Skip to main content
Dryad logo

Different functional networks underlying human walking with pulling force fields acting in forward or backward directions


Ogawa, Tetsuya et al. (2022), Different functional networks underlying human walking with pulling force fields acting in forward or backward directions, Dryad, Dataset,


Walking with pulling force fields acting at the body center of mass (in the forward or backward directions) is compatible with inclined walking and is used in clinical practice for gait training. From the perspective of known differences in the motor strategies that underlie walking with the respective force fields, the present study elucidated whether the adaptation acquired by walking on a split-belt treadmill with either one of the force fields affects subsequent walking in other directions. Walking with the force field induced an adaptive and de-adaptive behavior of the subjects, with the aspect evident in the anterior breaking and posterior propulsive impulses of the ground reaction force as parameters. In the parameters, the adaptation acquired during walking with a force field acting in one direction was transferred to that in the opposite direction only partially. Furthermore, the adaptation that occurred while walking in a force field in one direction was rarely washed out by subsequent walking in a force field in the opposite direction and thus was maintained independently of the other. These results demonstrated possible independence in the neural functional networks capable of controlling walking in each movement task with an opposing force field.



Sixteen volunteers (15 males and 1 female; mean ± SD age, 28.6 ± 6.6 y; weight, 66.0 ± 12.0 kg) without a history of neurological or orthopedic disorders were included in this study. Each participant was tested using two of the four experimental protocols (Fig. 1B). Eight of them participated in Experiments 1 and 3, while the other eight participated in Experiments 2 and 4, with the order of participation randomly distributed among subjects to overcome any ordering effects. All participants were naïve to the purpose of the study and provided written informed consent before participation. All experimental procedures were approved by the local Ethics Committee of the School of Arts and Sciences of the University of Tokyo and were conducted following the Declaration of Helsinki.


The experiments consisted of walking under one of two physical conditions (either with an “aiding” the force field or an “impeding” force field, see Fig. 1A). Force fields were applied to the participants via a belt stranded around the torso near the COM, which was then attached to the counterweight (2 kg) through two carabiners, a cable, and two low friction pulleys. Subjects were pulled horizontally forward in the aiding force field, while a backward pull was applied in the impeding force field.

The participants were instructed to walk on a split-belt treadmill (Bertec, Columbus, OH, USA) with two separate belts, and the speed of each belt was controlled independently. In the present experiment, the treadmill was operated under one of two conditions, tied (two belts moving together at the same speed) or split (separately at different speeds), using a custom-written computer program written in Lab-VIEW (National Instruments, Austin, Texas, USA). The speeds were set at 0.75 ms-1 for both belts under the tied, and while in the split, the belt on the left side was 0.5 ms-1 and another on the right was 1.0 ms-1 (ratio, 1:2). The limb on the slower (left) side speed of the treadmill under the split was defined as the “slow limb” and the limb on the faster (right) side speed as the “fast limb.”

The experimental protocols consisted of baseline, adaptation, re-adaptation, and three washout periods, as dictated by the protocol (Fig. 1B). Participants always accompanied one of the two force fields (impeding or aiding) throughout the experiments. During the baseline, the treadmill was tied and participants walked with the impeding and aiding force fields for 1 min each. The treadmill was then operated in a split and the participants underwent a 10-min adaptation, followed by a 1-min catch trial (washout 1) to walk on the treadmill in tied. The treadmill was then returned to split, and the participants again underwent adaptation to walking on a split-belt (re-adaptation) for 5 min, which was again followed by a 1-min catch trial (washout 2) to walk on a tied belt. The force fields in the two catch trial periods (washouts 1 and 2) were different in direction (impeding washout 1 and aiding in washout 2 or vice versa) to address both the degree of adaptation by evaluating the magnitude of the aftereffect and how it could transfer to walking with the opposite force field. In Experiment 1, for example, the degree of adaptation was tested by assessing the magnitude of the aftereffect while walking with the aiding force field on the tied belt during washout 1 (catch trial) after adapting to walk on the split-belt with the force field in the same (aiding) direction. The transfer of adaptation, on the other hand, was tested by walking with the impeding force field in washout 2 (post-adaptation) after adapting to walking (re-adaptation period) with the force field in the opposite (aiding) direction. Given that the emergence of the aftereffect is not stable but can decay throughout the experiments (Ogawa et al. 2015; 2018) (4,12), the order of exposure to the washout periods with different force fields was alternated depending on the experiments (between experiments 1 and 2, 3, and 4, respectively) to overcome possible ordering effects.

In Experiments 1 and 3, subjects underwent an extra washout of three periods to walk on a tied belt with force fields in the same direction as the adaptation periods, but were different from those during the washout 2 period. The purpose of this additional washout period was to evaluate the degree to which the adaptation acquired through walking on the split-belt could be maintained (or washed out) after walking with a force field in a different direction. Between each testing period, upon changing the belt speeds and/or direction of the force field, there was a 15-s time interval in which subjects stepped on platforms on both sides of the treadmill. They were then allowed to step on the treadmill with the left leg when a sufficient belt speed was reached and the appropriate force fields were mounted. During the experiments, subjects were instructed to walk while watching a wall approximately 3 m in front of them and not to look down at the belts. They were allowed to hold onto the handrails mounted on either side of the treadmill in case of risk of falling. All subjects completed the test sessions without holding on. To ensure safety, one experimenter stood on the treadmill.

Data recording and analysis

Force sensors mounted underneath each treadmill belt were used to determine the dimensional ground reaction force (GRF) components: mediolateral (Fx), anteroposterior (Fy), and vertical (Fz). Force signals were sampled at 1 kHz, stored on a computer via an analog-to-digital converter, and low-pass filtered at a cut-off frequency of 8 Hz (Power Lab; AD Instruments, Sydney, Australia). The magnitude of the GRF components was evaluated for each stride cycle. The timing of foot contact and toe-off for each stride cycle was determined based on the vertical Fz component of the GRF for both fast and slow sides using custom-written software (VEE Pro 9.3, Agilent Technologies, Santa Clara, CA, USA).

To address the degree of adaptation and transfer of motor patterns across walking with force fields in opposite directions, the degree of asymmetry in the anteroposterior (Fy) component of the GRF was calculated for each stride cycle of walking. As depicted in Fig. 2A, this GRF component includes anterior (breaking) and posterior propulsive components that appear at different phases during the gait cycle. For each component, the peak amplitude during each gait cycle was calculated as the absolute value for both the fast and slow sides (upper panels of Figs. 3 and 4). The degree of asymmetry, which represents the difference in the absolute values, was then calculated by subtracting the value of the slow limb from that of the fast limb on a stride-by-stride basis (lower panels of Figs. 3 and 4).

As exposure to different force fields (aiding and impeding) influences the magnitude of the GRF components during walking, it does not allow for direct comparisons of the degree of asymmetry between walking with different force fields. In addition, to consider the influence of the natural walking movement of the subjects, which is not perfectly symmetrical and allows for comparisons between different force fields, the obtained degree of asymmetry underwent a normalization process. For both the anterior and posterior components of the GRF, the degree of asymmetry obtained in the washout periods (1, 2, and 3) was subtracted from the mean values of those under the respective baseline with different force fields. The normalized values were then divided into bins of 5 s and averaged for each bin.


Two-way analysis of variance (ANOVA) with repeated measures was used to test for statistically significant differences in the degree of asymmetry (in terms of both acquisition and transfer of adaptation) between walking with different force fields (either aiding or impeding) and different time periods of the experiment ( initial or final phase of washout periods). When ANOVA revealed significant results, Bonferroni’s post hoc comparisons were performed to identify significant differences between variables. In addition, to test whether the adaptation acquired during walking with one force field was washed out (or maintained) by walking with the other force field, a paired Student’s t-test was performed to compare the degree of asymmetry between the final phase of washout 2 and the initial phase of washout 3 periods. A paired Student’s t-test was used to compare the magnitude of the GRF components and the cadence between walking under the two force conditions. Data are presented as mean ± standard error of the mean (SEM) values. Statistical significance was set at P < 0.05.


Japan Society for the Promotion of Science