Professional Documents
Culture Documents
42nd Annual Meeting of the American Society of Biomechanics, Rochester, MN, USA, August 8th – 11th, 2018
consecutive steps via state space reconstruction and
Wolf et al’s algorithm. Briefly, the state space was
reconstructed by determining the appropriate time
Mediolateral LyE
delay and embedding dimension via the average
Step width β
mutual information and far nearest neighbors
method, respectively [6]. Following the state space
reconstruction, Wolf et al’s algorithm was
implemented to compute the largest Lyapunov
exponent of the mediolateral plane trajectories in
state space [6].
Figure 1. Step width β (blue) was affected only by
To test our hypotheses, we used a four way
the initial level of difficulty of the attention-
ANOVA with repeated measures (p≤ 0.05) to test
demanding task. The divergence of the trunk
for significant differences in step width β and
trajectory in the mediolateral plane (red) was
mediolateral LyE. Then paired t-tests (p≤ 0.05)
insensitive to the attention-demanding task.*p<0.05
were used to determine specific differences between
the baseline and 2-digit conditions, 2-digit and 3-
digit conditions, then 3-digit and 4-digit conditions. CONCLUSIONS
A single attention-demanding task that can be
RESULTS AND DISCUSSION
performed during gait and for which the difficulty
The step width β of three participants were removed may be manipulated, is attractive from both clinical
before the final analysis because the data did not and research perspectives. Further aspects related
show a linear pattern. Step width β shifted away to the parameters of the attention-demanding task
from 1/f as a function of the attention-demanding used in the present study, including the number of
task. The four way ANOVA with repeated measures digits in the string, the frequency with which the
was significant (p=0.03). However, step width β, digits are displayed and the time available to
was significantly higher for the 2-digit task provide the correct response, are the focus of further
compared to the baseline (p = 0.01; Figure 1) but study.
not for the other comparisons.
REFERENCES
These results are consistent with a previous study 1. Hasson CJ. PLoS Comput Biol. 12(8):e1005044 ,
that showed a shift in step width β away from 1/f 2016
[3]. However, contrary to our expectations, the shift 2. Faisal AA et al. Nat Rev Neurosci. 9: doi:
away from 1/f did not increase as the difficulty of 10.1038/nm2258, 2008
the attention-demanding increased. Furthermore, the 3. Grabiner MD et al. Gait Posture. In press
mediolateral LyE appeared to be insensitive to the 4. Hurt CP et al. Gait Posture. 31: 461 – 464, 2010
performance of the attention-demanding task 5. Kahneman D. Attention and Effort. Prentice-Hall
(p>0.05). Inc, 1973
6. Kurz MJ et al. Nonlinear Dynamics Psychol Life
The results merit consideration in the context of the Sci. 14, 151 – 164, 2010
relatively small sample of healthy young older
adults that participated and whose results
demonstrated relatively large variability. More
importantly, perhaps, is that the method used to
increase the attention-demanding did not appear to
achieve its purpose, which was to impose a dose-
response of the dependent variables.
42nd Annual Meeting of the American Society of Biomechanics, Rochester, MN, USA, August 8th – 11th, 2018