Split-belt walking: adaptation differences between young and older adults

Abstract

Human walking is highly adaptable, which allows us to walk under different circumstances. With aging, the probability of falling increases, which may partially be due to a decreased ability of older adults to adapt the gait pattern to the needs of the environment. The literature on visuomotor adaptations during reaching suggests, however, that older adults have little problems in adapting their motor behavior. Nevertheless, it may be that adaptation during a more complex task like gait is compromised by aging. In this study, we investigated the ability of young (n = 8) and older (n = 12) adults to adapt their gait pattern to novel constraints with a split-belt paradigm. Findings revealed that older adults adapted less and more slowly to split-belt walking and showed fewer aftereffects than young adults. While young adults showed a fast adjustment of the relative time spent in swing for each leg older adults failed to do so, but instead they were very fast in manipulating swing speed differences between the two legs. We suggest that these changes in adaptability of gait due to aging stem from a mild degradation of cortico-cerebellar pathways (reduced adaptability) and cerebral structures (decreased ability to change gait cycle timing). However, an alternative interpretation may be that the observed reduced adaptation is a compensatory strategy in view of the instability induced by the split-belt paradigm.

Figures

Fig. 1.

Definition of step and stride length of fast and slow leg during the early adaptation condition (A), late adaptation condition (B), and aftereffects condition (C). Vertical lines show the path of the lateral malleolus marker during the gait cycle; most forward positions of vertical lines indicate the anterior-posterior position of the lateral malleolus position and correspond to heel strikes, while most backward positions correspond to toe off. Note: figure is only for illustrative purposes and contains no actual data.

Fig. 2.

Self-selected walking speed (A), self-selected stride times (B), self-selected stride lengths (C), and stride times during treadmill walking at 1 m/s (D) of young and older adults. Error bars represent SE.

Fig. 3.

A: step length symmetry (dimensionless) during strides 1–400 of the adaptation condition of young and older adults. Shaded areas represent SE. Episodes over which the mean was taken for statistical testing have been indicated in alternating white and gray bars. If only the group × episode effect was significant, post hoc t-tests with Bonferroni correction were performed on the difference between groups per episode and significance per episode indicated in the corresponding episode. B: stride lengths of the fast and slow leg during strides 1–400 of the adaptation. Shaded areas represent SE. Thin horizontal lines represent means of the baseline condition. Episodes over which the mean was taken for statistical testing have been indicated in alternating white and gray bars. If only the group × episode effect was significant, post hoc t-tests with Bonferroni correction were performed on the difference between groups per episode and significance per episode indicated in the corresponding episode.

Fig. 4.

A: step length symmetry (dimensionless) during strides 1–200 of the aftereffects condition of young and older adults. Shaded areas represent SE. Episodes over which the mean was taken for statistical testing have been indicated in alternating white and gray bars. If only the group × episode effect was significant, post hoc t-tests with Bonferroni correction were performed on the difference between groups per episode and significance per episode indicated in the corresponding episode. B: stride lengths of the fast and slow leg during strides 1–200 of the aftereffects condition. Shaded areas represent SE. Thin horizontal lines represent means of the baseline condition. Episodes over which the mean was taken for statistical testing have been indicated in alternating white and gray bars. If only the group × episode effect was significant, post hoc t-tests with Bonferroni correction were performed on the difference between groups per episode and significance per episode indicated in the corresponding episode.

Fig. 5.

Relationship between adaptation and aftereffects. x-Axis: step length symmetry at the end of adaptation. y-Axis: step length symmetry at the onset of the aftereffects condition. Correlation coefficients are shown (P values in parentheses).

Fig. 6.

Relationships of age with step length symmetry at the end of adaptation (A), and step length symmetry at the onset of the aftereffects condition (B).

Fig. 7.

A: % of swing time of the fast and slow leg during strides 1–400 of the adaptation condition. Shaded areas represent SE. Thin horizontal lines represent means of the baseline condition. Episodes over which the mean was taken for statistical testing have been indicated in alternating white and gray bars. If only the group × episode effect was significant, post hoc t-tests with Bonferroni correction were performed on the difference between groups per episode and significance per episode indicated in the corresponding episode (asterisk). B: average swing speed of fast and slow legs during strides 1–400 of the adaptation condition. Shaded areas represent SE. Thin horizontal lines represent means of the baseline condition. Episodes over which the mean was taken for statistical testing have been indicated in alternating white and gray bars. If only the group × episode effect was significant, post hoc t-tests with Bonferroni correction were performed on the difference between groups per episode and significance per episode indicated in the corresponding episode (asterisk).