Sensorimotor control of gait: a novel approach for the study of the interplay of visual and proprioceptive feedback

HOW is your doctor using this to adjust your stroke walking protocol? Has your doctor EVER adjusted any of your stroke protocols based on new research?
http://journal.frontiersin.org/article/10.3389/fnhum.2015.00014/full?
Ryan Frost¹, Jeffrey Skidmore¹, Marco Santello² and Panagiotis Artemiadis^1*

• ¹Human-Oriented Robotics and Control Lab, School for Engineering of Matter Transport and Energy, Arizona State University, Tempe, AZ, USA
• ²Neural Control of Movement Laboratory, School of Biological and Healthy Systems Engineering, Arizona State University, Tempe, AZ, USA

Sensorimotor control theories propose that the central nervous system exploits expected sensory consequences generated by motor commands for movement planning, as well as online sensory feedback for comparison with expected sensory feedback for monitoring and correcting, if needed, ongoing motor output. In our study, we tested this theoretical framework by quantifying the functional role of expected vs. actual proprioceptive feedback for planning and regulation of gait in humans. We addressed this question by using a novel methodological approach to deliver fast perturbations of the walking surface stiffness, in conjunction with a virtual reality system that provided visual feedback of upcoming changes of surface stiffness. In the “predictable” experimental condition, we asked subjects to learn associating visual feedback of changes in floor stiffness (sand patch) during locomotion to quantify kinematic and kinetic changes in gait prior to and during the gait cycle. In the “unpredictable” experimental condition, we perturbed floor stiffness at unpredictable instances during the gait to characterize the gait-phase dependent strategies in recovering the locomotor cycle. For the “unpredictable” conditions, visual feedback of changes in floor stiffness was absent or inconsistent with tactile and proprioceptive feedback. The investigation of these perturbation-induced effects on contralateral leg kinematics revealed that visual feedback of upcoming changes in floor stiffness allows for both early (preparatory) and late (post-perturbation) changes in leg kinematics. However, when proprioceptive feedback is not available, the early responses in leg kinematics do not occur while the late responses are preserved although in a, slightly attenuated form. The methods proposed in this study and the preliminary results of the kinematic response of the contralateral leg open new directions for the investigation of the relative role of visual, tactile, and proprioceptive feedback on gait control, with potential implications for designing novel robot-assisted gait rehabilitation approaches.

More at link between these two sections.

Conclusions—Discussion

Figure 6 shows that VP and PO conditions have very similar effects on the right (unperturbed) leg. Specifically, both hip and knee joint kinematics are significantly affected right after the right leg starts the swing phase at ~280 ms (20% of the average gait cycle of 1.4 s) after the start of the perturbation delivered to the left leg. This latency is consistent with our previous studies (Artemiadis and Krebs, 2011a,b; Skidmore et al., 2014b). Most importantly, these observations support the hypothesis that inter-leg coordination involves supraspinal pathways, which would account for the long delay in the response of the non-perturbed response (~280 ms). Additionally, a similar response but a longer latency (~420 ms) relative to the onset of the perturbation is observed at the ankle joint. Thus, it appears that the right leg kinematics respond to the perturbation to the contralateral leg by accelerating the swing phase and bring the foot in contact with the treadmill earlier relative to the unperturbed case. This interpretation is consistent with the earlier HS in VP and PO conditions that is facilitated by an additional flexion of the hip and knee joints, and larger dorsiflexion combined with faster plantar-flexion of the ankle joint (40–50% of the gait cycle; Figure 6).

Figure 6 also shows that there is a kinematic effect of the perturbation on the right leg kinematics also in the VO condition. Therefore, even if there is a visual “warning” of the stiffness perturbation but the perturbation never happens, the contralateral (right) leg kinematics changes in an anticipatory rather than reactive fashion. Specifically, we observe an effect on the kinematics of all joints (hip, knee, ankle) that starts at ~630 ms after the perturbation onset and just before the HS of the right leg. These kinematic effects are similar to the effects observed in the VP and PO conditions, i.e., acceleration of the swing phase and shortened stride length, which can be explained by increased knee flexion and decreased ankle plantarflexion.

Figure 7 provides an additional representation of the effects of the perturbations on the right leg kinematics. The acceleration of the gait cycle evoked by the perturbations for VP and PO conditions can be seen clearly in the phase space representation by following the clockwise rotation denoted by the arrows. The phase space representation of the three leg joints for both VP and PO conditions exhibit a distinct acceleration of joint rotations through a pattern that resembles the normal gait cycle, but shifted earlier in time through the gait cycle. For example, the loop in the knee phase space representation, in the center of the plot where the HS is included, happens earlier in the VP and PO cases, when the knee is still at −20 degrees. A similar behavior can be seen in the hip and ankle joints. It is also worth noting that the phase representation of the VP and PO cases converges to the normal one before the end of the gait cycle, which can also be seen in Figure 6.

For the VO condition, the phase representation provides further insight about two main features: (1) the evoked responses have similar characteristics to the ones associated with VP and PO perturbations, but delayed with respect to the latter; and (2) the kinematics for the VO condition converge to the normal ones within the gait cycle. It should be emphasized that the VO response in phase space resembles that observed for the VP and PO conditions in terms of acceleration profile of the gait cycle, but lies between the normal cycle and the VP and PO cycles in the phase space. The latter is obvious when examining the loop of the hip phase that includes the HS on the right bottom corner of the graph. A similar behavior can be observed in the corresponding loops of the knee and ankle joints.

The presented method of analyzing the interplay between visual and proprioceptive and tactile feedback in gait resulted in important observations. First, when there is no physical perturbation, and therefore proprioceptive feedback is not elicited, visual feedback can evoke contralateral leg responses that resemble those caused by proprioceptive feedback in response to a mechanical perturbation of the opposite leg. This leads to the validation of the hypothesis that a learnt mapping between visual and proprioceptive feedback creates or activates mechanisms, that are probably supraspinally mediated, that control inter-leg coordination.

However, evoked responses associated with only visual feedback of floor stiffness changes (VO) were significantly delayed relative to those caused by a physical perturbation. These data can be interpreted as follows. Visual cues (warning) act to mediate anticipatory/predictive control of gait, however they only evoke late responses. These responses appear to be independent from proprioceptive feedback, as suggested by time shift of visually-cued responses relative to proprioceptive-dependent responses. Moreover, our results support the existence of only late responses associated with visual feedback of upcoming changes in floor stiffness. This is supported by the observation that in the early phases of the gait cycle, VP and PO responses are almost identical, which suggests that the predictive role of visual feedback does not activate any early motor mechanisms.

The results of the present study should be considered as preliminary due to the small sample of subjects. Furthermore, more work is needed to identify the neural mechanisms underlying the observed kinematic responses of the unperturbed leg to mechanical perturbations delivered to the contralateral leg. Nevertheless, our findings are promising as they shed new light on inter-leg coordination mechanisms and open new avenues for research, However, the scope of this paper is to introduce a novel method of investigating the inter-play of visual and proprioceptive feedback in gait. The proposed method, facilitated by a novel and unique technological architecture (the VST setup), can be potentially beneficial not only for understanding sensorimotor control of gait, but also for significantly improving neural rehabilitation protocols for impaired walkers by applying the identified principles and developing model-based protocols for gait therapy.