). There was no significant difference between legs (p = 0.466), with the addition of the cognitive task (p = 0.08), or interaction between leg and task (p = 0.994). Additionally, the difference in thresholds between conditions was not significantly different (p = 0.436) between subjects that performed the normal (0.015 ± 0.032 m/s) versus the cognitive (0.003 ± 0.015 m/s) condition first, confirming that task order did not matter.

There was a significant dual-task cost on cognitive performance for counting rate (−14.9 ± 21.8%; p = 0.042) such that rate increased during locomotor perturbations compared to seated counting. There was no significant dual-task cost on counting accuracy (0.05 ± 6.3%; p = 0.966) (Fig. 4A). The difference in perception thresholds between conditions was not significantly correlated with either dual-task cost on accuracy (r = 0.14, p = 0.691) or rate (r = −0.11, p = 0.748) (Fig. 4B).

Figure 4: Cognitive dual-task costs.

A) There was a significant dual-task cost on counting rate such that rate was increased during the cognitive condition (p = 0.042), but there was no significant dual-task cost on counting accuracy (p = 0.966; n=11, Wilcoxon Signed Rank test). Each marker type and color correspond to individual subjects. B) There was no correlation between dual-task cost and the change in the perception threshold between conditions (dominant leg, Cognitive – Normal; accuracy: r = −0.14, p = 0.691; r =−0.11, p = 0.748).

4. Discussion

The purpose of this study was to determine the conscious perception threshold of discrete locomotor disturbances in young adults and how this is affected by performing a distractive cognitive task. Consistent with our hypotheses, our results provide evidence that young adults can perceive subtle slip-like locomotor disturbances and that they maintain this ability when distracted. Although the current study is focused on the perception of slip-like disturbances in young adults, our results have important implications for interpreting changes that may occur in balance-impaired populations.

Young adults could consciously perceive slip-like locomotor disturbances of small magnitude, e.g., less than 0.10 m/s deviation from their self-selected walking speed. This sensitivity to small speed deviation between legs is similar to that previously found in locomotor inter-limb asymmetry studies. For example, Bekkers et al. 2017 found that healthy adults were capable of perceiving inter-limb asymmetry of approximately 0.08 m/s. Another study by Galbreath and colleagues found the inter-limb symmetry threshold could be as low as 0.01 m/s [26]. This similarity in thresholds occurs despite the fact that the slip-like disturbance in the current study occurs over a single stance phase (Fig. 2) rather than multiple strides as in the inter-limb asymmetry protocols. In contrast, our slip threshold is slightly faster than the presumably imperceivable “micro-slip” of approximately 0.10 m/s that occurs at heel-strike during normal walking over a period of approximately 50 ms [27]. Taken together, these studies suggest there is both a magnitude and temporal component contributing to the ability to perceive locomotor slips, which warrants future study. Our results provide evidence, however, that over a long enough slip duration young adults are sensitive to small and transient changes in sensory feedback during locomotion. In particular, post-hoc exploration of joint kinematics in response to perturbations around the perception threshold of 0.10 m/s revealed the peak difference in ankle kinematics was less than 10° with little to no differences in knee, hip, trunk, and head kinematics. This suggests that young adult’s perceptual threshold of locomotor disturbances in the current study is likely in large part due to proprioceptive feedback from muscles, tendons, and joint sensors at the ankle. Such reliance on ankle proprioception in young adults may explain why impaired conscious perception of ankle motion is associated with poor balance and walking performance in older adults [28].

Young adults maintained their ability to perceive subtle slip-like locomotor disturbances when performing a secondary cognitively distractive task. Subjects perceived disturbances less than 10% of their walking speed in both single and dual-task conditions, with no significant difference between conditions (Fig. 3C). Because pre-frontal brain areas are involved in both conscious perception of movement and executive function [4,5], this result suggests that young adults can focus their attention on small sensory cues when they are important. Prior dual-task studies in balance-impaired populations provide evidence of increased reliance on executive control during normal walking [29,30], with a larger effect in fallers compared to non-fallers [31,32]. A potential mechanism that may explain this finding is that increased reliance on executive control for walking negatively impacts the ability to recognize changes in sensory feedback to perceive locomotor disturbances.

The importance of consciously perceiving a locomotor disturbance for walking balance control remains unknown. Preventing a fall when a loss of balance occurs requires a quick and reactive response. This response is mediated by local muscle reflexes, a later brain-stem response, and an even later cortical response [33]. When the spinal and brain-stem mediated responses are insufficient (e.g., due to aging, disease, or injury), the later cortically mediated response that depends on high-level perception of a disturbance may become critical. Consistent with this idea, declines in the ability to consciously perceive single-joint and whole-body motion have been associated with poor balance and walking function in various populations, such as older adults [34,35], stroke survivors [7], and individuals with Parkinson’s disease [6,17]. As such, the ability to consciously perceive locomotor disturbances may become disproportionally relevant for preventing falls in balance-impaired populations and warrants future investigation.

There are several limitations in this study that should be discussed. First, the perturbation perception task was performed on a treadmill and not free-world walking and our slip duration was slightly longer than slip durations that occur overground. However, treadmill-based slips and trips such as that implemented in the current study have been demonstrated to produce fall kinematics that replicates those in realistic, real-world conditions [21], and as such we expect that young adults would have similar perceptual acuity in free-walking conditions that is maintained in the presence of cognitive distractions. Another limitation is that our baseline cognitive condition was performed during sitting and not while walking. While the difference between these two conditions is expected to be low, we cannot identify whether the dual task costs in the current study were only due to the addition of the walking balance disturbances. However, this limitation does not affect our main result that the perceptual acuity is the same in the presence of cognitive distraction. Finally, counting backwards by threes may have been too easy for our young adult cohort and thus future studies should examine whether the ability to maintain perceptual acuity in the presence of cognitive distraction is more strongly affected by a more challenging cognitive task.

5. Conclusions

To our knowledge, this is the first study to examine the conscious perception of subtle slip-like locomotor disturbances. We found that young adults could perceive very small slip-like disturbances (e.g., decelerations of a single treadmill belt over a single stance phase less than 10% of their SSWS), and this perceptual ability was maintained when performing a cognitively distracting task. These results serve as a baseline for further studies investigating to what extent sensory and cognitive declines affect the ability to consciously perceive slip-like locomotor disturbances and its relationship to walking balance performance in individuals at a high risk of falls (e.g., aging, cognitive decline, neurodegeneration, musculoskeletal diseases, etc.).

Highlights.

• We examined the conscious perception of slip-like disturbances in young adults

• Young adults could perceive very subtle slip-like locomotor disturbances

• Perception of slip-like locomotor disturbances unaffected by cognitive distraction