One of the fundamental motor skills during human locomotion is maintaining body balance, for which the visual system, among various sensory systems, is mainly responsible (Hallemans et al., 2010, Patla, 1997). This sensory input is critical to control specific gait characteristics since its absence could cause changes in step-time parameters, upper body acceleration, ground reaction forces and neural processing (Hallemans et al., 2009a, Iosa et al., 2012, Oliveira et al., 2017, Shoja et al., 2020). These changes are probably adopted as a strategy to control the center of mass (COM), which has been considered as an important control variable during normal walking (Black et al., 2007, Papi et al., 2015). Relative few studies considered the relationship between COM control and different gait parameters during eyes-closed walking (Hallemans et al., 2010, Shoja et al., 2020, Yang and Pai, 2014) and little is known about the underlying control mechanisms of multi-joint coordination. Coordination patterns are essential movement features and may serve as objective measurements to characterize movement quality in neurorehabilitation (Solnik et al. 2020). Thus, an important next step is to characterize the coordination of body segments in stabilizing COM (Latash et al., 2007) during no vision walking.

Coordination patterns during walking has often been studied using principal component analysis, i.e. by reducing high-dimensional data sets to a small number of modes (Daffertshofer et al., 2004). With this analysis, visual feedback does not appear to influence the intersegmental correlation during walking (Hallemans and Aerts, 2009). However, by this apporach variance structure that is not caused by correlation may be undetected. In fact, there are evidences that motor variability is not compressed uniformly, but rather is constrained to a subspace that does not interfere with task performance (Domkin et al., 2002, Scholz et al., 2000, Tseng et al., 2002). To fully understand the role of vision in intersegmental coordination, different amounts of variances in different segments need to be quantified with respect to task performance.

The uncontrolled manifold (UCM) analysis provides a powerful tool to evaluate how variations in elemental variables (e.g., joint angles) covary to achieve consistent output in a task variable (e.g., COM). The variance across repetitions is partitioned into two components: one lies within the UCM (VUCM), which leads to a constant performance variable thus termed as the “good variance”, whereas the other is orthogonal to the UCM (VORT), which leads to changes in the performance variable thus termed as the “bad variance”. We define stability as the strength of exploiting the good variance and suppressing the bad variance. Thus, increased stability would indicate evidence of motor synergy that stabilizes the performance variable (e.g. greater VUCM and lower VORT). The stability can be characterized by synergy index, which is the difference between VUCM and VORT relative to the total variance (see Methods). To maintain stable gait, joint angles should covary in a coordinated manner to place the COM in certain positions in time to ensure stability, i.e., to stabilize the COM trajectory. Current study explores how the nervous system stabilizes the COM trajectory during no vision walking in the UCM framework.

Since major causes of falls result in an increase of the COM movements in the sagittal plane (Smeesters et al., 2001), this study focuses on the relationships between sagittal COM trajectory and segmental coordination. The goal is to determine how the organization of cycle-to-cycle variability underlying sagittal COM stabilization is affected by the block of vision. Previous studies observed altered kinematic joint patterns of the lower-limb after visual block, e.g., reduced ankle plantar flexion (indicating flat foot contact) and reduced hip adduction (which increases step width and base of support) in early stance phase (Hallemans et al., 2009b, Hallemans and Aerts, 2009), implying a change of multi-joint control strategy to maintain balance. We thus hypothesized that walking with restricted vision requires altered coordination strategy on stabilizing sagittal COM trajectory (Hypothesis 1). We predicted that the synergy strength would be different during no vision walking from normal walking. Since mechanical interactions at different phases or events (with single or double support) imply different stability requirements for which the nervous system deals with visual perturbations (Hallemans and Aerts, 2009), we further hypothesized that the synergy strength would change across different phases and gait events, in both visual conditions (Hypothesis 2). If our hypotheses were confirmed, the results would reveal how visual feedback alters coordination among joints during the whole walking phase, thus provide a better understanding of the relationship between vision and coordination. In addition, quantification of motor variances with respect to task performance can serve as a diagnostic tool to evaluate motor performance in patients with visual impairments, e.g. as a biomarker to measure movement quality (Solnik et al., 2020). Advancing our knowledge on the functional purposes of gait variability during no vision walking would shed lights on how the nervous system deals with perturbations of gait in general and may be relevant for rehabilitation treatment as millions of people in the world suffer from visual impairments (Hallemans et al., 2011).