- The study of split-belt walking (i.e., walking with each leg at a different speed) has improved our understanding of the neural control of gait.
- It also has led to new successful gait rehabilitation protocols in individuals poststroke.
- This article specifically addresses the sensory aspects of split-belt walking.
- It also discusses the neurophysiological mechanisms underlying the perception of gait asymmetry and locomotor adaptation, addressing the cortical and spinal involvement in addition to the known importance of the cerebellum.
The field of split-belt walking research is evolving from studying interlimb coordination and locomotor adaptation to optimizing gait rehabilitation. Optimization of gait rehabilitation critically depends on our understanding of the roles of somatosensory perception and sensorimotor recalibration in split-belt walking. Yet, these sensory aspects have received little attention (41). Recently, important progress has been made in this domain. This article addresses the sensory aspects of split-belt walking and, more specifically, what recent findings tell us about the neurophysiological mechanisms underlying the perception of gait asymmetry.
Split-belt walking is a popular paradigm to study locomotor adaptation (for review see (14,32,38)). With a split-belt treadmill different walking speeds can be presented to each leg. When walking on a split-belt treadmill, people quickly adjust several aspects of their gait pattern to match the different belt speeds without falling (i.e., they walk asymmetrically). However, through time, people slowly adapt some gait parameters to minimize the overall gait asymmetry (e.g., step length and double-stance times; (33)). These slower gait adaptations are then retained; when the belts are returned to equal speed, people initially walk asymmetrically (33). These asymmetric aftereffects mirror the initial asymmetries during split-belt walking and typically wash out completely within a few minutes (33). These motor aftereffects are accompanied by sensorimotor aftereffects, indicative of sensory recalibration taking place during split-belt walking (22,39). The split-belt walking paradigm has been used to improve our understanding of the neural control of gait (4) and has led to new successful gait rehabilitation protocols for individuals poststroke (34).
Although some fast gait changes occurring during split-belt walking can be direct results of the changed mechanical constraints of walking with each leg at a different speed, other (slower) gait changes result from a modified neural drive. These latter changes depend on the sensitivity of the nervous system to asymmetry in somatosensory feedback (17). Both unconsciously and consciously perceived asymmetry can result in changes in the gait pattern (40). Improving our understanding of how the nervous system perceives gait asymmetry will help to optimize split-belt treadmill rehabilitation and will provide fundamental insights in the persistence of gait asymmetry in several patient populations with neurological (40) or musculoskeletal disorders.
This article specifically addresses somatosensory perception and sensory recalibration during split-belt walking. It is hypothesized that perception of gait asymmetry is based mainly on the detection of temporal mismatches between afferent inputs occurring at the spinal level and at higher levels.
Split-belt treadmills have become widely available and many different aspects of split-belt adaptation have been addressed. It should be acknowledged, however, that split-belt walking is a highly artificial task and does not occur outside the laboratory. The real-life overground situation that is most similar is walking in circles, during which somewhat similar gait asymmetries occur (6). Split-belt walking provides a window into the complex interactions between the sensory and motor systems and reveals underlying sensory mechanisms.
In a classic split-belt walking paradigm, people walk with one leg 100% or 200% faster than the other leg for a period of 10–15 min. During this period of split-belt walking, both somatosensory perception and sensorimotor recalibration play important roles in the coordination of the gait pattern. At first, the gait pattern needs to be modified to meet the demands of walking with one leg on a belt with a different speed than the other leg. Normal gait coordination with symmetry in both stance times and limb excursions cannot be maintained and will likely result in a fall. Interestingly, although these intralimb gait parameters (coordinated within one leg) change quickly on exposure to split-belt walking, interlimb coordination (coordination between legs) remains relatively unchanged at first. For example, initial foot placement location and timing are similar between sides (25), which results in asymmetries in step length and double support times because of the asymmetry in belt speeds. Although walking with all these initially observed asymmetries is possible, slow changes in the gait pattern typically occur through a period of more than 100 strides (33). These slower changes are most prominent for interlimb gait parameters; step lengths, for example, often completely return to symmetric values (Fig. 1).
Although the details of the neurophysiological pathways are yet to be established, it is apparent that the central nervous system perceives that the initial gait coordination is suboptimal and subsequently reduces the related error signals stride by stride. This suboptimality could be related to metabolic energy expenditure (10), gait instability, or neural control complexity. Conceptually, gait coordination can be expected to be computationally less expensive (more optimal) when interlimb neural networks can operate symmetrically (their default mode). The stride-by-stride reduction of error signals is done through sensorimotor recalibration because aftereffects are observed when the treadmill belts are returned to equal speeds. The most obvious example is the aftereffect in step length symmetry: during split-belt walking, initially the step length for the leg stepping onto the slow belt is longer. Then, during the adaptation period the slow leg's step length becomes shorter, eventually resulting in symmetric step lengths (Fig. 1). When the belts are returned to equal speeds, people initially keep walking with a shorter step length for the leg that used to be on the slow belt. This results in step length asymmetry, which mirrors the asymmetry initially observed during split-belt walking (Fig. 1; (33)).
Split-belt walking results in aftereffects in both the motor (e.g., step length asymmetry) and the sensory domains. Examples of aftereffects in the sensory domain are observations that after split-belt walking, participants perceive the leg that was on the slow belt to move faster than the other leg (33), and that participants are unable to set the belts to equal speeds (22). Both motor and sensory aftereffects provide a measure of the extent to which gait alterations are stored in the central nervous system and differentiate motor adaptation from conscious adjustment of movement (1). Traditionally, adaptation and aftereffects are attributed only to interlimb gait parameters (33); however, some recent studies have shown aftereffects in intralimb gait parameters as well (Fig. 1C; (18,28)).
The aftereffects of split-belt walking wash out within several minutes, and transfer to overground walking is only limited (35). However, promising results have been reported for split-belt training as a tool in gait rehabilitation in individuals poststroke (34). Gait asymmetry is a common symptom in individuals poststroke with hemiparesis. Considering an affected gait pattern with asymmetric step lengths, a session of split-belt walking has the potential to result in a more symmetric gait pattern (35). For that to work, two conditions must be met: first, patients need to display motor adaptation during split-belt walking and second, the aftereffects should result in long-lasting changes. Reisman et al. (34) showed that 12 sessions of split-belt treadmill training can result in improved gait symmetry in individuals poststroke, maintained for at least 3 months; however, it is important to note that these improvements were observed in only 7 of the 12 patients in the study. It is unclear why split-belt treadmill training works for some but not for all individuals poststroke with asymmetric gait.
An important related question is (12,18,20,21,27,35): which parts of the central nervous system contribute substantially to split-belt adaptation? In reviewing the literature on underlying neurophysiological mechanisms of split-belt adaptation and perception of gait symmetry, it should be acknowledged that many studies have aimed to draw neurophysiological conclusions, although only the kinematics and kinetics of split-belt walking were measured. Obviously, conclusions about underlying neurophysiological mechanisms based on kinematic and kinetic data only have their limitations. However, many of the reviewed studies also have assessed muscle activation (7,10,30), used specific patient populations as a model (17,18,30,34,35,40), or have applied brain stimulation or neuroimaging techniques (17,18,20,21).
Based on its known role in other forms of motor adaptation, the cerebellum has received substantial attention. Although observations on severely ataxic cerebellar patients and data from brain stimulation studies have confirmed the important role of the cerebellum in split-belt adaptation (20,21,27), recent observations indicate that in addition we need to look more into the contributions of spinal circuits and the cortex (17,18,24,36,40). Furthermore, mildly ataxic patients with cerebellar damage show normal split-belt adaptation (18,27). These observations that normal split-belt adaptation can still occur in case of cerebellar damage suggest either that the cerebellar role is smaller than presumed or that the central nervous system has a way to compensate for the cerebellar dysfunction in case of cerebellar damage.
An early study by Reisman et al. (35) showed that cortical stroke does not interfere with split-belt adaptation. This suggested that the cortex plays no major role in split-belt adaptation. However, cortical involvement should not be ruled out completely because specific cortical areas involved in the control of split-belt walking might not have been damaged in this patient sample. As mentioned previously, a follow-up study (34) showed that repeated split-belt training improved gait symmetry in only 7 of 12 individuals poststroke. From the presented data in the follow-up study, it is unclear whether the five nonresponders did not show any aftereffects at all or that adaptation to split-belt walking occurred and that aftereffects were short-lasting or did not transfer to overground walking (34). In any case, the presence of nonresponders suggests that cortical stroke interferes in some degree with locomotor adaptation or with its transfer to overground walking. Similarly, studies combining split-belt walking with a cognitive dual task have shown that (adaptation to) split-belt walking incurs a cognitive load, suggesting cortical involvement in the control of split-belt walking (24,36).
To better understand what roles different parts of the central nervous system play in split-belt adaptation, we could look at the adaptation process from a control perspective. As mentioned, the motor recalibration that occurs during split-belt adaptation is aimed at reducing an error signal that is perceived during split-belt walking. Classically, this error-reducing recalibration is considered to occur in the cerebellum (27). But where does this error signal originate? Apparently, one's initial gait coordination in response to walking on two belts with different speeds is suboptimal in some way. It is plausible that metabolic energy expenditure is being optimized during split-belt walking. Finley et al. (10) showed that during split-belt walking, reductions in full-body metabolic power correlate with reductions in step length asymmetry. However, less understood is how such an error signal quantifying full-body metabolic energy expenditure would be perceived by the cerebellum. Alternatively, the central nervous system perceives a continuous stream of somatosensory information about muscle forces (from Golgi tendon organs), muscle length changes (from muscle spindles), and contact forces (from cutaneous mechanoreceptors). This information could be used to minimize activation of specific muscles or integrated into a proxy for full-body metabolic energy expenditure. For instance, the information could help to optimize inverted pendulum mechanics or minimize individual limb center-of-mass work. This integration could very well take place in the cerebellum (42).
Several studies discussing the possible role of the cerebellum in split-belt adaptation have distinguished feedback versus feed-forward control (25,27,30). In that framework, quickly adjusted gait parameters are considered to be feedback controlled and slowly adapted gait parameters are considered to be feed-forward controlled, using stride-by-stride error reduction. However, split-belt walking has its biomechanical constraints, and different gait features are interdependent. Therefore, some slowly changing gait parameters might actually be feedback controlled and be changing in reaction to the adaptation in the overall gait pattern that is driven by other, feed-forward–controlled parameters. For instance, Ogawa et al. (30) assessed changes in ground reaction forces and muscle activity during split-belt adaptation and observed that braking forces and tibialis anterior muscle activity changed slowly during split-belt walking. These observations could indicate that ankle stiffness at heel contact is feed-forward controlled, as the authors suggested. Alternatively, when gait adaptation takes place by actively reducing the differences in the relative phasing of the steps (25), slow changes in braking forces would automatically result. Similarly, the slow changes in tibialis anterior muscle activity are observed during the early stance phase. This suggests that the muscle activity could very well be caused by reflexes induced by instability at heel contact. Such instability is likely to be reduced simultaneously with the overall gait changes during adaptation. Instability could equally well be either a resulting or a driving factor of split-belt adaptation. To further investigate whether ankle stiffness is feed-forward controlled, it will be interesting to evaluate how muscle activity during late swing, just before heel contact, changes during split-belt walking. To further address the role of feed-forward control, more insights might come from studies comparing adaptation during split-belt walking and split-belt running (29) because running depends more on feed-forward control. The shorter contact times in running allow less time for stable reflex (feedback) control (3).
In multiple ways, somatosensory perception can be expected to play important roles in split-belt walking; both in relation to the motor adaptation process and during the initial exposure to split-belt walking when fast adjustments in stance time and limb excursion take place. Theoretically, one could walk on a split-belt treadmill with either symmetric stance times or symmetric limb excursions. Because one belt is moving faster than the other, maintaining symmetric stance times will result in asymmetric distances covered during the stance phase (i.e., asymmetric limb excursion; (16)). Similarly, the alternative of maintaining symmetric limb excursions will result in asymmetric stance times (17). However, most commonly, people walk with asymmetries in both stance times (the temporal domain) and limb excursions (the spatial domain). The fast changes to enable walking with the legs at different speeds could be coordinated purely through stretch reflexes. We have argued elsewhere that stretch of hip flexors could be the trigger for the onset of flexion at the end of the stance phases (15), similar to what has been observed repetitively in walking cats (8). However, our recent observations that limb excursions are more asymmetric than stance times and that this is more pronounced in participants who are better able to perceive differences between belt speeds suggest a prominent role for walking rhythm (temporal symmetry) (17).
Together, these observations support a hybrid control model. Both the magnitude of afferent input and its timing are important. Interestingly, one of the earliest studies on split-belt walking in humans (7) suggested that split-belt walking is predominantly coordinated through interactions between peripheral sensors and spinal interneuronal circuits. This notion is in line with observations from both classic and recent studies on spinal-transected cats, which display coordinated gait when walking on a split-belt treadmill (11,12). Similarly, human infants have been shown to display interlimb coordination when walking with support on a split-belt treadmill even before they display voluntary locomotion, confirming the importance and autonomy of spinal neurons coordinating the legs during gait (37).
To further address the role of proprioceptive feedback to the spinal cord on split-belt adaptation, Mukherjee et al. (9,28) have applied tactile vibration to the plantar surface of the feet during split-belt walking. The rationale for this intervention was that disrupting the function of the plantar mechanoreceptors with tactile vibration would make the nervous system more reliant on other somatosensory inputs (28). If sensorimotor recalibration of those reliable, alternative somatosensory channels could be enhanced, the overall motor adaptation effects might be stronger and longer lasting (28). Indeed, tactile vibration applied during the adaptation period resulted in longer lasting aftereffects in stance and swing time symmetry (28). Furthermore, when plantar tactile vibration was combined with artificial optic flow, simulating the visual input of walking through a corridor, these temporal aftereffects were reduced, suggesting that somatosensory recalibration was less strong in the presence of moving visual information (9). These studies did not report double support time symmetry, and step length symmetry was estimated based on temporal measures only. This is unfortunate because those interlimb gait features are known to undergo the largest and most robust changes during split-belt adaptation (33), and such changes seem to be important for gait rehabilitation (34). To confirm the suggested benefits of plantar tactile vibration in split-belt gait rehabilitation, more research is needed quantifying the changes in interlimb coordination.
PERCEPTION OF GAIT ASYMMETRY
To gain more insights in the role of somatosensory perception in split-belt walking, several groups have focused directly on how well specific patient populations perceive differences in belt speeds. Lauziere et al. (23) used a task where participants walked with each leg on a separate belt, although initially, both belts ran at the same speed. Then, after a random interval, one of the belt speeds started to accelerate slowly, causing the difference between the belt speeds to increase gradually. Participants had to indicate when they perceived the belts to be moving at different speeds. This difference in belt speeds at that moment is referred to as the perception threshold of gait asymmetry. A similar, but mirrored, task was performed where the belt speeds were different initially, and then the faster belt started to decelerate until the participant perceived the speeds of both belts to be equal. This identified the perception threshold of gait symmetry. With this protocol, the perception thresholds were assessed for healthy, older individuals. Through the whole sample, the perception thresholds were correlated to the stance time asymmetry (23). This finding supports our hypothesis that perception of gait asymmetry is based mainly on detection of temporal mismatches between afferent inputs. However, a role for detection of spatial mismatches cannot be ruled out based only on this correlation. As we have argued elsewhere (15), stance times are known to change quickly in reaction to belt-speed changes (33). As a result, stance time asymmetry is coupled to belt-speed asymmetry. Based on this coupling, the observed correlation between perception thresholds and stance time asymmetry was to be expected. A similar correlation can be expected between the perception threshold and limb excursion asymmetry (15) because the latter also is known to change quickly in reaction to belt-speed changes (33). This potential correlation with limb excursion asymmetry was not assessed in the study by Lauziere et al. (23).
When we (17) followed up on Lauziere et al. (23), we observed a similar strong relation between perception of gait asymmetry and temporal asymmetry. We applied a similar perception threshold detection test in healthy individuals and a population of mildly ataxic cerebellar patients. Then, we assessed the correlation between perception thresholds and initial asymmetries in stance time and limb excursion during a split-belt adaptation trial. Strikingly, on exposure to split-belt walking, participants who were better able to perceive belt-speed differences (lower perception thresholds) walked with the smallest asymmetry in stance time and the largest asymmetry in limb excursion (17). Furthermore, we did not observe differences between cerebellar patients and healthy control participants in their ability to perceive belt-speed differences. The latter suggests that the cerebellum is not involved in the conscious perception of gait asymmetry. Rather, the timing of somatosensory input from both legs is likely to be compared at a different level, either spinal or cortical.
Based on the observation that individuals poststroke who normally walk asymmetrically are able to walk more symmetrically after split-belt training (34), Wutzke et al. (40) hypothesized that poststroke gait asymmetry is partly due to sensory deficits; i.e., assuming that individuals poststroke would not recognize asymmetry as a movement error. To test this premise, they used an adaptive staircase design with short bouts (five strides) of various randomized belt-speed differences. When participants perceived belt speeds to be different, the difference was reduced for the next bout. When participants did not perceive belt speeds to be different, the difference was increased for the next bout. This was repeated until the threshold speed difference was identified. The authors evaluated whether overground asymmetries in step length and stance time increased or decreased because of the imposed belt-speed difference. They observed that participants for whom step length asymmetry was increased during split-belt walking were less able to detect belt-speed asymmetries than participants for whom stance time asymmetry was increased during split-belt walking.
Furthermore, Wutzke et al. (40) showed that individuals who normally walk with a large step length asymmetry were more likely to use step length asymmetry values to perceive belt-speed asymmetry. Similarly, individuals who normally walk with a large stance time asymmetry were more likely to use stance time asymmetry values to perceive belt-speed asymmetry. Note that this is counterintuitive and suggests that gait asymmetry poststroke is related to a recalibrated sense of symmetry rather than because of sensory deficits. If gait asymmetries would partly be due to sensory deficits, one would expect that individuals who walk with asymmetric step lengths would be less able to perceive step length asymmetry and less likely to use step length asymmetry values to perceive belt-speed asymmetry. Wutzke et al. (40) were the first to address perception of gait symmetry in individuals poststroke; their results suggest that perception of gait symmetry during treadmill walking does not occur at the cortical level.
Altogether, the observations of these studies suggest that detection of gait asymmetry is based mainly on detection of temporal mismatches between afferent inputs, most likely at the spinal level (Fig. 2). At that level, the pattern-generating neuronal networks that coordinate the movements of both legs based on afferent inputs can detect asymmetry in their inputs and communicate this to supraspinal centers. Marques et al. (26) addressed the interaction between the spinal pattern generations and their supraspinal controllers. They tested how well healthy individuals are able to manually set the belt speed of one leg to match that of the other, with perturbed visual or vestibular input. Although artificial optic flow did not affect this ability, cold-water irrigation of one ear (perturbing vestibular function) impaired the participants' ability to match the belt speeds. This suggests that set point for symmetry in afferent input as evaluated at the spinal level can be modified by supraspinal actors such as the vestibular system.
Although most split-belt walking studies have addressed motor adaptation, only a few studies have assessed sensorimotor recalibration, which occurs simultaneously. Immediately after split-belt walking, when the belts are returned to equal speeds, people commonly perceive the belt that moved slower during the adaptation period to be moving faster than the actual speed (33). Reisman et al. (35) observed that 7 of 13 individuals poststroke did not experience this belt-speed perception aftereffect, even though motor adaptation was unaffected. Jensen et al. (22) addressed the sensory aftereffects of split-belt walking more quantitatively. After 10 min of walking with one leg at 1.5 km·h−1 and the other at 4.5 km·h−1, when healthy individuals aimed to manually set the belts to equal speeds, a mean speed difference between the belts of 0.85 km·h−1 resulted. To further address the role of somatosensory input on this sensory aftereffect, they studied the effects of body loading and unloading on this task. To increase body loading by 30%, participants wore a lead-filled vest; to reduce body loading by 30% participants were suspended over the treadmill in a parachute harness connected to a winch. As hypothesized, the sensory aftereffects were load dependent. For split-belt walking, when aftereffect assessment was in the same body-loading conditions, the aftereffects were large. When aftereffects were assessed in a different body-loading condition than the split-belt walking adaptation, the aftereffects were small. This suggests that the sensory recalibration that occurs during split-belt walking is task specific. Even when a similar task is performed (i.e., walking), but with different load-dependent input, the sensory aftereffects are substantially smaller.
Recently, Vazquez et al. (39) followed up on the experiments by Jensen et al. (22), addressing the role of sensorimotor recalibration in split-belt adaptation. In addition to evaluating aftereffects in perception of belt speed, they investigated whether foot position perception and stepping force perception were altered by split-belt adaptation. They confirmed the existence of aftereffects in perception of belt speed. But, passive and active foot position perception and stepping force perception were all unaffected by 15 min of split-belt walking at a 1:3 belt-speed ratio. Together, these findings suggest that sensory recalibration mainly occurs in the temporal domain. Although the sensitivity to the magnitude of position and force feedback seems to be unchanged during split-belt adaptation, the perception of belt speed is recalibrated. Still, these observations do not fully rule out the possibility that position and force recalibration also occur. As we have argued elsewhere (19), the findings of Jensen et al. (22) indicate that it is important to consider task specificity when measuring transfer of sensorimotor recalibrations. Because Jensen et al. (22) observed that sensorimotor recalibration only slightly transfers from walking with increased body loading to walking with reduced body loading, it is likely that recalibration also does not transfer to a task that involves stepping, rather than walking.
By combining observations from the studies reviewed herein, one can argue that perception of gait asymmetry is based mainly on detection of temporal mismatches between afferent inputs at the spinal level. This conclusion seems to be supported by animal studies. Temporal coordination during split-belt walking is similar in intact and spinal-transected cats (11,12), indicating that this temporal coordination is established at the level of the spine. However, addressing split-belt adaptation in cats will be valuable for determining at which level sensory recalibration occurs. The changes in the gait coordination to accommodate split-belt walking are most likely produced by sensory inputs from the periphery that project to spinal locomotor central pattern generators. A similar controller that combined distal input with a rhythm generator and pattern formation was recently used in a bipedal robot able to walk on a split-belt treadmill (13). The robot also had a comparator unit predicting the timing of sensory events and reducing the error between predicted and actual input. Classically, the latter function is attributed to the cerebellum, but theoretically it is possible that even this error-driven motor adaptation could be controlled at the spinal level (2). It is unlikely that this theoretical spinal motor adaptation has a major role in split-belt adaptation, but it might become more involved when motor adaptation at the level of the cerebellum is impaired, for example in case of focal cerebellar lesions after tumor resection. Such reorganization could explain why individuals with focal cerebellar lesions show unimpaired split-belt adaptation (18).
Recently, important progress has been made in the domains of somatosensory perception and sensorimotor recalibration in split-belt walking. Observations from studies that have assessed muscle activation, used specific patient populations as a model, or have applied brain stimulation or neuroimaging techniques have all provided preliminary insights into the underlying neurophysiological mechanisms. It seems that perception of gait asymmetry is based mainly on detection of temporal mismatches between afferent inputs. These temporal mismatches can be detected in the spinal cord by pattern-generating neuronal networks that coordinate the movements of both legs. In addition, the timing and magnitude of afferent input could be integrated in the cerebellum into a proxy for metabolic energy expenditure, providing an error signal that is reduced stride by stride during split-belt adaptation. Our understanding of split-belt walking and its application as a tool in gait rehabilitation will benefit from future studies that address the neurophysiological mechanisms more directly by combining state-of-the-art neurostimulation and neuroimaging techniques. Specifically, magnetic resonance imaging–based lesion symptom mapping (18) in individuals poststroke and functional near infrared spectroscopy (5) or electroencephalography (31) during split-belt walking could provide more insights in the (lack of) cortical and cerebellar involvement in perception of gait asymmetry.
The author thanks Jacques Duysens, Asher Straw, and Rodger Kram for helpful feedback and comments regarding an earlier version of this manuscript.
1. Bastian AJ. Understanding sensorimotor adaptation
and learning for rehabilitation. Curr. Opin. Neurol
. 2008; 21:628–33.
2. Brownstone RM, Bui TV, Stifani N. Spinal circuits for motor learning. Curr. Opin. Neurobiol
. 2015; 33:166–73.
3. Capaday C, Stein RB. Difference in the amplitude of the human soleus H reflex during walking and running. J. Physiol
. 1987; 392:513–22.
4. Choi JT, Bastian AJ. Adaptation
reveals independent control networks for human walking. Nat. Neurosci
. 2007; 10:1055–62.
5. Clark DJ. Automaticity of walking: functional significance, mechanisms, measurement and rehabilitation strategies. Front. Hum. Neurosci
. 2015; 9:246.
6. Courtine G, Papaxanthis C, Schieppati M. Coordinated modulation of locomotor muscle synergies constructs straight-ahead and curvilinear walking in humans. Exp. Brain Res
. 2006; 170:320–35.
7. Dietz V, Zijlstra W, Duysens J. Human neuronal interlimb coordination during split-belt locomotion
. Exp. Brain Res
. 1994; 101:513–20.
8. Duysens J, Clarac F, Cruse H. Load-regulating mechanisms in gait and posture: comparative aspects. Physiol. Rev
. 2000; 80:83–133.
9. Eikema DJ, Chien JH, Stergiou N, et al. Optic flow improves adaptability of spatiotemporal characteristics during split-belt locomotor adaptation
with tactile stimulation. Exp. Brain Res
. 2016; 234:511–22.
10. Finley JM, Bastian AJ, Gottschall JS. Learning to be economical: the energy cost of walking tracks motor adaptation
. J. Physiol
. 2013; 591:1081–95.
11. Forssberg H, Grillner S, Halbertsma J, Rossignol S. The locomotion
of the low spinal cat. II. Interlimb coordination. Acta Physiol. Scand
. 1980; 108:283–95.
12. Frigon A, Hurteau MF, Thibaudier Y, Leblond H, Telonio A, D'Angelo G. Split-belt walking alters the relationship between locomotor phases and cycle duration across speeds in intact and chronic spinalized adult cats. J. Neurosci
. 2013; 33:8559–66.
13. Fujiki S, Aoi S, Funato T, Tomita N, Senda K, Tsuchiya K. Adaptation
mechanism of interlimb coordination in human split-belt treadmill walking through learning of foot contact timing: a robotics study. J. R. Soc. Interface
. 2015; 12:0542.
14. Helm EE, Reisman DS. The split-belt walking paradigm: exploring motor learning and spatiotemporal asymmetry poststroke. Phys. Med. Rehabil. Clin. N. Am
. 2015; 26:703–13.
15. Hoogkamer W, Bruijn SM, Duysens J. Gait parameters affecting the perception threshold of locomotor symmetry: comment on Lauzière, et al. (2014). Percept. Mot. Skills
. 2014; 119:474–7.
16. Hoogkamer W, Bruijn SM, Duysens J. Stride length asymmetry in split-belt locomotion
. Gait Posture
. 2014; 39:652–4.
17. Hoogkamer W, Bruijn SM, Potocanac Z, Van Calenbergh F, Swinnen SP, Duysens J. Gait asymmetry during early split-belt walking is related to perception of belt speed difference. J. Neurophysiol
. 2015; 114:1705–12.
18. Hoogkamer W, Bruijn SM, Sunaert S, Swinnen SP, Van Calenbergh F, Duysens J. Adaptation
and aftereffects of split-belt walking in cerebellar lesion patients. J. Neurophysiol
. 2015; 114:1693–704.
19. Hoogkamer W, O'Brien MK. Sensorimotor
recalibration during split-belt walking: task-specific and multisensory? J. Neurophysiol
. 2016; 116:1539–41.
20. Jayaram G, Galea JM, Bastian AJ, Celnik P. Human locomotor adaptive learning is proportional to depression of cerebellar excitability. Cereb. Cortex
. 2011; 21:1901–9.
21. Jayaram G, Tang B, Pallegadda R, Vasudevan EV, Celnik P, Bastian A. Modulating locomotor adaptation
with cerebellar stimulation. J. Neurophysiol
. 2012; 107:2950–7.
22. Jensen L, Prokop T, Dietz V. Adaptational effects during human split-belt walking: influence of afferent input. Exp. Brain Res
. 1998; 118:126–30.
23. Lauzière S, Miéville C, Duclos C, Aissaoui R, Nadeau S. Perception threshold of locomotor symmetry while walking on a split-belt treadmill in healthy elderly individuals. Percept. Mot. Skills
. 2014; 118:475–90.
24. McFadyen BJ, Hegeman J, Duysens J. Dual task effects for asymmetric stepping on a split-belt treadmill. Gait Posture
. 2009; 30:340–4.
25. Malone LA, Bastian AJ, Torres-Oviedo G. How does the motor system correct for errors in time and space during locomotor adaptation
? J. Neurophysiol
. 2012; 108:672–83.
26. Marques B, Colombo G, Müller R, Dürsteler MR, Dietz V, Straumann D. Influence of vestibular and visual stimulation on split-belt walking. Exp. Brain Res
. 2007; 183:457–63.
27. Morton SM, Bastian AJ. Cerebellar contributions to locomotor adaptations during splitbelt treadmill walking. J. Neurosci
. 2006; 26:9107–16.
28. Mukherjee M, Eikema DJ, Chien JH, et al. Plantar tactile perturbations enhance transfer of split-belt locomotor adaptation
. Exp. Brain Res
. 2015; 233:3005–12.
29. Ogawa T, Kawashima N, Obata H, Kanosue K, Nakazawa K. Distinct motor strategies underlying split-belt adaptation
in human walking and running. PLoS One
. 2015; 10:e0121951.
30. Ogawa T, Kawashima N, Ogata T, Nakazawa K. Predictive control of ankle stiffness at heel contact is a key element of locomotor adaptation
during split-belt treadmill walking in humans. J. Neurophysiol
. 2014; 111:722–32.
31. Oliveira AS, Schlink BR, Hairston WD, König P, Ferris DP. Proposing metrics for benchmarking novel EEG technologies towards real-world measurements. Front. Hum. Neurosci
. 2016; 10:188.
32. Reisman DS, Bastian AJ, Morton SM. Neurophysiologic and rehabilitation insights from the split-belt and other locomotor adaptation
paradigms. Phys. Ther
. 2010; 90:187–95.
33. Reisman DS, Block HJ, Bastian AJ. Interlimb coordination during locomotion
: what can be adapted and stored? J. Neurophysiol
. 2005; 94:2403–15.
34. Reisman DS, McLean H, Keller J, Danks KA, Bastian AJ. Repeated split-belt treadmill training improves poststroke step length asymmetry. Neurorehabil. Neural Repair
. 2013; 27:460–8.
35. Reisman DS, Wityk R, Silver K, Bastian AJ. Locomotor adaptation
on a split-belt treadmill can improve walking symmetry post-stroke. Brain
. 2007; 130(Pt 7):1861–72.
36. Sawers A, Kelly VE, Hahn ME. Effects of gradual versus sudden training on the cognitive demand required while learning a novel locomotor task. J. Mot. Behav
. 2013; 45:405–14.
37. Thelen E, Ulrich BD, Niles D. Bilateral coordination in human infants: stepping on a split-belt treadmill. J. Exp. Psychol. Hum. Percept. Perform
. 1987; 13:405–10.
38. Torres-Oviedo G, Vasudevan E, Malone L, Bastian AJ. Locomotor adaptation
. Prog. Brain Res
. 2011; 191:65–74.
39. Vazquez A, Statton MA, Busgang SA, Bastian AJ. Split-belt walking adaptation
estimates of leg speed, but not position or force. J. Neurophysiol
. 2015; 114:3255–67.
40. Wutzke CJ, Faldowski RA, Lewek MD. Individuals poststroke do not perceive their spatiotemporal gait asymmetries as abnormal. Phys. Ther
. 2015; 95:1244–53.
41. Wutzke CJ, Mercer VS, Lewek MD. Influence of lower extremity sensory function on locomotor adaptation
following stroke: a review. Top. Stroke Rehabil
. 2013; 20:233–40.
42. Zhang XY, Wang JJ, Zhu JN. Cerebellar fastigial nucleus: from anatomic construction to physiological functions. Cerebellum Ataxias
. 2016; 3:9.