A model of the cerebellum generates gait adaptations in a reflex-based neuromusculoskeletal model during split-belt walking

Sophie Fleischmann, Julian Shanbhag, Jörg Miehling, Sandro Wartzack, Carmichael Ong, Bjoern M. Eskofier, Anne D. KoelewijnbioRxiv 2024 · DOI

TL;DR

This paper introduces a novel neuromusculoskeletal model that integrates a biologically inspired cerebellum to simulate how humans adapt their gait to split-belt treadmill walking. The model successfully replicates key aspects of gait adaptation, demonstrating that timing adjustments alone can drive adaptation, offering new insights into cerebellar function in motor control.

What Problem Does This Paper Address?

Split-belt treadmill walking, where each foot walks on a belt moving at a different speed, is a common experimental paradigm to study how humans adapt their gait. Neurotypical individuals gradually adjust their step length asymmetry (SLA) to compensate for the perturbation, a process thought to involve the cerebellum. While existing neuromusculoskeletal models can simulate walking, they struggle to predict these continuous, adaptive changes. This paper aims to bridge this gap by developing a model that can capture these complex gait adaptations, thereby providing a better understanding of the cerebellum's role in motor learning and adaptation.

Methods

The authors extended a pre-existing spinal reflex model of walking with a biologically inspired cerebellar model. This cerebellar model operates on the principle of error-based adaptation: it predicts the outcome of motor commands, compares it to the actual sensory feedback, and uses the resulting prediction error to update its internal model and adjust motor commands. Crucially, in this study, the cerebellum's adaptation was limited to modulating only the timing of swing initiation in each leg, keeping other reflex parameters constant. This allowed for a focused investigation of timing's contribution to gait adaptation. The model was then simulated walking on a split-belt treadmill with a 2:1 speed ratio for 80 seconds, and its performance was compared to a reflex-only model and experimental data.

Datasets

Not discussed in the paper. The study relies on simulations and comparisons to existing experimental literature.

Key Insights and Ideas

  • Timing is Key (for Temporal Adaptation): The study demonstrates that adapting only the timing of gait phases (specifically, swing initiation) is sufficient to drive exponential adaptation in the temporal domain of gait. This is a significant finding, suggesting that precise timing control is a fundamental mechanism for adaptation.
  • Spatial Adaptation Needs More: While timing adaptation drove temporal changes, it only resulted in linear changes in the spatial domain. This implies that to fully replicate human spatial gait adaptations, additional control mechanisms (like reflex gain changes) are likely necessary, hinting at a hierarchical control structure.
  • Cerebellum as a Predictive Engine: The model reinforces the idea of the cerebellum acting as a forward model that predicts motor outcomes and uses prediction errors to drive adaptation, a concept well-established in reaching tasks but now extended to locomotion.
  • Modular Approach for Understanding: By isolating the adaptation of swing initiation timing, the researchers could dissect the contribution of a single mechanism, a powerful advantage of neuromusculoskeletal modeling over purely experimental approaches.

Key Results

  • The reflex-only model could walk on the split-belt treadmill but showed no step length asymmetry (SLA) adaptation.
  • The extended model with the cerebellum successfully adapted SLA, moving from an initial asymmetry of -0.17 towards -0.03 by the end of the simulation, mirroring experimental observations.
  • The adaptation followed an exponential time course, with the fast leg contributing more significantly to the adaptation than the slow leg.
  • Both spatial (center of oscillation difference) and temporal (limb phasing) components contributed to the SLA adaptation.
  • Simulated joint kinematics closely matched experimental data, though some deviations were observed, particularly in the fast ankle trajectory.

Limitations and Future Work

The authors acknowledge several limitations:

  • Manual Readjustment of Initial Guess: Certain reflex gains had to be manually adjusted to obtain an initial guess that approximated experimental observations of asymmetry. More sophisticated methods for finding initial guesses are suggested.
  • No Capture of Reactive Responses: The model does not capture the rapid, reactive responses that occur immediately at the onset of the perturbation or during transitions between adaptation and washout.
  • Simplified Cerebellar Model: The model represents a computational abstraction of the cerebellum, and the concept of single learning and adaptation rates might not perfectly reflect biological reality.
  • Limited Scope of Adaptation: The study focused solely on swing initiation timing. Future work could explore the interplay with reflex changes and other potential adaptation mechanisms.
  • Deviations in Kinematics: While overall kinematics matched well, some deviations in joint trajectories (e.g., fast ankle) and RMSE values suggest room for improvement in the spinal controller or objective function.

Future work could involve extending the model to include other locomotor tasks, investigating the role of central pattern generators (CPGs), and further refining the spinal control and objective functions to better capture experimental data. The framework is designed to be modular and can be extended to test various hypotheses about motor control and adaptation.

Summary updated

Abstract

Abstract Background During split-belt treadmill walking, neurotypical humans exhibit adaptations characterized by a gradual decrease in step length asymmetry (SLA) toward or beyond symmetry, whereas individuals with cerebellar damage do not show these motor adaptations. Neuromusculoskeletal simulations may help to better understand individual aspects of the underlying neural control, but are currently incapable of predicting adaptations to the continuous perturbations imposed by split-belt walking. Methods We extend a spinal reflex model with a biologically inspired model of the cerebellum, which enables error-based motor adaptation by modulating spinal control parameters in response to mismatches between a continuously updated internal prediction and actual motor outcomes. In this work, the cerebellum modulates only a single spinal control parameter, the timing of swing initiation in each leg, which allows examining its isolated contribution to gait adaptation as all other reflex pathways are held constant. We created 80 s predictive simulations of the model walking on a split-belt treadmill with a 2:1 belt-speed ratio, and compared predicted spatiotemporal parameters and kinematics to the reflex-only model and literature. Results The reflex-only model could walk on the split-belt treadmill, but showed no step length adaptations. In contrast, the extended model adapted SLA from an initial asymmetric value toward symmetry or beyond, following an exponential time course similar to that observed in experiments. The model could adapt at varying rates and converge to different asymmetry levels. We found that, in simulation, SLA adaptation during split-belt walking is possible without changes in reflex gains, by adapting the timing of swing initiation. The modulation of timing alone also predicted the experimentally observed exponential adaptation in the temporal domain, but only a linear change in the spatial domain, indicating that additional control mechanisms are likely required to reproduce the full spatial adaptation observed in split-belt walking. Conclusion We propose a computational model of the cerebellum which, when integrated into a spinal reflex model, autonomously drives feedforward gait adaptations during split-belt walking. This advances the current state of predictive simulations and may eventually help to better understand specific adaptation processes. The modular framework can be extended to test different hypotheses about motor control and adaptation during continuous perturbation tasks.
BibTeX
@misc{fleischmann2024model,
  author = {Sophie Fleischmann and Julian Shanbhag and Jörg Miehling and Sandro Wartzack and Carmichael Ong and Bjoern M. Eskofier and Anne D. Koelewijn},
  title = {{A model of the cerebellum generates gait adaptations in a reflex-based neuromusculoskeletal model during split-belt walking}},
  year = {2024},
  doi = {10.1101/2024.12.12.628122},
  howpublished = {bioRxiv},
}
Figure