Progress in Motor Control: From Neuroscience to Patient Outcomes

Approx.242 pages

• Translates the principles of motor control to improve sensorimotor outcomes in patients
• Reviews coordination topics including locomotor coordination, visual perception and head stability
• Explores movement analysis knowledge in rehabilitative tools

• Language: English
• Publisher: Academic Press
• Release date: Nov 17, 2023
• ISBN: 9780443239861

Book preview

9780443239861_FC

Progress in Motor Control

From Neuroscience to Patient Outcomes

First Edition

Mindy F. Levin

School of Physical and Occupational Therapy, Faculty of Medicine and Health Sciences, McGill University, Montreal, QC, Canada

Maurizio Petrarca

Movement Analysis and Robotics Laboratory, Bambino Gesu` Children’s Hospital, Research Institute, Rome, Italy

Daniele Piscitelli

Doctor of Physical Therapy Program, Department of Kinesiology, University of Connecticut, Storrs, CT, United States

Susanna Summa

Movement Analysis and Robotics Laboratory, Bambino Gesu` Children’s Hospital, Research Institute, Rome, Italy

publogo

Table of Contents

Cover image

Title page

Copyright

Contributors

Preface

Contents

Acknowledgments

References

Section A: Action perception coupling

Chapter 1 The equilibrium-point hypothesis: A major framework for the understanding of action and perception

Abstract

Introduction

Conclusions

References

Chapter 2 Synergic control of movement: From single muscles to the whole body

Abstract

The concept of synergy and its history

The two aspects of synergies

Neural control with spatial referent coordinates

Synergies in spaces of control variables

Synergies in spaces of motor units

Synergies in whole-body actions

Lessons from clinical and subclinical studies

Synergies in kinesthetic perception

Emerging issues and challenges

References

Chapter 3 Can nonlinear analysis of movement patterns reveal the status of the musculoskeletal system?

Abstract

Introduction

Capturing adaptive changes in response to perturbations

The effects of sleep deprivation on the structure of the postural control

The effects of muscle fatigue on the regularity of running

The effects of strenuous activities on the movement’s coordinative structure

Final considerations

References

Chapter 4 Toward a neural theory of goal-directed reaching movements

Abstract

Introduction

Functional principles for a neural theory of reaching movement

Skeleton of a neural theory of reaching movements

Discussion

References

Section B: Coordination

Chapter 5 The perception-action coupling in collective dynamics

Abstract

Introduction: Perception and action

The perception-action coupling

Behavioral dynamics

Complexities

Collective dynamics

Conclusion

References

Chapter 6 Locomotor coordination, visual perception, and head stability

Abstract

Introduction

Reflexive mechanisms and head control and locomotion

Shock attenuation and head stability

Whole body adaptations and head stability during unconstrained forward locomotion

Whole body adaptations and head control during reorienting

Head control and visual task demands

Locomotor asymmetries and head control

Summary and conclusions

References

Chapter 7 Computational joint action: From emergent coordination to artificial partners

Abstract

Introduction

Emergent coordination

Artificial partners

Conclusions

References

Section C: Translation of motor learning principles and rules of interaction

Chapter 8 High-fidelity interfacing for bionic rehabilitation

Abstract

Introduction

Challenges of establishing bionic interfaces

Interfacing neuromuscular structures

Bionic interfacing solutions

Surgical approaches for advanced interfacing

Toward high-fidelity interfaces

Summary

References

Chapter 9 Exploring to learn synergies and its applications in injuries affecting the upper limb

Abstract

Acknowledgments

Introduction

What is learned during motor learning?

Synergies at the level of the movement system

Motor learning as searching for new synergies

Could rehabilitation benefit from focusing on learning new synergies?

Learning new synergies in rehabilitation

Conclusion

References

Chapter 10 Translating movement analysis knowledge in rehabilitative tools

Abstract

Acknowledgments

Gait initiation

Gait sequence

Funding

References

Section D: Goal-oriented action

Chapter 11 Translation of principles of motor control to improve sensorimotor outcomes following brain injury

Abstract

Introduction

Controversies in the understanding of motor control principles—A shift in the paradigm for the understanding of how motor actions are controlled: Parametric control

The concept of abundance in movement production

The set of motor equivalent actions is decreased after central nervous system lesions

A new biomarker of sensorimotor pathology

Clinical relevance of the TSRT measure

The corticospinal tract is involved in the regulation of TSRT

Spasticity may mask the capacity for normal motor learning

Implications for training

References

Chapter 12 Goal-oriented action: New perspectives with special emphasis on neurorehabilitation

Abstract

Acknowledgments

Introduction

Funding

References

Index

Copyright

Academic Press is an imprint of Elsevier

125 London Wall, London EC2Y 5AS, United Kingdom

525 B Street, Suite 1650, San Diego, CA 92101, United States

50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States

The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom

Copyright © 2024 Elsevier Inc. All rights reserved, including those for text and data mining, AI training, and similar technologies.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

ISBN: 978-0-443-23987-8

For information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

fm01-9780443239878

Publisher: Nikki Levy

Acquisitions Editor: Joslyn Chaiprasert-Paguio

Editorial Project Manager: Pat Gonzalez

Production Project Manager: Swapna Srinivasan

Cover Designer: Victoria Esser Pearson

Typeset by STRAIVE, India

Contributors

Numbers in parenthesis indicate the pages on which the authors' contributions begin.

Mariana R.C. Aquino 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Priscila A. Araújo 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Laura Bandini 167     Department of Informatics, Bioengineering, Robotics and Systems Engineering, University of Genoa, Genoa, Italy

Lukas Bildheim 71     Faculty for Computer Science, Institute for Neural Computation, Ruhr-University Bochum, Bochum, Germany

Raoul M. Bongers 261     Department of Human Movement Sciences, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands

Renatha Carvalho 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Camila G.M. Castor 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Vinil T. Chackochan 167     Department of Informatics, Bioengineering, Robotics and Systems Engineering, University of Genoa, Genoa, Italy; Odstock Medical Ltd, Salisbury, United Kingdom

Anatol G. Feldman 3     Department of Neurosciences, University of Montreal; Feil/Oberfeld Research Centre, Jewish Rehabilitation Hospital, Center for Interdisciplinary Research in Rehabilitation (CRIR), Montreal, QC, Canada

Sergio T. Fonseca 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department; Centro de Treinamento Esportivo, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Joseph Hamill 139     Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA, United States

Mark L. Latash 25     Department of Kinesiology, The Pennsylvania State University, University Park, PA, United States

Mindy F. Levin 325     School of Physical and Occupational Therapy, McGill University; Feil/Oberfeld Research Centre, Jewish Rehabilitation Hospital, Centre for Interdisciplinary Research in Rehabilitation of Greater Montreal, Montreal, QC, Canada

C. Dane Napoli 139     Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA, United States

Juliana M. Ocarino 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department; Centro de Treinamento Esportivo, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Liria A. Okai-Nobrega 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Maurizio Petrarca 283     Movement Analysis and Robotics Laboratory (MARLab), Neurorehabilitation and Robotics Department, Bambino Gesù Children's Hospital, IRCCS, Rome, Italy

Renan A. Resende 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department; Centro de Treinamento Esportivo, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Vittorio Sanguineti 167     Department of Informatics, Bioengineering, Robotics and Systems Engineering, University of Genoa, Genoa, Italy

Thiago R.T. Santos 49     Faculty of Physical Education and Physical Therapy, Universidade Federal de Uberlândia, Uberlândia, Brazil

Gregor Schöner 71     Faculty for Computer Science, Institute for Neural Computation, Ruhr-University Bochum, Bochum, Germany

Thales R. Souza 49     Graduate Program in Rehabilitation Sciences, Physical Therapy Department, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil

Richard E.A. van Emmerik 139     Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA, United States

Cecilia De Vicariis 167     Department of Informatics, Bioengineering, Robotics and Systems Engineering, University of Genoa, Genoa, Italy

Ivan Vujaklija 213     Department of Electrical Engineering and Automation, Aalto University, Espoo, Finland

William H. Warren 105     Department of Cognitive, Linguistic, and Psychological Sciences, Brown University, Providence, RI, United States

Carolee J. Winstein 349     Biokinesiology and Physical Therapy, Herman Ostrow School of Dentistry, Department of Neurology, Keck School of Medicine, University of Southern California, Los Angeles, CA, United States

Samuel R. Zeff 139     Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA, United States

Lei Zhang 71     Faculty for Computer Science, Institute for Neural Computation, Ruhr-University Bochum, Bochum, Germany

Preface

Continuing in the tradition of the book series Progress in Motor Control, started in 1998 with the publication of the first volume of, Bernstein’s Traditions in Movement Sciences by Human Kinetics (Latash, 1998), this eighth book in the series represents the most up-to-date knowledge in motor control with a focus on the translation of basic neuroscience concepts to clinical rehabilitation outcomes. The Progress in Motor Control series addresses fundamental notions of motor control originally described by Nikolai Alexandrovich Bernstein (1899–1966), a pioneer of motor control neuroscience who worked in the former Soviet Union. Bernstein wrote extensively on the relationship between structure and function in the central nervous system. The English translation of his seminal 1935 paper has influenced scientific inquiry and the development of motor control theory since it was published in 1967 (Bernstein, 1967). For example, the motion capture technology we employ today for the analysis of kinematics and kinetics traces its origins back to Bernstein’s novel method for observing movements with the kimocyclograph. This method was inspired by the pioneering work on photography by Etienne-Jules Marey (1830–1904) and Eadweard J. Muybridge (1830–1904). Bernstein achieved a remarkable frequency of 500 Hz in recording movements. The kimocyclograph was employed in the renowned study of professional blacksmiths, which served as the foundation for one of the central challenges in motor control: the problem of motor redundancy. Aside from being a neuroscientist, Bernstein was a keen observer of human movement and had extensive experience with movement disorders in clinical populations. His earliest publication on motor pathology was the paper, Clinical paths of contemporary biomechanics in 1929 (Bernstein, 1929), which remains relevant today. In this work, he classified kinematics in patient populations and expressed the degrees of freedom problem as influencing disordered motor control. Bernstein used examples of disorders of movement due to various neuropathologies to illustrate his concept of the hierarchical model of the neural control of movement, which he described in five levels from the muscle level to the level of the symbolic representation of motor behavior. It is thus fitting that the theme of the latest book in the Progress in Motor Control series is the translation of motor control principles to clinical populations.

This book is intended for basic neuroscientists and clinical neuroscientists interested in gaining a better understanding of fundamental concepts of motor control and how these concepts can be translated into meaningful rehabilitation interventions to improve sensorimotor outcomes in people with motor disorders, from neuropathology to musculoskeletal impairments. The topics are covered by international experts who participated in the Progress in Motor Control XIX conference held in Rome, Italy from September 28–30, 2023.

Contents

The book is composed of 12 chapters from leaders in the fields of motor control, motor learning, and rehabilitation. Four chapters address themes related to perception-action coupling.

Chapter 1 (Anatol G. Feldman) brings us up to date on Feldman’s seminal theory of motor control, the equilibrium-point hypothesis. The EP theory of motor control has been a driving force in motor control neuroscience since the mid-1960s. Importantly, this chapter provides a balanced view of the strengths and weaknesses of competing theories of motor control as we understand them today.

Chapter 2 (Mark L. Latash) reviews the fundamental idea expressed by Bernstein in his multilevel scheme for the construction of movement—that of the formation of synergies to group elemental variables in order to reduce redundancy and to ensure stability of performance variables. The chapter compares and contrasts two commonly used methodologies to characterize synergies—matrix factorization and analysis of covariation. The measurement approaches describe movements in healthy subjects and in people with neurological disorders, providing evidence for the concept of movement control based on neural variables that define spatial referent coordinates as described in the equilibrium-point approach to motor control, now advanced to the theory of referent control of action and perception.

Continuing with the same concepts, Chapter 3 (Sergio T. Fonseca et al.) discusses how action-perception systems self-organize to adapt actions to organismic, environmental, and task constraints in multiple ways. In particular, the chapter describes how nonlinear methodologies can be used such as entropy, as a manifestation of optimality in motor behavior, to capture how individuals respond to perturbations to identify injury and/or recovery.

Chapter 4 (Gregor Schöner et al.) approaches the understanding of motor control from an integrated theoretical framework elaborated in a neural theory of goal-directed reaching movements. The model draws on dynamical concepts leading to the production of stable movement at different levels of the central nervous system.

Chapter 5 (William H. Warren) reviews the concept of how perception-action coupling guides actions, originally described by James Gibson in the first half of the past century. Instead of considering visual/perceptual information separate from movement production, interactions with the environment both generate and are guided by perceptual input, and task-specific actions emerge from the dynamics of this interaction. In the chapter, the concept of perception-action coupling in one individual is extended to the actions of multiple individuals in crowd situations.

Chapter 6 (Richard E.A. van Emmerik et al.) uses locomotion as a model of a cyclic behavior to describe perception-action coupling during a dynamical action. The illustrations focus on head stability as a key feature of gait stability and adaptability. The chapter describes how healthy individuals integrate visual perception, rhythmic locomotor coordination, and shock attenuation patterns in conditions during which foot-ground collisions and visual task demands are altered based on different biomechanical and visual task demands.

Chapter 7 (Cecilia De Vicariis et al.) The concept of perception-action coupling is extended to how actions are coordinated between human-human dyads, when two people influence each other’s behavior through sensorimotor exchanges within continuous action spaces. The chapter discusses the underlying mechanisms of such coordinative action based on experimental and modeling studies.

Chapters 8 through 12 turn the focus to the translation of motor control concepts to rehabilitation interventions.

Chapter 8 (Ivan Vujaklija) discusses bionic interfaces with a focus on prosthetic limbs for upper limb impairments that are controlled by the brain, peripheral nerves, or muscle activity. The chapter reviews the challenges involved in using and adapting these interfaces, including the need to base bionic control on mechanisms of action-perception coupling in the nervous system.

Chapter 9 (Raoul M. Bongers) focuses on the need to integrate motor control theory in relation to motor learning theory based on dynamical systems to enhance sensorimotor recovery in individuals with neurological disorders. The concept of synergies (Chapter 2) is revisited from the perspective of dynamical systems. The chapter also discusses ways to translate knowledge on how new synergies are learned to the rehabilitation of individuals using a hand prosthesis.

Chapter 10 (Maurizio Petrarca) returns to the consideration of locomotor adaptability (Chapter 5) from a dynamical perspective in healthy individuals and in people with neurological disorders. How dynamical training may improve locomotion and locomotor adaptability in people with neurological disorders is discussed.

In Chapter 11 (Mindy F. Levin), the principles of motor control and learning in people with neurological disorders are discussed from the perspective of the equilibrium-point model of motor control. In line with the referent control theory of Feldman (Chapter 1) and Latash (Chapter 2), it is proposed that an impaired ability of the central nervous system to regulate stretch reflex thresholds underlies sensorimotor deficits and decreased motor adaptability in people with stroke. Translation of these concepts to rehabilitation interventions is discussed.

Chapter 12 (Carolee J. Winstein) also takes a dynamical approach to motor learning by addressing how movement emerges from a complex interaction of the individual, the task, and the environment (Chapters 5 and 9). An emphasis is placed on the brain’s inherent neuroplasticity and the confluence of physical, cognitive, and psychosocial factors and their influence on functional recovery after brain injury.

Acknowledgments

We extend our deepest appreciation and gratitude to all the individuals who played a part in the development of this book. Your unwavering support, expertise, and dedication have been instrumental in shaping this work into its final form. As this book is an outcome of the Progress in Motor Control XIV Conference, we are immensely grateful to the International Society of Motor Control (ISMC) and the Bambino Gesù Children's Hospital (Rome, Italy) for their invaluable support in organizing the Conference. We also extend our special thanks to the other members of the Organizing Committee, namely Mark L. Latash and Monica Perez, for their essential contributions to our endeavor.

References

Bernstein N.A. Klinicheskie Puti Sovremennoi Biomekhaniki (Clinical paths of contemporary biomechanics). [in Russian] Kazan: Trudy Instituta usovershenstvovaniya vrachei; 1929.249–270.

Bernstein N.A. The co-ordination and regulation of movements. 1st ed. Oxford: Pergamon Press; 1967.

Latash, 1998 Latash M.L., ed. Progress in motor control, Vol. 1: Bernstein’s traditions in movement studies. Champaign, IL: Human Kinetics; 1998.

Section A

Action perception coupling

Chapter 1 The equilibrium-point hypothesis: A major framework for the understanding of action and perception

Anatol G. Feldmana,b    a Department of Neurosciences, University of Montreal, Montreal, QC, Canada

b Feil/Oberfeld Research Centre, Jewish Rehabilitation Hospital, Center for Interdisciplinary Research in Rehabilitation (CRIR), Montreal, QC, Canada

Abstract

As in other areas of science, research in behavioral neuroscience is motivated by the desire to consider experimental facts and bring them into a logical system, a theory. In the field of motor control, we have a controversial situation resulting from apparently competing theories of motor control, some of which cannot be considered as physiologically feasible. Moreover, there are persistent signs of regressive, rather than progressive, tendencies in the development of the understanding of basic principles underlying motor control. This can be seen from the fact that despite its origin more than half a century ago, the equilibrium-point (EP) hypothesis remains poorly understood, resulting in numerous, unfounded claims that it should be rejected, despite systematic explanations that all such claims are misleading. In the present review, we will try to rectify this situation by comparing the explanatory and predictive power of different theories and approaches to motor control, starting from those motivated by the idea of Sir Charles Sherrington that the motoneuron (MN) is the final common path for all control processes involved in motor actions. We will also consider the presently dominant computational internal model theory (CIMT) and two versions of the EP hypothesis. The physiological validity of these theories and approaches will be compared by considering how different actions including locomotion are controlled. It is concluded that the λ version of the EP hypothesis, now advanced to the referent control theory, was and remains a valuable framework for guiding research on action and perception.

Keywords

Motor control; Referent control theory; Locomotion; Parametric control; Posture-movement problem; Vision

Introduction

Parametric, referent control of motor actions

A set of motor control theories were formulated under the influence of the ideas of Sir Charles Sherrington. Most seminal of them was the notion that the motoneuron (MN) is the final common path, i.e., the site of convergence of all the central and peripheral pathways involved in motor behavior. Based on this metaphor, it is usually suggested that the control of movement is defined by the output MN activity adjusted by sensory feedback resulting from the interaction of the body with the environment. This activity is also referred to as the motor commands to muscles or EMG patterns. In the biomechanical analyses of actions, it is usually assumed that knowledge of EMG patterns describing the motor outcome is basically sufficient for the understanding of how motor actions are controlled. Originated from robotics (Hollerbach, 1982), the computational internal model theory (CIMT) also postulates that the neural control of motor actions is eventually reduced to the specification (preprogramming) of output signals of MNs or motor commands to muscles. A similar idea was proposed long before Sherrington by researchers who considered how eye movements and visual perception are related (Helmholtz, 1866; Mach, 1897). They suggested that action and perception are interrelated and that a copy of motor commands called, efference copy (EC) by Von Holst (1954) or corollary discharge by Sperry (1950), is transmitted to sensory brain areas to provide visual constancy—the sense that the world is not moving despite the motion of its retinal image during eye movements. Later, it was suggested that EC also plays an essential role in active sensing of limb position (Gandevia et al., 2006). Bernstein (1967) also shared the notion that motor control is reduced to the specification of motor commands adjusted by sensory feedback.

However, this approach seems to disregard the existence of a neurological process that can be independent of, and accomplished prior to, motoneuronal activation, i.e., in a feedforward manner. Specifically, from a physical point of view, many motor actions can be considered as different forms of maintaining or changing the balance (equilibrium) and stability of the neuromuscular system. In particular, according to recent studies (Feldman et al., 2021; Shoja et al., 2023; Zhang et al., 2018), human locomotion can be defined as primarily resulting from transferring stable body balance to a targeted place in the environment, whereas rhythmic activation of multiple body muscles emerges secondarily, due to a central pattern generator (CPG).

There is an important physical principle of how equilibrium states can be set and changed in dynamical systems. In each equilibrium state, muscle and external forces and torques are balanced, but the choice of a specific body posture at which balance can be achieved is defined not by EMG patterns describing the motor outcome, but independently of them, by parameters of physical and physiological laws. This principle was identified in thermodynamics of dynamical systems (Glansdorff & Prigogine, 1971). The applicability of this principle to human motor control was demonstrated experimentally, more than half a century ago, when parameter λ, the threshold muscle length, and respective joint angle, R, at which proprioceptive reflexes become functional in activating elbow muscles, were identified, and integrated in the equilibrium-point (EP) hypothesis (Asatryan & Feldman, 1965). It was found that voluntary actions result from feedforward changes in λ and R, i.e., in advance of changes in the activity of motoneurons (MNs) and thus independently of variables describing the motor outcome. It has also been shown that in involuntary motor actions such as the unloading reflex (a change in the arm position resulting from the sudden removal of a load held by the subject) λ and R can remain invariant (Ilmane et al., 2013; Sangani et al., 2011). Motor actions usually result from several phases or segments of changes in parameters. The transition from one phase to another can be accomplished depending on appropriate sensory signals. For example, sit-to-stand movement is accomplished by initially flexing and then extending the referent body configuration. However, the referent body extension starts only if the afferent feedback from feet receptors ensures that after the referent flexion, the legs will touch the ground and be ready to accept the body weight during standing (Feldman et al., 2007). In other words, the independence of changes in parameters from afferent feedback is not absolute. The transitions from one pattern to another of changes in parameters can be accomplished depending on afferent feedback in a task-specific way.

A major implication of the EP hypothesis is that, rather than being directly predetermined by the nervous system, motor actions emerge indirectly due to setting parameters of physical and physiological laws influencing body balance and stability. A characteristic feature of parameters in the context of the EP hypothesis is that their values can be set independently of values of variables characterizing the motor outcome. This definition of parameters has been poorly understood. For example, it was suggested that equilibrium states of the neuromuscular system depend not on λ defining the spatial threshold for motoneuronal activation, but by the level of this activation or muscle stiffness related to it (Bizzi et al., 1992). However, due to the EMG-force relationship, muscle activation and stiffness are linked to the muscle force characterizing the motor outcome and therefore cannot be regarded as parameters influencing the equilibrium state of the neuromuscular system. The level of muscle activation can be converted into a parameter, either artificially, after transection of dorsal roots in animals, thus destroying the EMG-force relationship, or in humans, following pathological degeneration of dorsal root afferents (Guillain-Barre Syndrome). It was clarified (Feldman, 1986) that in the original version of the EP hypothesis, it is a feedforward change in the muscle activation threshold, λ, preceding MN activation, that is responsible for shifts in the equilibrium state of the intact neuromuscular system. In what follows, the term EP hypothesis refers to its original, λ version.

Despite several clarifications, the EP hypothesis continued to be misunderstood, resulting in a regressive, rather than progressive tendency in the understanding of basic principles underlying motor control during the last half a century. This misunderstanding gave rise to numerous claims that the EP hypothesis should be rejected, despite systematic explanations that all such claims are misleading. The reader can consult the section False rejections of the EP hypothesis in a recent review by Zhang et al. (2022) to see substantial errors in the interpretation of results of experiments based on which the opponents rejected the EP hypothesis (see also Feldman & Latash, 2005; Latash, 1993). Moreover, in this negative atmosphere surrounding the EP hypothesis, there was a tendency to promote the CIMT without realizing that it has essential problems that question its physiological validity. Thus, although the EP hypothesis was elaborated based on the experimental analysis of human voluntary and involuntary actions (Asatryan & Feldman, 1965), it was ignored in the mainstream of studies of perception and action.

To rectify this situation, it is necessary to systematically clarify the basic notions of the EP hypothesis considering that it has been advanced into the referent control theory (RCT) of action and perception (Feldman, 2015). Thus, we will try to resolve the controversy in the understanding of motor control principles using comparatively simple examples of motor actions before considering control processes underlying locomotion. When using the abbreviation RCT, it should be kept in mind that the EP hypothesis is an integral part of it.

Explanations of simple motor actions in the context of the EP hypothesis

Consider the tasks of the production of isometric torques and the choice of a desirable joint angle in isotonic conditions in terms of the EP hypothesis (Fig. 1). These actions result from shifts in the EP in the interaction between the muscles and an external load. In isometric conditions (Fig. 1A), the muscle torque results from pushing with a hand against a motionless object, e.g., a wall or table that counteracts the arm muscles. On the graph of torque versus joint angle, this counteraction (load) is characterized by a vertical line at the joint angle (Q) at which isometric torque is produced. In isotonic conditions (Fig. 1B), the load is represented with a horizontal line at the level of acting torque, T. The EP is the point of intersection between the muscle torque-angle characteristic of the stretch reflex and the load.

Fig. 1

Fig. 1 Reaching a desired muscle torque T in an isometric condition (A) or a joint angle Q in an isotonic condition (B), in the context of the equilibrium-point (EP) hypothesis. In both cases, the system can monotonically decrease the threshold joint angle, R , (horizontal arrows) until the predetermined torque T or joint angle Q , respectively, is reached. Changes in R result in a gradual shift in the EPs, points of intersection between the force-angle characteristics of the stretch reflex (solid curves) and the external load. In the isometric condition (A), the load is represented with a vertical line at joint angle Q at which the isometric torque is generated. In the isotonic condition (B), the load is represented with a horizontal line at the level of constant muscle torque T . In both cases, the motor goal is reached without computations or an internal model of the muscle-load interaction. Note that the threshold position R of the hand, during isometric torque production (in A) is behind the load line, i.e., it virtually penetrates the motionless object against which the subject pushes to produce the isometric torque. The muscle torque is generated since the solid object prevents the hand from reaching the threshold position. This notion also explains how grip force is produced (see Fig. 2). The notion of virtual threshold position is not only essential in the explanation of grip force production but also of how subjects counteract ground reaction forces during standing, walking, running or jumping (Shoja et al., 2023). Panel C shows that the joint angle Q resulting from EP shifts can be sensed by considering the deflection of P from R delivered by proprioceptive feedback (Q = R + P).

The system can monotonically shift parameter R (horizontal arrow, Fig. 1), the spatial threshold at which MNs begin to be recruited due to the stretch reflex until the required isometric torque (Fig. 1A) or joint angle in isotonic conditions (Fig. 1B), respectively, is reached. This method is purely physiological, not involving computations or internal models. Shifts in R can result from a gradual increase in the facilitation of MNs, for example, by the corticospinal system (Raptis et al., 2010).

Fig. 2A shows that the threshold position of the hand, during isometric force production is inside the object, i.e., it virtually penetrates the object. Isometric muscle force is generated since the solid object prevents the hand from reaching the threshold position. This notion is helpful in the explanation of how grip force is produced. It results from the deflection of the actual hand aperture Q, defined by the size of the object, from the threshold hand aperture, R. One can see that the hand can reach the virtual position when the object is suddenly removed (Fig. 2B). The notion of virtual threshold position is not only essential in the explanation of grip force production but also to how ground reaction forces are produced during standing, walking, running, or jumping (Feldman et al., 2021).

Fig. 2

Fig. 2 Threshold control of grip force. (A) By influencing the activation threshold of motoneurons of hand muscles, the nervous system specifies a referent aperture ( R a ) that defines a virtual distance between the index and the thumb. In the presence of the object, the actual aperture ( Q a ) is constrained by the size of the object held between the fingers, whereas in the referent position, the fingers virtually penetrate the object. Deviated by the object from their thresholds of activation, hand muscles generate activity and grip forces in proportion to the gap between the Q a and R a . Thus, grip forces emerge because the object prevents the fingers from reaching the referent position. Both central modifications in the threshold position ( R a ) or/and changes in the size of the object ( Q a ) influence the grip force. (B) The referent aperture is reached when the object is forcefully pulled away from the fingers (horizontal arrow), a phenomenon similar to the unloading reflex usually demonstrated by sudden unloading of other limb segments. Reproduced from Pilon, J.-F., Feldman, A.G. (2006). Threshold control of motor actions prevents destabilizing effects of proprioceptive delays. Experimental Brain Research 174(2), 229–239, with permission.

Limitations of computational motor control theories

In everyday life, it is essential to not only make movements but also make motionless motor actions, such as generating forces holding a solid object, pushing with hands against a wall, producing ground reaction forces during standing or at certain phases of walking, running, jumping, etc.

Consider how such an action is produced in the context of CIMT—generation of a flexor muscle torque at a certain elbow angle, i.e., in the isometric condition. This can be accomplished in the CIMT by gradually increasing the activity of flexor elbow muscles until the desired muscle torque is established. To make this process physiologically feasible, we need to be sure that in the CIMT framework, the system can deliver input signals forcing MNs to produce the required motor commands to muscles. According to the CIMT, this can be done using an internal model that inverts the input/output functions of MNs. However, MNs are fundamentally nonlinear objects having irreversible input/output functions (e.g., Feldman, 2019). This means that the internal models required in the CIMT do not exist in isometric tasks as well as in other motionless tasks listed above in which generation of muscle torques in isometric conditions is required. Note, however, that in robotics, one can replace MNs with artificial neurons having reversible, e.g., linear functions to generate isometric torques, but here we consider only physiologically feasible theories of motor control.

Similarly, the CIMT is not helpful in the explanation of how the system sets, say, a desired elbow angle in isotonic conditions in which muscle and external torques are balanced at each elbow angle. The choice of a specific joint angle at which arm balance is achieved is defined not by muscle torques or forces or other kinematic and kinetic variables describing the motor outcome, but independently of them by parameters that are absent in the CIMT.

Advancing the EP hypothesis into the referent control theory of action and perception

The question of how multiple muscles are controlled is addressed in the RCT by assuming that all possible body postures or configurations comprise a spatial frame of reference (FR) or system of coordinates in which each posture is represented as a point (Fig. 3) The origin or referent point in this FR is a spatial, threshold body posture at which all muscles of the body can be quiescent but can be activated depending on the deflection of the emergent, actual body configuration Q from the threshold, referent body configuration R. If necessary, the R is modified until the emergent action meets the task demands. The notion of referent posture can be applied to muscles of body segments, e.g., arms or legs with the respective change in the term—the referent arm or leg postures or configurations. The RCT was supported by verifying its prediction that in several motor actions Q and R can transiently match each other bringing the activity of multiple body muscles to a minimum (Mullick et al., 2018). Most often this occurs in movements with reversals in direction, when the R begins to be reversed, whereas the body, by inertia, continues to move without a change in direction for some time. Moving toward each other, Q and R overlap (Q ≈ R), as was observed during head motion with reversals in direction in monkeys (Lestienne et al., 2000). However, transient movement reversals in direction occur in many motor actions—during vertical or forward jumps, sit-to-stand movements, dancing, and hammering. In all such actions, transient matching (Q ≈ R) and minimization of EMG activity of multiple muscles have been observed (Lepelley et al., 2006; Lestienne et al., 2000; Mullick et al., 2018; St-Onge & Feldman, 2004).

Fig. 3

Fig. 3 Referent control theory, an extension of the EP hypothesis. (A) It is assumed that all possible body postures or configurations comprise a spatial frame of reference (FR) or system of coordinates in which each posture is represented with an n -dimensional point ( n is the number of degrees of freedom of the body or its segments involved in the motor task). The origin or referent point in this FR is a spatial, threshold body posture R at which all muscles of the body or segments involved in the motor task can be quiescent. Muscles can be activated depending on the deflection of the emergent, actual configuration Q from the threshold, referent configuration R . If necessary, the R is modified until the emergent action meets the task demands. (B) An explanation of a motor action in the framework of the referent control theory (RCT). To balance on the beam, the athlete specifies a referent body configuration ( gray silhouette). Under the influence of gravity, the body is deflected from R to the actual posture Q (color silhouette) at which a stable equilibrium of the body is reached. The stable equilibrium is set due to muscle activation depending on proprioceptive reflexes and other sensory feedback sensitive to deflection of the body P from Q such that, symbolically Q  =  R  +  P . The resulting body posture can be corrected by modifying the R posture. For an explanation of how the theory has been verified experimentally, see text. Modified from Feldman, A.G. (2011). Space and time in the context of equilibrium-point theory. Wiley Interdisciplinary Reviews. Cognitive Science, 2(3), 287–304, with permission.

The origin or referent point is only one of the attributes of an FR. Another attribute is metrics showing how far a given posture, Q, is deflected from the referent posture R. It is assumed that, physiologically, proprioceptive feedback, P, and other sensory signals are sensitive to the deflection of Q from R such that, symbolically:

Equation    (1)

that has been defined for single joints (Fig. 1C), but now refers to the entire body postures Q and R (Fig. 3). This formula has several implications. First, changes in R underlie intentional motor actions, and since the R is the origin point of the FR, one can say that motor actions result from shifts in the FRs in which motor actions are produced. The above formula also implies that with each intentional motor action, sensory information is reinterpreted and perceived in the new, shifted FR. In contrast, during the unloading reflex, the R usually remains unchanged (Ilmane et al., 2013; Sangani et al., 2011), and the change in the arm position is sensed based on changes in proprioceptive feedback, P, when the FR remains stationary. Similarly, tendon vibration affects component P in the above formula, resulting in an illusion of arm motion (Feldman & Latash, 1982). In some cases, tendon vibration affects the central component, R, in the above formula, resulting in an inversion in kinesthetic illusion (Feldman & Latash, 1982). In most cases, changes in the arm position are sensed based on changes in both components of position sense (see Feldman, 2016) in which the position sense rule (1) is used to explain the phantom limb and phantom limb pain phenomena and why mirror therapy can be helpful in treating these symptoms.

Visual constancy in the context of RCT

The notion of action-perception coupling is an integral part of practically all motor control theories, including the RCT, but the question of how such coupling is achieved is a matter of controversy. Physiologically, neurons that reflect motor commands to muscles or EC have been found in different supraspinal brain areas (Robinson, 1968). Renshaw cells that are responsible for recurrent inhibition of MNs can be considered as reflecting EC at the spinal level. However, Bridgeman and his colleagues showed that visual constancy would be accomplished with substantial errors if it were based on EC (Bridgeman, 2007, 2010; Bridgeman et al., 1975). The role of EC in active sensing of limb position and motion and visual constancy has also been questioned (Feldman, 2016; Zhang et al., 2022). In particular, according to Von Holst (1954), visual constancy results from suppression of afferent signals resulting from eye motion by efferent signals reflecting motor commands to oculomotor muscles or EC. It is unclear how afferent and efferent signals of different dimensions can compensate each other.

In the context of referent control, visual constancy can be considered as a natural consequence of shifts of visual perception from one spatial FR to another during saccades (Zhang et al., 2022). Specifically, R is initially centered on the fovea and then on the final foveal position at which the image of the visual target is anticipated to appear after the saccade. Topologically, rotation of the eyes maintains the relative distances between the retinal images of objects. Therefore, only moving objects will be perceived as such after a saccade. Although the depiction of world retinal images in the shifted spatial FR is likely a major cause of visual constancy, it might also be facilitated by trans-saccadic memory of stationary landmarks, as suggested by Deubel et al. (2010). It is also essential that with the transition from a pre- to post-saccadic FR, the image of the world is presented not only in the fovea but also in the entire retina.

Referent control of motor actions by descending systems

The corticospinal system

It has been shown that, rather than directly influencing motor commands to muscles, descending systems primarily influence parameters λ and R such that, say, the corticospinal system can be involved in production of the motor actions considered above (Raptis et al., 2010). This has been shown by asking subjects to establish different wrist positions at which tonic EMG levels were similar (equalized). Using transcranial magnetic stimulation (TMS) of the motor cortex projecting to MNs of wrist muscles, we tested two alternative hypotheses. According to the traditional view, the corticospinal system specifies motor commands to muscles. Since these commands were equalized, TMS responses in muscles at different wrist positions should also be similar. In contrast, according to the RCT, active changes in the wrist position result from changes in the referent, threshold wrist position, R. Therefore, TMS responses of wrist flexor MNs should be higher during wrist flexion, and responses of wrist extensor MNs should be higher during wrist extension. Testing supported the RCT (Fig. 4) and rejected the traditional view that the motor cortex is involved in the specification of motor commands. This view was accepted after findings that the activity of M1 neurons correlate with the activity of targeted MNs (e.g., Evarts & Tanji, 1974). It should be noted that since correlations do not necessarily imply causality, without additional tests, conclusions based on correlations can be unreliable. It was shown in decerebrated cats (Feldman & Orlovsky, 1972) that descending systems, including M1, primarily set parameter λ, thus indirectly influencing the activity of MNs. Using transcranial magnetic stimulation (TMS) techniques, Raptis et al. (2010) showed in humans that M1 activity can be decorrelated from that of MNs. This demonstrated that M1 is directly involved in setting parameters λ and R (Fig. 4) and only indirectly influences the activity of MNs, contrary to the traditional view and the major postulate of CIMT.

Fig. 4

Fig. 4 Corticospinal influences reset the referent (threshold) position underlying intentional wrist motion without being directly involved in the specification of electromyographic (EMG) patterns. (A) When subjects intentionally changed wrist position from extension (E) to flexion (F) or vice versa, wrist flexors (flexor carpi radialis, FCR; flexor carpi ulnaris, FCU) were tonically active in the flexion (F) position and extensors (extensor carpi radialis, ECR; extensor carpi ulnaris, ECU) in the extension (E) position. (B) By compensating passive muscle torques with a torque motor, the tonic EMG activity of wrist muscles at F and E positions was equalized at near-zero (threshold) levels, showing that the threshold position at which muscles begin to be activated was reset with the changes in wrist position. (C) Although EMG activity was equalized, transcranial magnetic stimulation (TMS) of the primary motor cortex at the E position elicited a small extensor jerk (top curves, left and right) and a flexor jerk at the F position, (top curve, middle). Motor evoked potentials (MEPs) elicited by TMS also changed reciprocally for wrist flexors and extensors with the transition from one wrist position to another. (D) and (E) Group data from 16 subjects, showing that although EMG activity was similar at F and E positions (D), MEPs for wrist flexors and extensors were substantially different at these positions (E), implying that the referent (threshold) position was shifted when the intentional wrist motion was produced. This is consistent with the hypothesis that the corticospinal system accomplishes referent control of motor actions without being directly involved in the specification of EMG patterns. Reproduced with modifications from Raptis, H.A., Burtet, L., Forget, R., Feldman, A.G. (2010). Control of wrist position and muscle relaxation by shifting spatial frames of reference for motoneuronal recruitment: Possible involvement of corticospinal pathways. The Journal of Physiology 588: 1551–1570, with permission.

Thus, complex movements in robots designed in artificial intelligence (AI) frameworks cannot be considered as models of biological motor actions. Moreover, it is now fashionable to use word computation in reference to what the real