Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The influence of sensorimotor loop delays in maintaining upright stance McKendry, Geoffrey James 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_september_mckendry_geoffrey.pdf [ 7.94MB ]
Metadata
JSON: 24-1.0368918.json
JSON-LD: 24-1.0368918-ld.json
RDF/XML (Pretty): 24-1.0368918-rdf.xml
RDF/JSON: 24-1.0368918-rdf.json
Turtle: 24-1.0368918-turtle.txt
N-Triples: 24-1.0368918-rdf-ntriples.txt
Original Record: 24-1.0368918-source.json
Full Text
24-1.0368918-fulltext.txt
Citation
24-1.0368918.ris

Full Text

 The Influence of Sensorimotor Loop Delays in Maintaining Upright Stance     by Geoffrey James McKendry BSc., The University of British Columbia, 2014  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in The Faculty of Graduate and Postdoctoral Studies (School of Kinesiology)  THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)  July 2018  © Geoffrey James McKendry, 2018   ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:   submitted by Geoffrey James McKendry in partial fulfillment of the requirements for the degree of Master of Science in Kinesiology  Examining Committee: Dr. Jean-Sébastien Blouin Supervisor  Dr. Mark Carpenter Supervisory Committee Member  Dr. Romeo Chua  Supervisory Committee Member Dr. Eli Puterman Chair            iii Abstract Sensorimotor delays are inherent to the control of standing balance. Increases in sensorimotor delays observed during healthy aging and in people with Multiple Sclerosis (MS) have been theorized to reduce stability. The aim was to determine if balance stability declines with increasing artificial delays in balance control. Furthermore, I sought to determine whether there is an association between estimates of sensorimotor loop delays and the artificial delays at which people become unstable.    Thirty healthy participants (19 to 75 years) and three participants with MS were recruited. A rotating platform elicited balance responses to pitch rotations. Surface electromyography recorded muscle activity from soleus (Sol), medial gastrocnemius (mGas), and tibialis anterior (TA) to determine onset latencies of balance responses, providing estimates of sensorimotor loop delays in balance control. A robotic balance simulator introduced artificial delays (50-400 ms) between the participants’ motor commands and resulting whole-body movements. Participants balanced for up to 20 seconds or until a ‘virtual fall’ occurred (exceeding 3° posterior or 6° anterior). Logistic regression analysis using the number of ‘virtual falls’ at each delay was performed to obtain a threshold of stability (5% probability of falling).  Maintaining balance became more difficult as delays increased: the number of ‘virtual falls’ and standard deviation of angular backboard displacement increased, while the time to fall decreased as delays increased. The mean threshold of stability was 98 ± 38 ms. In response to the tilt perturbations, short and medium latency responses were observed in Sol (50.8 ± 3.2 ms) and TA (98 ± 10 ms), respectively. Long latency responses in TA, Sol and mGas were 147 ± 12 ms, 166 ± 22 ms, and 178 ± 27ms respectively. There were no correlations between thresholds of balance stability on the robot or the evoked-balance responses and age. A weak association   iv between long latency responses in TA and thresholds of stability was observed. My results support the postulation that balance control becomes unstable as delays in balance control increase. The weak correlation between estimates of sensorimotor delays and thresholds of stability on the robot, however, suggests these measures target different pathways involved in standing balance.        v Lay Summary In order to maintain balance our central nervous system must receive information about our position through various sensory systems and respond to signals that indicate we are off balance. These signals do not happen instantaneously, as it takes time for neurons to conduct signals. It has been theorized that balance stability decreases as neural response times increase. In this study my aim was to determine whether balance stability declines when someone experiences imposed delays in their own balance control. This can be done using a robotic balance simulator. Furthermore, I sought to determine whether estimates of a person’s overall neural delays are associated with the imposed balance delay in which people become unstable on the robot. This study shows that balance stability declines as imposed delays increase. However, a weak connection was found between peoples’ own neural delay and the imposed delay in which they become unstable.    vi Preface All work henceforth was conducted in the Sensorimotor Physiology Laboratory as well as the Neural Control of Posture Movement Laboratory at the University of British Columbia’s Point Grey Campus. The experimental protocols used in this thesis were reviewed by the University of British Columbia’s Clinical Research Ethics Board (certificate# H17-00600).   I was the lead investigator on this project for all major areas of concept development, data collection and analysis, and document composition. Dr. Jean-Sébastien Blouin, Dr. Mark Carpenter, and Dr. Romeo Chua were supervisory authors of this project and were involved in concept formation and thesis revisions. Dr. Rajiv Reebye also consulted on this research project providing clinical expertise and recruitment assistance in participants with Multiple Sclerosis. At the time of thesis submission the experiment contained in this thesis have not been submitted for publication.      vii Table of Contents  Abstract ......................................................................................................................................... iii	  Lay Summary .................................................................................................................................v	  Preface ........................................................................................................................................... vi	  Table of Contents ........................................................................................................................ vii	  List of Tables ................................................................................................................................ ix	  List of Figures ................................................................................................................................. x	  List of Abbreviations ................................................................................................................... xi	  Acknowledgements ..................................................................................................................... xii	  Chapter 1: Introduction to Standing Balance Control ...............................................................1	  1.1	   The Control of Human Stance ........................................................................................... 2	  1.2	   Perturbation-Evoked Balance Responses: Estimating Sensorimotor Loop Delays ........... 5	  1.3	   Supraspinal Influence of Perturbed Stance: Differences in Sensorimotor Loop Delays ... 6	  1.4	   Artificial and Estimated Sensorimotor Delays in Balance .............................................. 11	  1.4.1	   Artificial Delays Using a Robotic Balance Simulator .............................................. 11	  1.4.2	   Age-Related Changes in Balance Response Latencies ............................................. 13	  1.4.3	   Multiple Sclerosis: An Alternative Model for Slowed Sensorimotor Loop Delays . 14	  1.5	   Aim and Hypotheses ........................................................................................................ 15	  Chapter 2: Experimental Design, Protocol, and Results ..........................................................17	  2.1	   General Experimental Setup ............................................................................................ 17	  2.1.1	   Participants ................................................................................................................ 17	  2.1.2	   Balance Simulator ..................................................................................................... 17	  2.1.3	   Adjustable Time Delays ............................................................................................ 19	  2.1.4	   Rotating Platform ...................................................................................................... 20	  2.1.5	   Electromyography ..................................................................................................... 22	  2.2	   Experimental Procedures, Analysis, and Results ............................................................. 22	  2.2.1	   Robotic Balance Simulator: Investigating Thresholds of Stability ........................... 22	  2.2.1.1	   Methods.............................................................................................................. 22	  2.2.1.1.1	   Balance Stability ......................................................................................... 22	  2.2.1.1.2	   Analysis: Variables of Stability .................................................................. 24	  2.2.1.1.3	   Statistical Analysis ...................................................................................... 27	  2.2.1.2	   Results ................................................................................................................ 28	  2.2.2	   Rotating Platform: Determining Postural Response Latencies ................................. 32	  2.2.2.1	   Methods.............................................................................................................. 32	  2.2.2.1.1	   Balance Response Latencies ....................................................................... 32	  2.2.2.1.2	   Analysis: Quantifying Balance Response Latencies ................................... 33	  2.2.2.1.3	   Statistical Analysis ...................................................................................... 34	  2.2.2.2	   Results ................................................................................................................ 34	    viii 2.2.3	   Association Between Thresholds of Stability and Balance Response Latencies ...... 42	  2.2.3.1	   Statistical Analysis ............................................................................................. 42	  2.2.3.2	   Results ................................................................................................................ 42	  Chapter 3: Discussion ..................................................................................................................44	  3.1	   Robotic Balance Simulator .............................................................................................. 44	  3.1.1	   Sensorimotor Delays and Loss of Stability ............................................................... 44	  3.1.2	   Adaptation Effect ...................................................................................................... 45	  3.2	   Perturbation-Evoked Balance Responses ........................................................................ 48	  3.2.1	   EMG Onset Latency Comparisons with Current Literature ..................................... 48	  3.2.2	   Negative Results of Correlation Between Age and Perturbation-Evoked Balance Responses .............................................................................................................................. 49	  3.3	   Correlation Between Balance Stability Thresholds and Balance Response Latencies .... 52	  3.3.1	   Interpretation of Results ............................................................................................ 52	  3.4	   Observations in Participants with Multiple Sclerosis ...................................................... 54	  3.5	   Considerations and future directions ............................................................................... 56	  Chapter 4: Conclusion .................................................................................................................58	  Work Cited ...................................................................................................................................59	  Appendix - Individual EMG onset latencies ..............................................................................63	     ix List of Tables Table 2.1 Average EMG onset latencies for each age group…………………………..……..37 Table A.1 Onset latency data for all participants…………………….………………..….…..61  	     x List of Figures Figure 1.1 Short, medium, and long latency balance responses in humans……………………..10 Figure 2.1 Servo-controlled balance simulation device…………………………………………19 Figure 2.2 Rotating platform setup……………………………………………...………………21 Figure 2.3 Example of angular displacement traces for 9 trials…………………………….…...26 Figure 2.4 Logistic regression curves and thresholds of stability………………………...……..27 Figure 2.5 Linear regression of age vs. thresholds of stability………………………………….30 Figure 2.6 Dependent variables for robotic simulations………………………….……………..31 Figure 2.7 Regression lines for age vs. onset latencies………………………………….……....36 Figure 2.8 Example of average onset latencies for on participant………………………………39 Figure 2.9 Example of bilateral difference in one individual with Multiple Sclerosis…….……40 Figure 2.10 Relationship between age and long latency responses in tibialis anterior……..…..41 Figure 2.11 Long latency responses vs. thresholds of stability……………………………..…..43 Figure 3.1 Average number of ‘virtual falls’ per round…………………………………………46 Figure 3.2 The average number of falls per round per delay……………………………………47 Figure 3.3 Comparison of perturbation-evoked EMG responses with Nardone et al. (1995).….51   xi List of Abbreviations CI: confidence interval df: degrees of freedom   EMG: electromyography  LLR: long latency response mGas: medial gastrocnemius MLR: medium latency response MS: Multiple Sclerosis SD: standard deviation  SLR: short latency response  Sol: soleus TA: tibialis anterior   xii Acknowledgements I would like to offer my gratitude to the faculty members and students who have been instrumental in my learning over the last two years. Dr. Blouin, I cannot thank you enough for the support you have provided me. Your door was always open and you always took the time to make sure I was on the right track. I appreciate your expertise and your incredible ability to provide guidance on a wide range of projects that I was involved in over the course of this degree. Dr. Carpenter, I would like to thank you for taking the time to guide me through this project. Having little experience in analyzing EMG or operating the tilt platform, I sincerely appreciated your patience with me throughout this process. Dr. Chua, I thank you for your guidance and input throughout this degree. The discussions generated in your 568 class were instrumental in formulating this experiment. Furthermore, I would like to thank Dr. Reebye for believing in my project and providing clinical expertise regarding patients with Multiple Sclerosis.   I would like to thank John Luu and Taylor Cleworth for their assistance in setting up part of this experiment in the Neural Control of Posture and Movement laboratory. Like Dr. Carpenter, you were both patient with my lack of experience I appreciated your continuous support when needed. I would also like to express an enormous amount of gratitude to all the members of the Sensorimotor Physiology Laboratory. Everyone in the lab was always keen to help in anyway possible and it truly felt like a team environment. I also will miss our new traditions: post lab meeting games, group lunches, and sports competitions. You have all made coming to work a joy!       To my family, I would like to thank you for your unwavering support throughout this degree. From start to finish you helped in everyway possible, whether it was meals, access to a   xiii vehicle or morning pick ups (thanks for all the rides dad!). You also were an enormous help in completing this thesis through participating in the experiment and helping with recruitment. Benny, I enjoyed our meals and games nights while you were here and I cannot wait to catch up on some bro time this summer in Finland. I’m always proud of you! Finally, I would like to give my utmost appreciation to Krisula. Thank you for being so patient with me in my academic pursuits and always supporting me in my goals.           1 Chapter 1: Introduction to Standing Balance Control Standing upright is an apparently easy task for most humans yet its sensorimotor control is complex. Through neural circuits that include monosynaptic and polysynaptic spinal reflexes, as well as subcortical and trans-cortical sensorimotor loops, the human balance controller constantly adjusts muscle activity to maintain upright stance. Inherent to the sensorimotor control of standing balance are small delays due to the time it takes neurons to integrate and conduct signals. Thus, the central nervous system must coordinate balance with visual, vestibular, and somatosensory information reaching control centres at different times. To investigate how the controller processes and responds to sensory information, researchers commonly use imposed perturbations of the visual, vestibular, and/or somatosensory environment to measure evoked balance responses. Data from these methods can be used to formulate computational models of how standing balance is controlled. Testing these models, however, requires manipulating balance dynamics under active rather than imposed conditions. To do so a robotic device was developed to simulate the mechanics of a standing person while allowing subjects to experience novel balance conditions. One such condition is a simulation of increased sensorimotor loop delays in balance.  The present research focuses on how temporal changes in sensorimotor loop delays affect standing balance. The experiment involved using a robotic balance simulator that can create a delay between a participant’s motor commands and resulting whole-body movements. Using a logistic regression approach, I determined the probability of a participant falling within 20 seconds of balance at a given delay (50 – 400 ms).  The aim of this experiment was to determine whether balance stability declines when increasing sensorimotor delays in balance control are introduced. This experiment also sought to determine whether there is an association between   2 estimates of sensorimotor loop delays and the artificial delays in which people become unstable on the robotic balance simulator. To estimate sensorimotor loop delays in balance, this experiment used a rotary platform to determine onset latencies of perturbation-evoked balance responses. Therefore, this thesis seeks to characterize changes in balance stability when delays in sensorimotor balance control increase and how those changes in stability are related to perturbation-evoked balance responses. In Chapter 1 I provide an introduction to the basic principles and mechanics of human stance. I then introduce the concept of internal models in balance control along with a computational model suggesting that balance stability declines with increasing delays. I then describe characteristic muscle responses to imposed rotary perturbations and where these responses are theorized to be mediated in the central nervous system. Finally, I describe the robotic balance simulator’s ability to create artificial delays in balance control and real examples of increased delays due to age or disease, in this case Multiple Sclerosis. In Chapter 2 I describe the methods, analysis, and results for each device used in this experiment followed by the analysis and results from the comparison between balance responses on the rotary platform and balance stability on the robot. In Chapter 3 I discuss the results and how they relate to current theories regarding the sensorimotor control of standing balance.  The thesis then concludes with a summary of the results and suggested future directions. 1.1 The Control of Human Stance Standing balance requires controlling the body’s centre of mass within a base of support defined by the lateral, anterior, and posterior edges of the feet. Despite the apparent ease of this task, standing balance is mechanically unstable, as the whole body load must be balanced in the presence of gravitational forces and disturbances from breathing, heartbeat, and intrinsic noise in   3 our sensory and motor systems. In this research project, the ability of the central nervous system to maintain balance in the anteroposterior plane is of particular importance. For this reason, I will focus this section on describing balance control in the sagittal plane. Anteroposterior balance is commonly represented as a single-link inverted pendulum whereby the whole body sways about the ankle joints. Typically the body centre of mass rests slightly in front of the ankles during standing (Smith, 1957; Cotton, 1931). Thus, gravity generates a torque that would cause the body to topple forward. In order to prevent this from occurring, the triceps surae muscles must generate a plantar-flexion torque that is, at least, equal and opposite to the torque imposed by gravity. If the plantar-flexor torque is equal and opposite, the body will remain in a quasi-static posture. Otherwise the body’s centre of mass will accelerate in the direction of the net torque.  How the central nervous system controls balance-correcting torques is still being explored. Early research postulated that individuals can maintain upright stance through inherent stiffness of the ankle joint provided by the Achilles tendon, and reflexive muscle control from passive stretching of triceps surae (Fitzpatrick, Taylor, & McCloskey, 1992; Fitzpatrick, Rogers, & McColskey, 1994; Winter et al., 1998). This would suggest that anteroposterior balance could be maintained through mostly passive processes. However, more recent research has demonstrated that the torque generated from passive stretch of the ankle tissues during standing balance cannot adequately counteract the torques created by gravity acting on the centre of mass (Loram & Lakie, 2002; Morasso & Sanguineti, 2002; Morasso & Schieppati, 1999). Thus, active control of balance is required to maintain upright stance, supplementing the restoring torque associated with stretching the passive tissues.    4 To actively maintain upright stance the balance controller must integrate sensory information to generate corrective actions. Through visual, vestibular and somatosensory inputs, the balance controller filters, processes and integrates sensory cues generating compensatory motor commands to correct for errors in balance. Inherent to sensorimotor control is a delay between sensory input, cortical processing, and subsequent motor output. This poses a constraint to the controller’s ability to respond rapidly to imbalance. To overcome the inherent loop delays in sensorimotor control, it is widely theorized that general motor control relies on internal models to make online corrections and adapt to changing environments (Kawato & Wolpert, 1998; Schmidt, 1975; Scott, 2004; Shadmehr, Smith, & Krakauer, 2010). Through these models, the controller uses internal representations to compare sensory feedback with the model’s sensory expectation of self-generated movements. Differences between the expected and actual sensory feedback produces an error signal resulting in the generation of compensatory motor commands.  By using data on from studies investigating balance dynamics, researchers have also created computational models of standing balance (Kuo, 1995; Kuo, 2005; Morasso, Spada, & Capra, 1999; Peterka, 2002; van der Kooij et al., 1999). Through the use of such computational models, it has been shown that the ability of the balance controller to predict sensory feedback is important to overcome sensorimotor loop delays. For example, van der Kooij & Peterka (2011) used a computational model operating on proportional derivative principles to demonstrate that stabilization of balance increasingly declines at delays above 100 ms, with complete loss of stabilization above 340 ms. Additionally, the removal of a predictive mechanism in another computational model resulted in the controller not being able to stabilise balance (van der Kooij   5 et al., 1999). Thus, current models suggest that internal representations of balance control are necessary to overcome the sensorimotor loop delays in balance. 1.2 Perturbation-Evoked Balance Responses: Estimating Sensorimotor Loop Delays  Sensorimotor loop delays in balance control can be determined using mechanical or sensory perturbations. For example, afferent conduction velocity can be determined using somatosensory evoked potentials by measuring the time it takes sensory signals to travel between the initial stimulus and the cortex. Alternatively, the same principle can be applied to measuring efferent conduction velocity using transcranial magnetic stimulation. To estimate the total loop delays in balance responses, electromyography can be used to determine the time it takes for muscles to respond to a sensory and/or mechanical perturbation to balance. In this thesis estimates of sensorimotor loop delays in balance control are compared with stability on a robotic platform that operates in the sagittal plane. To obtain an estimate of sensorimotor delays related to balance in the sagittal plane it was postulated that pure pitch rotary perturbations would be closely related. I will therefore discuss the evoked responses to pure pitch rotary perturbations during stance. By suddenly rotating a support surface upwards or downwards, distinct activity in muscles of the lower limb is observed following perturbation onset for each direction (Allum & Büdingen 1979; Diener et al. 1983). For example, an upward rotation of the platform, ‘toes-up’, stretches the soleus and gastrocnemius muscles eliciting short (30 - 60 ms) and medium latency (70 - 120 ms) responses, while long latency responses (100 – 200 ms) are observed in tibialis anterior. Whereas a downward rotation of the platform, ‘toes-down’, is characteristically different in that only medium latency responses are generally observed in the stretched tibialis anterior, while long latency responses are observed in soleus and gastrocnemius muscles.   6 Irrespective of direction, the short and medium latency responses are considered destabilizing, as the reflexive contractions result in acceleration of the centre of mass in the same direction as the perturbation. Conversely, the long latency responses, which were observed in muscles that are antagonistic to the stretched musculature, contribute to stabilizing the body in response to a perturbation. For this reason it is referred to as a ‘balance-correcting response’ and will also be referred to in this proposal as the long latency response.  In addition to changes in direction, modulation of the balance responses occur through changes in the amplitude or velocity of perturbations as well as initial postural set (Nardone, Corrà, & Scieppati, 1990; Diener et al., 1983). For example, the amplitude and velocity of the perturbation also scale the amplitude of the balance response such that larger and faster perturbations evoke larger muscle responses (Nardone, Corrà, & Scieppati, 1990). Initial postural set can also influence postural responses, where backwards lean prior to a toes-up rotations elicits greater and earlier tibialis anterior responses compared to when standing normal (Diener et al., 1983). Despite the influence of perturbation type and the state of the individual on the characteristics of each response, each balanace response described above has been replicated under numerous conditions and when perturbation parameters are constant these responses demonstrate consistent phases and activation patterns (Allum & Büdingen, 1979; Campbell, Dakin, & Carpenter, 2009, Carpenter, Allum, & Honegger, 1999;  Diener et al., 1984; Nardone, Corrà, & Schieppati., 1990; Nashner 1976).  1.3 Supraspinal Influence of Perturbed Stance: Differences in Sensorimotor Loop Delays  Standing balance, like breathing, is a largely subconscious task. However, when balance is perturbed the cortex can influence balance responses. Historically, subcortical regions such as the brainstem and spinal networks were thought to control automatic balance responses to   7 perturbations without much cortical influence. This was due to observations that animals with transections of the midbrain retained reflexes that maintain upright stance (Magnus, 1926 & Sherrington, 1910). Furthermore, scientists believed that the temporal window of evoked responses to perturbations was too short to be under volitional control (Diener et al., 1984). More recent research has demonstrated that the cerebral cortex may have more of an influence on automatic balance responses than previously thought, particularly at later latencies (See review by Jacob & Horak 2007). Here I will discuss how differences in automatic balance responses can be used to make assumptions about where in the central nervous system sensorimotor loops involved with perturbation-evoked balance responses are associated. I will also stipulate that sensorimotor loops involved in perturbation-evoked balance responses may not necessarily reflect processes occurring during normal active stance.   Pathological conditions that affect the central nervous system allow researchers to gain insight into cortical and subcortical contributions to standing balance. For example, (Diener et al., 1985) compared balance responses to toes-up perturbations on a rotating platform between patients with spinal lesions, patients with lesions of the internal capsule or sensorimotor cortex, patients with lesions in the frontal or occipital lobes, and healthy individuals. They observed that participants with spinal lesions had abolished or considerably reduced medium latency responses with significantly delayed long latency responses compared to controls (Diener et al., 1985). The absence of medium latency responses and delays in long latency responses suggests that medium and long latency responses are mediated in regions rostral to the spinal cord. In participants with cortical lesions, a slight but significant increase in short and medium latency responses along with a greater delay in the long latency responses was observed (Diener et al., 1985). Thus   8 indicates that the cortex has an influential role on perturbation-evoked balance responses, particularly in the late latency response.   The role of the cortex in perturbation-evoked long latency balance responses has also been investigated through the use of transcranial magnetic stimulation and peripheral nerve stimulation. Taube et al. (2006) timed transcranial magnetic stimulation or peripheral stimulation so that either motor evoked potentials or H reflexes coincided with the onset of short, medium, or long latency responses during a posterior surface-translation task while measuring muscle activity in soleus. Results indicated that only long latency balance responses were enhanced by transcranial magnetic stimulation when accounting for spinal motoneuron excitability, giving further evidence that cortical centres are able to modulate late latency responses to perturbations in stance.      Cortical influence on perturbation-evoked balance responses can also be investigated through tasks that alter cognitive set prior to perturbations. For example, Carpenter et al., (2004) induced postural anxiety by exposing individuals to balance perturbations at height. Postural threat condition increased amplitudes of balance-correcting responses in all leg, trunk, and arm muscles, demonstrating that higher brain centers associated with anxiety and emotional behaviour can influence late latency perturbation-evoked balance responses (Carpenter et al., 2004). Additional research has also demonstrated that the cerebral cortex can modulate postural responses based on intention and knowledge of the perturbation, provide online monitoring of balance status, and modulate long latency and later-phase responses to changes in support surfaces (see review by Bolton (2015)).  So far I have given examples of how researchers use imposed perturbations to evoke balance responses to explain how cortical and subcortical regions respond to imbalance. These   9 examples, however, cannot make assumptions about the cortical and subcortical influences of unperturbed stance. To investigate central contributions to balance without imposing a perturbation during stance, Luu (2010) asked participants to generate voluntary torques that matched the torques generated during quiet stance. He found that during the voluntary task, participants would generate one third of the torque compared to the torque generated during quiet stance accounting for the influence of passive torques. This lead to the conclusion that quiet stance is primarily controlled sub-cortically.  Recent advances in robotics have also allowed researchers to gain more insight into where in the central nervous system active balance is being controlled. Luu et al., (2012) used a robotic device (described in section 1.4.1) that mimicked the mechanics of standing balance to investigate the relationship between muscle responses and vestibular signals while standing. In this experiment, subjects were asked to maintain quiet standing balance on the robot while being perturbed with electrical vestibular stimulation. Intermittently, motion would become computer-controlled, whereby the subject would receive motions from earlier in the trials. During this time, coherence between vestibular stimuli and soleus EMG decreased significantly 175 ms following onset of computer-controlled motion. Perception of the change, however, did not occur until 2247 ms following onset of computer-controlled motion. This demonstrated that vestibular-motor systems are sensitive to congruent sensory and motor signals during standing in the absence of conscious perception. Thus, subcortical regions may play a more vital role in active stance compared to the cortex.  Although the exact role of subcortical and cortical structures in controlling balance is not yet understood, research suggests that trans-cortical loops are more involved in later responses. Thus, it may be that subcortical circuits initiate early balance responses to perturbations and also   10 control the active balance state, while cortical circuits modify long latency balance responses to imbalance. For the purposes of this thesis, I will assume that evoked responses to perturbations involve cortical control of long latency responses as illustrated in Figure 1. The long latency responses will also be used as an estimate for participants’ own sensorimotor delay in responding to imbalance when artificial delays are imposed on the robotic balance simulator.   Figure 1.1 - Short, medium, and long latency balance responses in humans. In this diagram, short latency balance responses occur through reflexive pathways at the spinal cord. The medium latency balance responses occur through supraspinal sensorimotor loops that synapse in the midbrain/brainstem. Long latency balance responses involve transcortical sensorimotor loops that modulate responses to imbalance. Adapted from Jacobs & Horak (2007).   11 1.4 Artificial and Estimated Sensorimotor Delays in Balance  As mentioned above, researchers have extensively used mechanical and/or sensory perturbations in stance to probe balance responses. However, many of the methods rely on imposed perturbations. This poses concern when comparing perturbation-evoked balance responses with standing balance control, which is a quasi-continuous rather than transient process. It is partly for this reason that researchers began using continuously varying stimuli to perturb balance (Fitzpatrick, Taylor, & McCloskey, 1992; Peterka, 2002; Peterka & Loughlin, 2004). This allows the nervous system to achieve a steady state, which can then be used to obtain a transfer function characterizing specific aspects of balance. More recent advances in robotics have allowed researchers to further investigate how we control balance and adapt our movements to novel physical environments in a dynamic state. Here I will discuss a novel robotic system that is capable of manipulating sensorimotor loop delays (1.4.1), differences in perturbation-evoked sensorimotor delays due to age (1.4.2), followed by a pathological model causing delays in sensorimotor loops (1.4.3).  1.4.1 Artificial Delays Using a Robotic Balance Simulator Huryn et al. (2010) developed a robotic system that permits subjects to experience novel body, environment and/or sensory dynamics while maintaining active standing balance. The device works by having subjects stand with their weight supported on a force plate and their body braced to a backboard that is rigidly mounted to the robot. While operating the robot, the subject conceptually enters the control loop of robotic motion, having full control of anteroposterior movement based on forces/moments applied to the force plate. The balance simulations are programmed based on the physical principles of an inverted pendulum, with the ability to alter the physics of the controlled body (stiffness, damping, inertia, height, mass),   12 environment (simulated gravity, laws of motion) and/or sensory dynamics (ankle rotation, electrical vestibular stimuli, visual field). Thus, the robotic device makes it possible to explore proposed models of whole-body stabilization and the control of standing balance when body mechanics are changed.  One particular use of this robotic system is the ability to explore balance control when increased delays in sensorimotor control are introduced. Through the balance simulator, a delay between forces/moments applied to the force plate and the resultant whole-body movement can be manipulated. Thus, the balance simulator manipulates the balance control loop by creating a delay between motor commands and sensory information from whole-body movement. Assuming the central nervous system relies on an internal model to control standing balance, this would result in an error between the actual and the expected sensory feedback from self-generated movements on the robot. Currently unpublished research using this balance simulator has demonstrated that balance performance declines in healthy individuals as artificial sensorimotor loop delays increase up to 500 ms. van der Kooij & Peterka’s (2011) model also suggests that balance declines with increasing delays above 100 ms including complete loss of balance control at delays above 340 ms. Given these findings, how might individuals with slower balance responses to imposed perturbations perform on the robot when artificial sensorimotor loop delays in balance are present compared to individuals with faster balance responses? To investigate these questions I will use age differences in postural response latencies as a model for delayed neural conduction in the central nervous system. As a case study, I will also compare people with Multiple Sclerosis to healthy individuals to see how pathology may affect outcomes for this experiment.   13 1.4.2 Age-Related Changes in Balance Response Latencies  A number of studies have investigated the relationship between postural response latencies to imposed perturbations in stance and age. Early research primarily used surface translations to elicit postural responses, finding modest but significant differences between young and elderly individuals (Peterka & Black 1990; Stelmach & Phillips 1989, Woollacott). This is consistent with decreased nerve conduction velocity observed in the elderly (Wolfson et al. 1985).   Differences in postural response latencies between age groups have also been observed using rotational perturbations in stance. Nardone et al. (1995) saw a significant positive correlation between age and onset latencies for short, medium, and long latency responses in tibialis anterior and soleus between 75 subjects aged 15 to 79. Furthermore, age-related changes in postural response were greater in the medium and long latency responses than short latency response (Nardone et al., 1995). In a separate investigation, Allum et al. (2002) compared postural response latencies between three age groups: young (20-34), middle-aged (35-55), and elderly (60-75). They observed delayed long-latency responses in the middle-aged and elderly groups compared to young but no significant difference between middle-aged and elderly, albeit group comparison were used rather than a correlation analysis.   Whether or not changes in postural response latencies lead to deficits in balance is harder to discern in elderly populations due to the various aspects of balance that can be affected by age, such as decreased vestibular function (see review by Anson and Jeka, 2016), impaired proprioception (see review by Goble et al. 2009), and/or impaired motor function (Porter, Vandervoort, & Lexallm, 1995). For this reason alternative groups that exhibit slow sensorimotor conduction due to pathology may provide an opportunity to specifically compare   14 differences based on neural conduction times. In the following section I will discuss how people with Multiple Sclerosis may provide this opportunity.   1.4.3 Multiple Sclerosis: An Alternative Model for Slowed Sensorimotor Loop Delays In addition to robotic simulations and age-related differences in postural response latencies, it is possible to look at the effects of increased sensorimotor loop delays on standing balance using Multiple Sclerosis as a model for delayed balance responses. Multiple sclerosis is an autoimmune disease of the central nervous system, which is characterized by damage to the myelin sheath. Subsequently, nerve conduction slows in regions in which lesions occur. This provides an opportunity to compare balance ability between people with Multiple Sclerosis and healthy controls based on differences in temporal responses to imbalance.    Perturbation-evoked balance responses in people with Multiple Sclerosis differ from the typical balance responses described in section 1.2. For example, Diener et al., (1984) compared perturbation-evoked balance responses between patients with Multiple Sclerosis and healthy controls using toes-up rotations. Patients with Multiple Sclerosis exhibited normal short and medium latency stretch responses, but delayed long latency responses compared to healthy controls. The observation that long latency responses are delayed in people with Multiple Sclerosis was corroborated in Cameron et al., (2008), where patients with Multiple Sclerosis had significantly delayed balance responses to a surface translation compared to healthy controls (161 ± 31ms vs. 102 ± 21ms respectively). Although the medium latency responses appeared normal in the Diener et al. (1984) study, a disproportionate number of Multiple Sclerosis participants were missing the medium latency response, 28.6% compared to 10% in healthy controls. Diener et al. (1984) did not report lesion location in the Multiple Sclerosis population, but absent medium latency responses could be related to the observation that people with spinal   15 lesions also have disproportionately absent medium latency responses compared to healthy controls, 49% compared to 17% (Diener et al. 1985). This was not observed when lesions were located in the cortex (Diener et al. 1985). Thus, it appears that patients with Multiple Sclerosis have delayed long latency balance responses irrespective of lesion location, while subcortical lesions may abolish or reduce medium latency responses.   Ultimately, delays in central nervous system conduction can lead to impairment in the ability of people with Multiple Sclerosis to maintain standing balance. For example, in addition to delayed balance responses, people with Multiple Sclerosis sway more than healthy controls and have limited and slowed movement toward limits of stability compared to healthy controls (see review by Cameron & Lord (2010)). Huisinga et al. (2014) also reported a correlation between somatosensory evoked potentials and centre of pressure variables during quiet stance. Thus indicating that increased sensorimotor delays in balance responses to perturbations for people with Multiple Sclerosis leads to deficits in their ability to balance. 1.5 Aim and Hypotheses  The first aim of this study was to investigate whether balance stability declines with increasing artificial delays in active balance control. Using the previously described robotic balance simulator (1.4.1), I introduced artificial sensorimotor loop delays (50 to 400ms) between the participants’ motor commands and resulting whole-body movements. I hypothesized that the probability of falling for all participants would increase as delays increased. The second aim of this experiment was to determine whether there is an association between estimates of sensorimotor loop delays and balance stability when presented with artificial sensorimotor loop delays on a robotic simulator. Using pure pitch rotations, onset latencies of balance responses were used as an estimate of sensorimotor loop time. From the logistic regression curves obtained   16 from the robotic balance simulations, I extracted thresholds of stability based on a 5% probability of falling at a given delay. I then compared these thresholds of stability with postural response latencies to rotational perturbations about the ankle joint in individuals aged 19 to 75 under the assumption of variability with age.  As previously described, short and medium latency responses to rotational perturbations are destabilizing, while the long latency response provides a balance-correcting contraction. Furthermore, long latency responses are assumed to have cortical influence that can modify balance responses to sudden states of imbalance and may be the best estimate of overall sensorimotor delay. For these reasons, I hypothesized that thresholds of stability on the robot when artificial sensorimotor delays were present would be associated with the onset of long latency balance responses to rotational perturbations, but not the onset of medium and short latency balance responses.               17 Chapter 2: Experimental Design, Protocol, and Results 2.1 General Experimental Setup  2.1.1 Participants  Thirty healthy individuals aged 19 to 75 (16 Females, 14 Males; 49 ± 19 yrs; 173 ± 9 cm; mean ± SD) were recruited to participate in the experiment. Each participant had no known neurological deficits that would affect their balance based on self-report. Three participants with Multiple Sclerosis with low disability were also recruited for comparative interest (2 Female, 1 Male; 33, 46, & 61 yrs; predominantly supra-spinal lesions). Participants with Multiple Sclerosis were recruited through co-investigator Dr. Rajiv Reebye’s private clinic. The experiment was approved by the University of British Columbia’s Clinical Research Ethics Board (H17-00600) and was conducted in accordance with the ethical guidelines set forth by the Declaration of Helsinki. 2.1.2 Balance Simulator  A robotic balancing apparatus that simulates the mechanics of a standing person (Figure 2.1) was used to manipulate the relationship between the action of the leg and ankle muscles and the movement of the body in the anterior-posterior plane. The robotic platform is comprised of two mechanical pieces (ankle-tilt platform and rotating backboard), each of which is independently controlled by two rotary motors (SCMCS-2ZN3A-YA21, Yaskawa, Japan). For the purposes of this experiment the ankle-tilt platform remained immobile during trials. A real-time computer (PXI-8119; National Instruments, TX, USA) running at 2000 Hz controlled a single actuator that rotated the single-degree of freedom backboard about the ankle joint in the sagittal plane. Participants were braced against the backboard and controlled the motion of the backboard by applying ankle torques to the force platform (OR6-7-1000, AMTI, MA, USA)   18 fixed atop of the ankle-tilt platform. For example, when a participant pushed down with their toes (plantarflexion) with a greater torque than the torque generated by gravity on their centre of mass the backboard accelerated posteriorly. The motion of the backboard was programmed with soft limits of 6° anterior and 3° posterior from vertical. These limits have been shown to be greater than normal angular deviations during stance in young (15-25 years), middle-aged (45-55 years), and elderly (65-75 years) individuals (Gill et al., 2001). Soft motion limits of 50 °/s and 1000°/s2 were also set to encompass the physical limits of sway during standing balance (Pospisil et al., 2012). During trials, the angular displacement of the backboard was digitized and recorded at 2000 Hz using a data acquisition board (PXI-7853R; National Instruments, Austin, TX, USA). The motion of the backboard could be stopped at any time by the experimenter using an emergency stop button. The balance simulations were programmed based on the physical characteristics of an inverted pendulum based the participant’s mass and centre of mass height (Luu et al., 2011). Mass (kg) was determined by taking the participants vertical force (N) while standing on the force plate divided by acceleration due to gravity (9.8 m/s2). Each participant’s centre of mass height was measured by balancing participants on a board positioned on a dowelling. The balance point of the board on the dowelling was found before participants lay on the board in a supine position. They were then asked to shift either up or down the board to find the balance point. Once this was found, the distance between the dowelling and ankle joint determined the centre of mass height. These values were then entered into the balance simulator.       19  Figure 2.1 - Servo-controlled balance simulation device Participants were asked to stand barefoot on a force plate fixed atop of an ankle tilt platform. Participants were braced to a foam-covered metal frame using two seat belts. One seat belt was strapped firmly around the participant’s waist, while the other was strapped over the arms and around the participant’s chest. The frame/backboard independently hinged forwards or backwards in the anterior-posterior plane based on ankle torques applied to the force plate. In all trials participants were facing forwards. Image adapted from Shepherd (2014).  2.1.3 Adjustable Time Delays  On the robot there is a baseline mechanical delay of 20 ms present between ankle torques applied to the force plate and the resultant backboard motion (Shepherd, 2014). This delay can be increased through our computer-controlled balance simulation program. For the purposes of   20 the present research this delay did not exceed 400 ms (i.e. 380 ms entered into the program + 20 ms baseline delay).  2.1.4  Rotating Platform  A dual-axis rotating platform was used to manipulate the ankle joint in the pitch plane while standing (Figure 2.2). Participants stood barefoot with their feet lightly strapped to the platform and their heels placed 15 cm apart. Brackets on the surface of the platform were adjusted so that each lateral malleolus was aligned with the axis of pitch rotation. Throughout the experiment, participants had handrails placed in front and behind the platform and a spotter was present to assist the participant if needed.   The rotating platform was used to present perturbations in the pitch plane with a peak displacement of 7.5° up or down and with a peak velocity of 50°/s (Carpenter, Allum, & Honegger, 2001; Carpenter, Allum, & Honegger, 1999). The displacement of the platform was determined prior to the experiment using Optotrak (Optotrak, Northern Digital Inc., Canada), while the velocity was measured throughout the experiment using an angular rate sensor (LCG50, Sytron Donner Inertial, USA) placed on the rotating platform. A computer monitor was placed directly in front of the participants to give visual feedback about their stance preceding perturbation onset. The monitor displayed a rectangle box representing three standard deviations of a two-minute quiet standing trial for both anterior-posterior and medial-lateral directions. A red dot on the computer monitor was used to represent the current centre of pressure position. Participants were instructed perturbations would be initiated at any time once the red dot was within the target. The researcher initiated perturbations randomly between 1 and 10 seconds once participants were positioned with the red dot in the target area. Participants were also instructed that they may use the safety bar in front of them once a perturbation had been initiated if they felt   21 like they could not maintain their balance. Once the perturbation reached its final position, the rotated platform was held in place for approximately 5 seconds before slowly (1.1°/s) returning back to its initial position (0°).              Figure 2.2 - Rotating platform setup A: Heel brackets are adjusted so that each participant’s lateral malleoli were aligned with the pitch axis of rotation. Velcro foot straps were lightly placed across the top of the foot to maintain foot position. An angular rate sensor located on the rotating platform measured angular rate during perturbation.  B: Participants stood on the rotating platform while looking directly at a computer monitor. The computer monitor displayed a red dot representing their anteroposterior and mediolateral centre of pressure along with an outline of a box representing 3 standard deviations of their 2-minute quiet standing trial.  A B   22 2.1.5 Electromyography  Surface electromyography (EMG) was recorded bilaterally from soleus, medial gastrocnemius, and tibialis anterior. Two electrodes (H69P Cloth Electrodes, Covidien/ Kendall, USA) were placed over each muscle belly approximately 2 cm apart. EMG data were collected at 3000 Hz, amplified 500×, and bandpass filtered between 10 and 500 Hz (Telemyo, 2400R, Noraxon, USA). The data were digitized and sampled at 2000 Hz (Power 1401 with Spike2 software, Cambridge Electronic Design, UK).  2.2 Experimental Procedures, Analysis, and Results  In this experiment, a rotating platform was used to test postural response latencies and a robotic balance simulator was used to test balance stability while artificial sensorimotor delays in balance control were added. Here, I will describe the methods and results for the rotating platform and robotic balance simulator separately. The robotic section also addresses the hypothesis that balance stability declines with increasing imposed delays on the robot, while the final section combines the results from the robot and tilt platform to address the hypothesis relating to associations between perturbation-evoked balance responses and stability on the robotic balance simulator.   2.2.1 Robotic Balance Simulator: Investigating Thresholds of Stability  2.2.1.1 Methods  2.2.1.1.1 Balance Stability  Participants were exposed to 10 induced artificial delays in balance control (50 - 400 ms). Prior to the trials, participants were given time to familiarize themselves with the mechanics of the robotic balance simulator without artificial delays in balance present. Once participants were comfortable with controlling the robot, their feet position on the force plate was marked using   23 tape. They were asked to return to this position if they moved or stepped off the force plate between trials. An initial stance calibration was also performed prior to beginning the experimental trials. Participants were asked to maintain quiet stance starting at zero degrees on the robot for two minutes. The zero degree starting position was used to ensure participants’ starting position wasn’t close to the backward limit, as many people naturally lean back on the robot because of its design. Following the two-minute quiet standing, the mean angle, standard deviation of angular displacement, and the standard deviation of mean angular velocity were calculated. The following parameters were then used to create a range in which the perturbations would be initiated: three standard deviations from the mean angle and one standard deviation from the mean angular velocity (= 0). The deviation from the mean angle was selected to be consistent with the rotating platform protocol. The velocity tolerance was selected to limit oscillations pre-perturbation that may make balance more difficult once the artificial delay is added.  Participants were told that they were going to go through a series of 80 trials (8 repetitions of the 10 time delays). At the beginning of each trial the robotic system was initiated under normal balance conditions. Participants received verbal feedback from the researcher about their whole-body position and angular velocity. They were then instructed that once their upright posture was within the desired position/velocity, the researcher would trigger a setting that may make their balance more difficult. Their task was to maintain their balance without falling to either the anterior (6°) or posterior (3°) limits set by the simulation for 20 seconds. If within those 20 seconds a forwards or backwards limit was hit, the robot was halted and the participant was returned to the starting position for the next trial. This was done to eliminate additional exposure to the induced delays and minimise learning. If the participant balanced for   24 20 seconds without hitting a limit, the robot was stopped after 20 seconds and returned to the starting position for the next trial.   Ten experimentally-induced delays were tested over the 80 trials:  50, 100, 125, 150, 175, 200, 225, 250, 300, and 400 ms. These delays were selected to provide resolution where we predicted balance instability to occur based on pilot observations. The range is also inclusive of the critical delay (340 ms) presented by van der Kooij & Peterka (2011). The 80 trials consisted of 8 groups each with each group containing all 10 delays and with each group of delays randomised in order. The 80 trials were delivered sequentially and participants were told they could take breaks at any point. A baseline catch trial (20 ms) was not used because it was assumed that all participants would be able to balance at 50 ms based on pilot observations. 2.2.1.1.2 Analysis: Variables of Stability  The angular position of the backboard was recorded in each trial on the robotic platform (See Figure 2.3 for raw angular traces). To test the hypothesis that balance stability declines with increasing artificial delays in balance control three variables were obtained. The first variable was the average number of ‘virtual falls’ at each delay, as defined by a forward or backward lean that exceeded the angular limits of the robotic balance simulator (3° posterior and 6° anterior) within 20 seconds. The second variable was the average time to a  ‘virtual fall’. This was defined as the time from delay onset to the occurrence of a ‘virtual fall’. The third variable was the average standard deviation of angular backboard displacement at each delay measured from delay onset to the occurrence of a ‘virtual fall’.  Balance stability thresholds were determined based on the number of ‘virtual falls’ out of 8 trials for each delay. A logistic regression curve was then fitted to each participant’s data using   25 the ‘glmfit’ ‘logit’ function in Matlab (R2016a, MathWorks Inc, USA). This function returns two values that correspond to the slope and intercept of the following transformation function:    log (p/1-p) = mx + b  Where p is the probability of a ‘virtual fall’ occurring, m is the slope, b is the intercept, and x is the artificial delay. Thus the logistic regression curve provides the probability of a fall occurring at a given artificial delay (x). From each participant’s logistic regression curve, the delay in which the probability was 0.05 was determined (See Figure 2.4 for example of fitted curve). Given the threshold of stability is meant to indicate a point in which a participant is no longer able to reasonably control balance, this value was believed to indicate a reasonable probability of falling.                  26            Figure 2.3 - Example of angular displacement traces for 9 trials  For illustrative purposes angular displacement traces from 9 of the 10 delays are presented (125 ms not included). In each trial the participant attempted to remain within the angular limits set by the simulation (3° posterior and 6° anterior) for 20 seconds. If a limit was hit the trial was manually stopped. In this diagram the red dots represent a ‘virtual fall’ occurring within the 20 seconds trial. The number of ‘virtual falls’ per delay, the time to ‘virtual fall’, and the standard deviation of angular backboard displacement were recorded. Grey traces indicate angular displacement after a ‘virtual fall’ occurred, but before the device was manually stopped data not included in the analysis).     27  Figure 2.4 Logistic regression curves and thresholds of stability  Representative data from a single participant. A logistic regression curve was fit to the number of ‘virtual falls’ out of 8 for each delay. From the curve the delay in which a 5% probability of falling (represented by the red dot) was determined and used in the correlation analysis between balance stability and postural response latencies. Background represents the range of curves across all participants.  2.2.1.1.3 Statistical Analysis A one-way repeated measures ANOVA was also used to test the differences in group means for the following three dependent variables: average SD of angular backboard displacement, average time to fall, and average number of falls. A linear regression analysis was   28 performed with age as the independent variable and thresholds of stability as the dependent variable. Additionally, a one-way ANOVA was performed to determine whether mean thresholds of stability differed between three age groups: young (20-34 years), middle-aged (35-55 years), and elderly (60-75 years) for consistency with EMG analysis. The statistical software JASP (0.8.6, The JASP Team, Netherlands) was used to run the ANOVAs and post-hoc analysis, while the linear regression analysis and t-tests were performed using RStudio (Rstudio Inc, USA).  Statistical significance was set at p < 0.05. Results are presented as means ± standard deviation.  2.2.1.2 Results  All participants were able to balance at the 50 ms delay with the exception of a single trial in one elderly participant (>70 yrs).  Conversely, no participants were able to balance for the duration of the 400 ms condition and in only one trial was a participant able to balance for the duration of a 300 ms condition (see Figure 2.6 for mean number of falls). The standard deviation of angular backboard displacement increased as the delays increased with the exception of the 400 ms condition, which was slightly lower than the 300 ms condition (see Figure 2.6). The average time from the beginning of a trial to a ‘virtual fall’ decreased as delay conditions increased (see Figure 2.6).   A one-way repeated measures ANOVA revealed that balance stability decreased with increasing delays for each of the three variables presented in Figure 2.6. This was demonstrated by the increase in the standard deviation of angular backboard displacement (F5,142 = 101.1, p < 0.001), increase in the average number of falls (F3, 82 = 218.2, p < 0.001), and decrease in the average time to fall (F3, 77 = 147.2, p < 0.001) as delays increased. In all cases a Tukey HSD test revealed a significant difference between the 50 ms condition and all other delays (p < 0.001).   29 The mean threshold of stability across all participants on the robot was 98 ± 38 ms. The linear regression analysis showed no significant association between age and the threshold of stability (R2 = 0.01; p = 0.58; see Figure 2.5 for regression line). A quantile-quantile (Q-Q) plot was used to confirm normality visually. Furthermore, age group comparisons revealed no significant differences in the mean thresholds of stability for the three age groups (F2,25 = 2.7, p = 0.09; young: 93 ± 28 ms; middle-age: 120 ± 49 ms; elderly: 83 ± 24 ms).        30  Figure 2.5 – Linear regression of age vs. threshold of stability  A linear regression analysis was performed between age and the threshold of stability determined from the thresholds of stability for each participant (see Figure 5). The regression line is given by y = -0.21x + 109.9 (R2 = 0.01; p = 0.58).   31    Figure 2.6 – Dependent variables from robotic simulations Illustration of the changes in the average number of falls, average time to fall, and average standard deviation of angular backboard displacement for all participants (n = 31) on the robot.  The error bars represent the 95% confidence intervals for each condition.   32 2.2.2 Rotating Platform: Determining Postural Response Latencies  2.2.2.1 Methods  2.2.2.1.1 Balance Response Latencies  Prior to beginning the experimental trials on the rotating platform normal quiet standing was measured. Standing on the rotating platform, initial stance was measured by asking participants to maintain quiet stance with their arms by their side looking directly forward for two minutes. Between trials, participants were asked to return to this stance by receiving visual feedback about their anterior-posterior and medial-lateral centre of pressure position from a computer monitor placed directly in front of them at eye level (See section 2.1.4).  Following the normal stance measurement, participants were informed that they would go through a series of 40 trials consisting of toes-up and toes-down perturbations. They were not informed about the three separate conditions: Serial Toes-Up, Serial Toes-Down, and Random Toes-Up & Toes-Down. The Serial Toes-Up condition consisted of 10 upward tilts of the platform rotating the ankle into dorsiflexion. The Serial Toes-Down condition consisted of 10 downward perturbations of the platform rotating the ankle into plantarflexion. These were done to habituate the participants to the perturbation, eliminating first trial responses in the Random Toes-Up Toes-Down trials (Nashner, 1976). For further investigations not related to this thesis, Serial Toes-Up always preceded the Serial Toes-Down condition. The Random Toes-Up & Toes-Down condition was performed after the serial perturbation conditions. It consisted of 20 random toes-up or toes-down perturbations, with 10 of each being presented. Data from the Random Toes-Up & Toes-Down were used to determine short, medium, and long latency balance responses in tibialis anterior, soleus, and medial gastrocnemius. For all conditions the perturbations were manually initiated between 1 and 10 seconds once participants had returned   33 to their normal quiet standing position; i.e. the red dot within the box. Participants were given the opportunity to sit and rest between trials if needed.  2.2.2.1.2 Analysis: Quantifying Balance Response Latencies Surface EMG was used to bilaterally record muscles activity in soleus, medial gastrocnemius, and tibialis anterior. The data were full-wave rectified offline to determine the onset latency of each trial. An angular rate sensor adhered to the rotating platform was used to determine perturbation onsets. This was defined as platform rotation exceeding 2 standard deviations of the mean background noise. A semi-automated algorithm was used to identify the onset of EMG activity in each trial. The onset was identified as the first muscle activity exceeding the mean plus 2 standard deviations of background muscle activity 500 ms prior to stimulus. Additionally, the activity had to remain supra-threshold for at least 50 ms in long latency responses, 30 ms in medium latency responses, and 20 ms in short latency responses with returns below the threshold allowed for no more than 3 ms. All EMG onsets were visually inspected to ensure that the algorithm accurately identified the onset. Any mistakes from the algorithm were adjusted manually and any trials in which the onset could not be identified were removed from the analysis. A minimum of 3 identifiable responses out of 10 in at least one leg was required for onset latencies to be included in the analysis for each participant. If both legs had 3 identifiable responses or more the onset latencies were averaged between legs. The balance response latencies of the participants with Multiple Sclerosis were analyzed separately to identify potential bilateral differences due to pathology.  For comparison with previous research, the mean onset latencies were also analyzed based on the age groups used by Allum et al. (2002): young (20-34 years), middle-aged (35-55 years), and elderly (60-75 years).    34 2.2.2.1.3 Statistical Analysis  A linear regression analysis between age as the independent variable and postural response latency for short latency responses in soleus, medium latency responses in tibialis anterior, and long latency responses in both soleus and tibialis anterior. Normality was visually verified for all regression lines using quantile-quantile (Q-Q) plots in RStudio. For comparison with previous research, a one-way ANOVA was used to investigate differences in mean onset latencies between the following age groups: young (20-34 years), middle-aged (35-55 years), and elderly (60-75 years). The statistical software JASP (0.8.6, The JASP Team, Netherlands) was used to run the ANOVAs, while the linear regression analysis and t-tests were performed using RStudio (Rstudio Inc, USA). Statistical significance was set at p < 0.05. Results are presented as means ± standard deviation. 2.2.2.2 Results  All participants, including the three individuals with Multiple Sclerosis completed the tilt platform portion of this experiment. Short latency responses were determined from soleus responses to toes-up perturbations only. This was due to unclear short latency responses in a large number of trials from medial gastrocnemius. The medium latency response was also difficult to distinguish from short latency responses in a large number of participants. For this reason the medium latency responses were only obtained from tibialis anterior (toes-down). The long latency responses were identifiable in all three muscles (Sol, mGas, TA).  The short latency responses in soleus were identifiable in 26 out of the 30 healthy participants and all three of the Multiple Sclerosis participants. Of the healthy participants the average short latency response was 50.8 ± 3.2 ms and 52.7 ± 3.1 ms in participants with Multiple   35 Sclerosis. In the healthy population, no association between age and the short latency response was observed (R2 = 0; p = 0.79).  The medium latency responses in tibialis anterior were identifiable in 28 out of the 30 healthy participants. The average medium onset latency for healthy individuals was 98 ± 11 ms and no significant correlation was found between age and the medium latency stretch response (R2 = 0.04; p = 0.26).  In one participant with Multiple Sclerosis the medium latency responses were absent in the right leg with an average onset latency in the left leg of 115 ± 6 ms. In another participant a 25 ms difference between limbs was observed (Left: 153 ± 12 ms; Right: 128 ± 9 ms), while a bilateral difference of 15 ms was observed in the third participant (Left: 114 ± 3 ms; Right: 99 ± 12 ms). The average bilateral difference in healthy participants for the medium latency response was 6 ± 4 ms.  Long latency responses were observed in 26 participants in soleus, 25 in medial gastrocnemius, and 29 in tibialis anterior. The mean onset latencies for the three muscles were 167 ± 22 ms in soleus, 178 ± 27 ms in medial gastrocnemius, and 147 ± 12 ms in tibialis anterior. In all cases there was no correlation between age and onset latency (Sol: R2 = 0.10, p = 0.12; mGas: R2 = 0.05, p = 0.29;  TA: R2 = 0.13, p = 0.72). See Figure 2.7 for illustration of short, medium, and long latency response regression lines.          36             Figure 2.7 – Regression lines for age vs. onset latencies  Linear regression analyses revealed no association between age and onset latencies for all responses (all p > 0.05). The regression lines for each response were given by: Sol SLR = 0.01x + 50.4; Sol LLR = 0.34x + 148.5; TA MLR = 0.13x  + 91.8; TA LLR = 0.05x + 144.7; mGas LLR = 0; mGas LLR = -0.32x + 193.6.  37  In participants with Multiple Sclerosis, bilateral differences in long latency responses were observed. In one participant a 35 ms difference in soleus (left = 161 ± 12 ms; right = 196 ± 6 ms), 55 ms difference in medial gastrocnemius (left = 169 ± 6 ms; right = 224 ± 13 ms.) and 104 ms difference in tibialis anterior (left: 135 ± 10 ms; right = 239 ± 4 ms) was observed.  In a second participant a bilateral difference of 22 ms in soleus (left: 198 ± 16 ms; right 176 ± 22 ms) and 19 ms difference in medial gastrocnemius (left = 213 ± 10 ms; right = 194 ± 26 ms) was observed with no difference between onset latencies in tibialis anterior (left = 160 ± 8 ms; right = 156 ± 10 ms). In the third participant with Multiple Sclerosis no difference between soleus (left = 143 ± 10 ms; right = 145 ± 8 ms) and tibialis anterior (left = 144 ± 9 ms; right = 148 ± 6 ms) muscles was observed and only responses in the right leg for medial gastrocnemius were identifiable. A table summary of each participant can be found in the appendix. The average bilateral differences in healthy individuals were as follows: Sol = 7 ± 7 ms, mGas = 10 ± 9 ms, TA = 4 ± 4 ms. Group comparisons between young, middle-aged, and elderly participants did not reveal significant differences in mean onset latencies for the short, medium, and long latency responses (see Table 2.1 for group averages and statistics).         38  Table 2.1 – Average EMG onset latencies for each age group  Muscle Response Young Middle-aged Elderly F (df) p  TA MLR 98 ± 13 100 ± 12 98 ± 7 0.14(2,24) 0.87 LLR 147 ± 12 147 ± 13 147 ± 10 0(2,25) 0.99 Sol SLR 51 ± 4 51 ± 3 51 ± 3 0.07(2,23) 0.93 LLR 155 ± 14 171 ± 23 171 ± 26 1.4(2,21) 0.27 mGas LLR 178 ± 24 179 ± 32 174 ± 25 0.09(2,21) 0.92  Mean (ms) ± SD               39    Figure 2.8 - Example of average onset latencies for one participant For illustrative purposes raw EMG was rectified and averaged over 10 trials for each of the toes-up and toes-down conditions. The characteristic short, medium, and long latency responses (SLR, MLR, and LLR respectively) are evident in the above traces. The medium latency responses in soleus and medial gastrocnemius were not analyzed and medium latency responses in tibialis anterior were used instead.    40  Figure 2.9 - Example of bilateral differences in one individual with Multiple Sclerosis  In one participant with Multiple Sclerosis there were pronounced bilateral differences in postural responses. Red traces represent muscles from the right leg, while black traces, muscles from the left leg. In both the toes-up and toes-down conditions, the medium latency responses were absent for both soleus and tibialis anterior respectively. Note the delayed long latency response in the right tibialis anterior compared to the left.       41  Figure 2.10 – Relationship between age and long latency responses in tibialis anterior Linear regression analysis between age and onset latency was performed for all responses analyzed. The linear regression line is given by y = 0.05x + 144.7 (R2 = 0.13; p = 0.72). For illustrative purposes, participants with multiple sclerosis are included in red with L and R representing the left and right legs respectively.       42 2.2.3 Association Between Thresholds of Stability and Balance Response Latencies  2.2.3.1 Statistical Analysis To test the hypothesis that there would be an association between long latency balance responses and balance stability on the robotic balance simulator, but not short and medium latency responses, a linear regression analysis was performed between thresholds of stability on the robot and onset latencies for all perturbation-evoked balance responses analyzed on the rotating platform for healthy participants. 2.2.3.2 Results  As expected there was no significant association between thresholds of stability and short latency responses in soleus (R2 = 0, p = 0.9). This was also true of the association between thresholds of stability and medium latency responses in tibialis anterior (R2 = 0, p =0.9). The hypothesis that there would be an association between long latency responses and thresholds of stability was also supported in tibialis anterior (Figure 2.11; R2 = 0.14, p = 0.047). However, the long latency responses in soleus and medial gastrocnemius did not reveal an association with thresholds of stability (Sol: R2 = 0, p = 0.83; mGas: R2 = 0.08, p = 0.16).     43  Figure 2.11 – Long latency responses vs. threshold of stability  Linear regression analysis between long latency responses in tibialis anterior and thresholds of stability on the robot. The linear regression line is given by y = -1.16x + 268.6, R2 = 0.13, p = 0.047.          44 Chapter 3: Discussion In this experiment, I characterized postural stability on a robotic system that can induce artificial delays in balance control, while also comparing stability with onset latencies evoked by rotational perturbations in stance. This was performed in both healthy subjects ranging from 19 to 75 years old and three individuals with Multiple Sclerosis for pilot comparisons. In section 3.1 I will discuss the results from the robotic simulator and how this portion of the experiment supported the hypothesis that balance stability would decrease as artificial delays increase (3.1). I will then discuss the results from the rotary platform (3.2). This will be followed by an interpretation of the regression analysis between long latency responses and stability thresholds on the robot (3.3). Furthermore, I will discuss how the participants with Multiple Sclerosis compared to the healthy individuals and how future studies may incorporate clinical populations to help answer the hypothesis of this experiment as well as potential rehabilitative uses for the robot (3.4). 3.1 Robotic Balance Simulator  3.1.1 Sensorimotor Delays and Loss of Stability  The sensorimotor control of balance is a mechanically unstable process requiring the central nervous system to constantly adjust muscle activity to maintain upright posture. Inherent to the sensorimotor control of balance are delays between sensory information, cortical processing, and subsequent motor commands. It has been theorized through a computational model that increased delays in balance control lead to decreased stability (van der Kooij & Peterka, 2011). This experiment has provided an opportunity to investigate how postural stability is altered when artificial delays in balance control are introduced.   Participants were exposed to artificial delays in balance control ranging from 50 to 400   45 ms. This range is inclusive of the critical delay of 340 ms proposed by van der Kooij & Peterka (2011). In their model, the normal time delay parameter was 97 ms, thus implying that an increase in the time delay of 243 ms or greater would lead to complete loss of stability. This is comparable to the results observed in this experiment. For example, both the 300 and 400 ms conditions were near impossible for all participants (1 successful trial out of 240 trials was observed at 300 ms). Furthermore, the 225 ms and 250 ms conditions were also difficult across participants with the average number of falls out of 8 trials being 7 ± 1.5 and 7 ± 1.2 respectively (Table 2.1).  Although this study was not designed to validate the model proposed by van der Kooij & Peterka (2011), the results support the postulation that balance control becomes unstable as delays in balance control increase. As was presented in section 2.3, both the average standard deviation of angular backboard displacement and the average number of ‘virtual falls’ were significantly larger than the 50 ms condition. Furthermore, the average time to fall was significantly smaller than the 50 ms condition. This would suggest that stability begins to decline between 50 and 100 ms, which is also reflected in the average threshold of stability across participants: 98 ± 38 ms.  3.1.2 Adaptation Effect  The protocol on the robotic balance simulator was designed to minimise adaptation. By randomizing the presentation of the delays used in each of the 8 groups it was expected that participants would be unable to adapt to the perturbations during a trial. Furthermore, halting a trial immediately after a ‘virtual fall’ was meant to reduce the amount of time participants would have to practice. However, post-experimental analysis revealed that an adaptation affect across participants did in fact occur.    46  To determine whether there was an adaptation effect on the robot the number of ‘virtual falls’ each participant had in each of the 8 rounds was determined. A one-way repeated measures ANOVA using a Greenhouse-Geisser correction revealed that there was a significant difference in the mean number of errors in at least one round (F4, 30 = 12.40, p < 0.001). A Tukey HSD test revealed that the number of ‘virtual falls’ in the first round was significantly greater than all other rounds (Figure 3.1; p = 0.024 between round 1 and 2; p < 0.024 between round 1 and all other rounds). For further reference Figure 3.2 illustrates the average number of virtual falls at a given delay in each round.   Figure 3.1 - Average number of ‘virtual falls’ per round  Here I present the average number of ‘virtual falls’ per round for all participants on the robot (mean ± 95% CI) A Tukey HSD post-hoc analysis revealed the number of ‘virtual falls in the first round was significantly greater than all other rounds (*p ≤ 0.024).    47     Figure 3.2 The average number of falls per round per delay The above figure illustrates the change in the average number of ‘virtual falls’ occurring at each delay in each round.   There are number of reasons why participants may adapt on the robot throughout the experiment. The first is that participants simply become more familiar with operating the robot as time goes on. To try to mitigate adaptation participants were given time to familiarize themselves with the robot without delays present. However, they never truly understood what the perturbations would be like until the experimental trials began. Given that the middle 4 rounds were not significantly different from each other in the number of ‘virtual falls’, future investigations may choose to have practice rounds to eliminate the initial adaptation observed.   48 This, however, would not account for later adaptation observed in rounds 7 and 8. Another reason for adaptation could be the realization that reducing responses to the perturbations can reduce the difficulty of the task. This occurred on at least two occasions where participants verbally expressed their belief that the task would be easier if they dampened their reactions to the perturbation, which is indeed true. Ultimately, the observation that adaptation occurred affects the test-retest reliability in this experiment and brings the validity of the stability thresholds into question. To avoid adaptation, future investigations that employ a similar approach should consider alternative methods to reduce or compensate for the learning effect. One possibility would be to use a staircase method commonly used in psychophysical experiments to determine a desired threshold. This could be done in either a top-down or bottom-up way. For example, participants could start at a difficult delay where ‘virtual falls’ are expected to occur 100% of the time, such as 400 ms. The delays could then be decreased incrementally until a successful trial is completed (no ‘virtual fall’). At this point the delay would be increased again in the same increments until failure occurs. This would be repeated until a desired range of successful responses is achieved. The same process could be implemented starting at easy delays and working in steps towards difficult delays. 3.2 Perturbation-Evoked Balance Responses 3.2.1 EMG Onset Latency Comparisons with Current Literature   Although Allum et al. (2002) only reported average long latency responses for pure pitch perturbations in tibialis anterior, their observed onset latencies were earlier than the onset latencies observed in this experiment: young: 114 ± 7 ms, middle-aged: 128 ± 19 ms, elderly: 132 ± 11 ms compared to 147 ± 12 ms. Despite using the same angular displacement, the onsets   49 reported in Allum et al. (2002) were obtained from rotations with a peak angular velocity 10°/s faster than the current experiment. This could be one explanation for the observed differences considering increased angular velocities results in earlier onsets (Nardone, Corrà, & Schieppati, 1990).   Further comparisons with publications involving perturbation-evoked responses to ankle rotation appear to validate the onset latencies observed in this experiment. For example, Nardone et al. (1995), presented group means ± standard deviation for young (<30 years), middle-aged (30-59), and elderly (>59 years) participants in soleus and tibialis anterior for short, medium, and long onset latencies. In Nardone et al. (1995) the long latency responses in tibialis anterior were 143 ± 15.2 ms, 145.2 ± 17.2 ms, and 156.1 ± 15.6 ms for the respective age groups compared to the group mean of 147 ± 12 ms observed in this thesis. Nardone et al. (1995) also presented mean long latency responses in soleus for each age group: 163.8 ± 10.1 ms, 165.8 ± 14 ms, and 180.9 ± 10 ms respectively. These are comparable to the mean long latency response observed in soleus for this thesis: 167 ± 22 ms. Despite the difference in angular displacement between experiments (3° vs. 7.5°), the onset latencies observed in this experiment appear to be within ranges reported in Nardone et al. (1995). Comparable results were also found in Nardone, Corrà, & Schieppati  (1990). Therefore, it is reasonable to believe that the onset latencies presented in this thesis are consistent with previous research.  3.2.2 Negative Results of Correlation Between Age and Perturbation-Evoked Balance Responses In reviewing the literature for this experiment it was assumed that onset latencies from the rotational perturbations would increase with age, particularly in the long latency responses. As was presented in section 2.5, a relationship between age and onset latencies was absent for all   50 responses analyzed. Here I will discuss possible reasons as to why this relationship was not observed in the experiment.   The methods in this experiment were most comparable to Nardone et al. (1995), where positive correlations between age and short, medium, and long latency responses were observed. In Nardone et al. (1995) the tilt perturbations used had the same angular velocity, however the angular displacement used was 3° rather than the 7.5° used here. The major difference, however, was the number of participants: 75 participants ranging from 15 to 79 years old (Nardone et al., 1995). Given that in this thesis less than 30 participants had identifiable responses in each of the short, medium, and long latency responses it is possible that increasing the sample size may have lead to a significant association consistent with Nardone et al. (1995). Indeed side-by-side comparisons of the data set from Nardone et al. (1995) and the data set in this thesis validates the observations of this experiment (Figure 3.2).  51  Figure 3.3 Comparison of perturbation-evoked EMG responses with Nardone et al. (1995) The above comparison was made to illustrate the similarities between the data set in this thesis (right) compared to Nardone et al. (1995) where a significant association between age and onset latencies was found. Image adapted with permission from Nardone et al. (1995).  TA Sol TA Sol Toes-­‐up Toes-­‐down   52 Allum et al. (2002) also reported significant differences in the onset of long latency responses between young individuals (20-34 years) and both middle-aged (35-55 years) and elderly (60-75 years) individuals. Allum et al. (2002) used group comparisons rather than correlation analysis and only reported average onsets for long latency responses in tibialis anterior (young: 114 ± 7 ms; middle-aged: 128 ± 19 ms; elderly: 132 ± 11 ms). However, comparisons using the same age groups as Allum et al. (2002) and the data form this thesis did not reveal significant differences in mean onset latencies between the age groups (See Table 2.1). The purpose of varying ages for this thesis, though, was to get a range of onset latencies, and whether or not there were significant differences between age groups was not a primary interest.  3.3 Correlation Between Balance Stability Thresholds and Balance Response Latencies 3.3.1 Interpretation of Results In this study I used perturbation-evoked balance responses to estimate sensorimotor loop delays in balance control. The temporal onset of short, medium, and long latency postural responses to the rotary perturbations were then compared to stability on the robotic balance simulator. As such, the present experiment provided an opportunity to compare estimates of sensorimotor loop delays with the active balance state under varying delays in sensorimotor balance control. The nature of short and medium latency responses to rotational perturbations are considered destabilizing, as contractions from stretch responses result in movement in the same direction as the perturbation. For this reason, it was hypothesized that the sensorimotor loops involved in these pathways would not contribute to balance stability on the robot when artificial delays in balance control were present. This was validated by the results in this study where both the short and medium latency responses showed insignificant associations in the linear regression   53 analysis between balance stability thresholds and onset latencies (SLR: R2 = 0, p = 0.9; MLR: R2 = 0, p = 0.9).   It was also hypothesized that long latency responses would be correlated with balance stability thresholds on the robot. This was based on van der Kooij & Paterka’s (2002) model of balance control that suggests balance stability declines as internal delays in balance control increase. Using the long latency response as an estimate of the sensorimotor loop delay in balance, I hypothesized that participants with longer balance response latencies would become unstable at a lower artificial delay on the robot compared to a participant with shorter balance response latencies. The assumption was that the two pathways targeted are similar enough that the estimate of the sensorimotor loop delay plus the artificial delay represents overall delay in balance control for each participant (balance response latency + artificial delay = total delay in balance control). The results of this experiment partially support this hypothesis in that a significant association between balance stability thresholds and long latency balance correcting responses was observed in one of the three muscles analyzed (TA: R2 = 0.14, p = 0.047; Sol: R2 = 0, p = 0.83; mGas: R2 = 0.08, p = 0.16).   The weak correlation between estimates of sensorimotor delays and thresholds of stability on the robot may suggest that these two measure target different pathways involved in standing balance. As was alluded to in the introduction, balance responses evoked by sudden imposed perturbations do not necessarily reflect how the central nervous system controls the active balance state under normal conditions. The two perturbations used in this study were similar in that they were both purely sagittal. The major difference, however, was that the balance control loop was being manipulated on the robot during a continuous balance task compared to the rotary perturbations, which were sudden externally imposed perturbations in   54 stance. Thus, it cannot be assumed that the central nervous system reacts to either perturbation in the same way and this could be the reason for such a weak association between the two variables.  3.4  Observations in Participants with Multiple Sclerosis As a sub-component to this experiment, I was also interested in how people with Multiple Sclerosis responded to perturbations and maintained their stability on the robot compared to the healthy controls. Here, I will discuss how each of the three individuals with Multiple Sclerosis compared with healthy participants.   Only one of the participants with Multiple Sclerosis was able to complete both the robotic and tilt platform portions. This particular individual was recently diagnosed with Multiple Sclerosis having lesions around the ventricles as well as in the occipital and frontal lobes. Additionally, this participant did not present any symptoms related to disability as reported by Dr. Reebye. On the robotic platform, this individual exhibited a high threshold of stability (4th highest of all participants), thus indicating that pathology was not a factor in decreasing stability on the robotic balance simulator compared to healthy participants. The pathology also did not appear to cause substantial differences in postural responses with latencies being approximately the same or faster than the group averages.  Only the medium latency response showed a difference between limbs (Left: 114 ± 3 ms; Right: 99 ± 12 ms; See MS participant #1 in Table A.1 of Appendix).  In another participant with Multiple Sclerosis there were marked differences between onsets in each limb for the medium and long latency responses. In this particular individual, over 20 lesions are located in supraspinal regions and three are located in the posterior cervical spine resulting in spasticity and weakness on the right side of the body among other symptoms. These lesions were seemingly associated with reduced or abolished medium latency responses in   55 muscle recorded in the right lower limb (See Figure 2.7). Diener et al. (1985) also observed that people with spinal lesions had abolished or reduced medium latency responses, however, responses were present when lesions occurred only in the cortex. As this participant had both spinal and supraspinal lesions, the results can only provide evidence that medium latency responses are mediated in regions rostral to the cervical spinal cord. The long latency was present, but delayed compared to the onset in the left leg, particularly in tibialis anterior (left: 135 ± 10 ms; right = 239 ± 4 ms). This is also consistent with observations in Diener et al. (1985) where long latency responses were delayed in participants with both spinal and cortical lesions. For this participant’s response latencies see MS participant #2 in Table A.1 of Appendix.  In the third participant with Multiple Sclerosis there were bilateral differences in onset latencies between limbs that were more pronounced in the medium latency response than the long latency responses. In the medium latency response in tibialis anterior, a 25 ms difference between limbs, while in the long latency responses 22 ms, 19 ms, and 4 ms differences were observed in soleus, medial gastrocnemius, and tibialis anterior. The variation in the long latency responses, however, was much greater and suggests that the differences may not have been as pronounced as the medium latency responses for this individual (See MS participant #3 in Table A.1 of Appendix). Pathologically, this participant had brainstem and cerebellar lesions only, and could thus explain why the long latency response was not as affected as the medium latency response. The fact a bilateral difference was observed in the medium latency response also helps support the postulation that medium latency responses are mediated in subcortical region rostral to the spinal cord.   As expected, pathology affected the perturbation-evoked balance responses analyzed in at least two of the participants tested with Multiple Sclerosis. The effects in the third participant   56 were less clear, possibly due to the early nature of that participant’s diagnosis. Furthermore, the effects appear to be dependent on lesion location and future studies that use participants with Multiple Sclerosis on the robotic balance simulator should consider how bilateral differences might affect postural stability. Future investigations should also consider how the robotic device could potentially function as a rehabilitative tool, as the robotic balance simulator provides a safe environment for participants with Multiple Sclerosis or other neurological disorders to practice and possibly strengthen balance with or without bilateral differences. Indeed unpublished pilot data indicates that participants with Multiple Sclerosis can reduce the number of ‘virtual falls’ at delays equal or less than 300 ms following training for 240 minutes over 3 weeks at a delay of 300 ms. This is consistent with adaptations seen in healthy individuals in a previous unpublished study. Anecdotal evidence also suggests the robot may help individuals with spasticity. It is unclear though whether training on the robot leads to real changes in balance stability off the robot. Thus, future investigations may focus on quantifying changes in balance off the robot following training.  3.5 Considerations and future directions  The experiment presented in this thesis was exploratory in nature. In particular, the methods used on the robotic simulator were novel and had not been used in any previous experiments. Previous experiments using the robot relied on outcome variables from trials in which participants continuously balanced for the duration of a trial irrespective of the number of ‘virtual falls’. For the purposes of this experiment, the desired outcome variable was the delay in which a participant could not stabilize themselves. For this it was believed that a protocol based on whether a participant could successfully complete a trial or failed  (‘virtual fall’) would be   57 most appropriate. The results from the logistic regression analysis appeared to support this method, as curves generally fit the data well. However the experimental protocol was unable to prevent adaptation effects, which affects the reliability of the results. Altogether though, the protocol developed for the robotic simulations is a positive step forward for future investigations that may wish to explore the influence of artificial sensorimotor delays on balance stability. This study also provided an opportunity to compare perturbation-evoked balance responses in healthy individuals and people with Multiple Sclerosis. As was expected pathology lead to differences in the muscular responses to perturbations compared to healthy participants. In particular, bilateral differences in onset latencies were apparent in at least two participants with Multiple Sclerosis, validating observations made previously (Cameron et al. 2008). Unfortunately, only one participant with Multiple Sclerosis was able to complete the protocol on the robotic balance simulator. This did not allow much comparison between perturbation-evoked balance responses and balance stability on the robot compared to healthy participants. In the future it may be worth comparing participants with severe bilateral or unilateral delays in perturbation-evoked balance responses due to Multiple Sclerosis with the results of balance stability on the robot in healthy individuals.            58 Chapter 4: Conclusion The experiment in this thesis was designed to investigate how balance stability changes as delays in sensorimotor balance control increase. Furthermore, this experiment sought to compare estimates of sensorimotor delays from perturbation-evoked balance responses with balance control during conditions where artificial delays in balance control were introduced. The experiment supported the hypothesis that balance becomes unstable as delays in balance control increase. The experiment also supported the postulation that there would be no significant association between balance stability thresholds on the robot and both the short and medium latency responses in soleus and tibialis anterior respectively. Finally, this experiment partially supported the hypothesis that balance stability thresholds would be correlated with long latency responses, as this was only found to be significant in tibialis anterior, but not soleus or medial gastrocnemius.  The weak correlation between estimates of sensorimotor delays and thresholds of stability on the robot suggests that these measures target different pathways involved in standing balance.   59 Work Cited Allum, JHJ., & Büdingen, HJ. (1979). Coupled Stretch Reflexes in Ankle Muscles: An Evaluation of the Contributions of Active Muscle Mechanisms to Human Posture Stability. Progress in Brain Research, 50, 185–195.  Allum, JHJ., Carpenter, MG., Honegger, F., Adkin, AL., & Bloem, BR. (2002). Age-dependent variations in the directional sensitivity of balance corrections and compensatory arm movements in man. The Journal of Physiology, 542(2), 643–663. Anson,	  E.,	  &	  Jeka,	  J.	  (2015).	  Perspectives	  on	  Aging	  Vestibular	  Function.	  Frontiers	  in	  Neurology,	  6,	  269.	  	  Beckley, DJ., Bloem, BR., Remler, MP., Roos, RAC., & Van Dijk, JG. (1991). Long latency postural responses are functionally modified by cognitive set. Electroencephalography and Clinical Neurophysiology/Evoked Potentials Section, 81(5), 353–358.  Bolton, DAE. (2015). The role of the cerebral cortex in postural responses to externally induced perturbations. Neuroscience and Biobehavioral Reviews, 57, 142–155.  Cameron, MH., Horak, FB., Herndon, RR., & Bourdette, D. (2008). Imbalance in multiple sclerosis: a result of slowed spinal somatosensory conduction. Somatosensory & Motor Research, 25(2), 113–22. Cameron, MH., & Lord, S. (2010). Postural Control in Multiple Sclerosis: Implications for Fall Prevention. Current Neurology and Neuroscience Reports, 10(5), 407–412. Campbell, AD., Dakin, CJ., Carpenter, MG., (2009). Postural responses explored through classical conditioning. Neuroscience, 164(3), 986–997.  Carpenter, MG., Allum, JH., & Honegger, F. (2001). Vestibular influences on human postural control in combinations of pitch and roll planes reveal differences in spatiotemporal processing. Experimental Brain Research, 140(1), 95–111.  Carpenter, MG., Allum, JH., & Honegger, F. (1999). Directional sensitivity of stretch reflexes and balance corrections for normal subjects in the roll and pitch planes. Experimental Brain Research, 129(1), 93–113.  Carpenter, MG., Frank, JS., Adkin, AL., Paton, A., & Allum, JH. (2004). Influence of postural anxiety on postural reactions to multi-directional surface rotations. Journal of Neurophysiology, 92(6), 3255–65.  Cotton FS (1931). A new method for finding the height of the centre of gravity in man, with some applications. Australian Journal Experimental Biolology & Medical Science 8: 53-67.  Diener, HC., Bootz, F., Dichgans, J., & Bruzek, W. (1983). Variability of postural “reflexes” in humans. Experimental Brain Research, 52(3), 423–428.  Diener, HC., Dichgans, J., Hülser, P.-J., Buettner, U.-W., Bacher, M., & Guschbauer, B. (1984). The significance of delayed long-loop responses to ankle displacement for the diagnosis of multiple sclerosis. Electroencephalography and Clinical Neurophysiology, 57(4), 336–342.  Diener, HC., Ackermann, H., Dichgans, J., & Guschlbauer, B. (1985). Medium- and long-latency responses to displacements of the ankle joint in patients with spinal and central lesions. Electroencephalography and Clinical Neurophysiology, 60(5), 407–416.  Fitzpatrick, RC., Taylor, JL., & McCloskey, DI. (1992). Ankle stiffness of standing humans in response to imperceptible perturbation: reflex and task-dependent components. The Journal of Physiology, 454(1), 533–547.  Fitzpatrick, RC., Rogers, DK., & McCloskey, DI. (1994). Stable human standing with lower-  60 limb muscle afferents providing the only sensory input. The Journal of Physiology, (Pt 2), 395–403.  Gill,	  J.,	  Allum,	  JHJ.,	  Carpenter,	  MG.,	  Held-­‐Ziolkowska,	  M.,	  Adkin,	  AL.,	  Honegger,	  F.,	  &	  Pierchala,	  K.	  (2001).	  Trunk	  Sway	  Measures	  of	  Postural	  Stability	  During	  Clinical	  Balance	  Tests:	  Effects	  of	  Age.	  The	  Journals	  of	  Gerontology	  Series	  A:	  Biological	  Sciences	  and	  Medical	  Sciences,	  56(7),	  M438–M447.	  	  Goble,	  DJ.,	  Coxon,	  JP.,	  Wenderoth,	  N.,	  Van	  Impe,	  A.,	  &	  Swinnen,	  SP.	  (2009).	  Proprioceptive	  sensibility	  in	  the	  elderly:	  Degeneration,	  functional	  consequences	  and	  plastic-­‐adaptive	  processes.	  Neuroscience	  &	  Biobehavioral	  Reviews,	  33(3),	  271–278.	  	  Huisinga, JM., St George, RJ., Spain, R., Overs, S., & Horak, FB. (2014). Postural Response Latencies Are Related to Balance Control During Standing and Walking in Patients With Multiple Sclerosis. Archives of Physical Medicine and Rehabilitation, 95(7), 1390–1397.  Huryn, TP., Luu, BL., Van der Loos, HFM., Blouin, J.-S., & Croft, EA. (2010). Investigating human balance using a robotic motion platform. In 2010 IEEE International Conference on Robotics and Automation (pp. 5090–5095). IEEE. Jacobs, JV., & Horak, FB. (2007). Cortical control of postural responses. Journal of Neural Transmission, 114(10), 1339–1348.  Kawato, M., & Wolpert, D. (1998). Internal models for motor control. Novartis Foundation Symposium, 218, 291-304–7.  Kuo, AD. (1995). An optimal control model for analyzing human postural balance. IEEE Transactions on Biomedical Engineering, 42(1), 87–101.  Kuo, AD (2005). An optimal state estimation model of sensory integration in human postural balance. Journal of Neural Engineering, 2(3), S235–S249.  Loram, ID, & Lakie, M. (2002). Human balancing of an inverted pendulum: position control by small, ballistic-like, throw and catch movements. The Journal of Physiology, 540(3), 1111–1124.  Luu BL. (2010). Perception, perfusion and posture (doctoral thesis). Sydney, New South Wales, Australia: Univ. of New South Wales.  Luu, BL., Huryn, TP., Van der Loos, HFM., Croft, EA., & Blouin, J.-S. (2011). Validation of a Robotic Balance System for Investigations in the Control of Human Standing Balance. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 19(4), 382–390.  Luu, BL., Inglis, JT., Huryn, TP., Van der Loos, HFM., Croft, EA., & Blouin, J.-S. (2012). Human standing is modified by an unconscious integration of congruent sensory and motor signals. The Journal of Physiology, 590(22), 5783–5794. Magnus, R. (1926). On Some Results of Studies in the Physiology of Posture, 531–536.  Morasso, PG., & Sanguineti, V. (2002). Ankle Muscle Stiffness Alone Cannot Stabilize Balance During Quiet Standing. Journal of Neurophysiology, 88(4), 2157-2162. Morasso, PG., & Schieppati, M. (1999). Can Muscle Stiffness Alone Stabilize Upright Standing? Journal of Neurophysiology, 82(3), 1622-1626. Morasso, PG., Spada, G., & Capra, R. (1999). Computing the COM from the COP in postural sway movements. Human Movement Science, 18(6), 759–767.  Nardone, A., Corrà, T., & Schieppati, M. (1990). Different activations of the soleus and gastrocnemii muscles in response to various types of stance perturbation in man. Experimental Brain Research, 80(2), 323–332.  Nardone, A., Siliotto, R., Grasso, M., & Schieppati, M. (1995). Influence of aging on leg muscle   61 reflex responses to stance perturbation. Archives of Physical Medicine and Rehabilitation, 76(2), 158–165.  Nashner, LM. (1976). Adapting reflexes controlling the human posture. Experimental Brain Research, 26(1), 59–72. Peterka, RJ. (2002). Sensorimotor Integration in Human Postural Control. Journal of Neurophysiology, 88(3).  Peterka, RJ., & Loughlin, PJ. (2004). Dynamic Regulation of Sensorimotor Integration in Human Postural Control. Journal of Neurophysiology, 91, 410–423. Peterka, RJ., & Black, FO. (1990). Age-relatied changes in human postural control: motor coordination tests. Journal of Vestibular Research, 1, 87–96.  Porter,	  MM.,	  Vandervoort,	  AA.,	  &	  Lexell,	  J.	  (2007).	  Aging	  of	  human	  muscle:	  structure,	  function	  and	  adaptability.	  Scandinavian	  Journal	  of	  Medicine	  &	  Science	  in	  Sports,	  5(3),	  129–142.	  	  Pospisil, ER., Luu, BL., Blouin, J.-S., Van der Loos, HFM., & Croft, EA. (2012). Independent ankle motion control improves robotic balance simulator. In 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society (Vol. 2012, pp. 6487–6491). IEEE.  Schmidt, RA. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82(4), 225–260. Scott, SH. (2004). Optimal feedback control and the neural basis of volitional motor control. Nature Reviews Neuroscience, 5(7), 532–546.  Shadmehr, R., Smith, MA., & Krakauer, JW. (2010). Error Correction, Sensory Prediction, and Adaptation in Motor Control. Annual Review of Neuroscience, 33(1), 89–108.  Shepherd, M. (2014). Intersensory vestibular control of standing balance (masters thesis). Vancouver, British Columbia, Canada: Univ. of British Columbia.  Sherrington, CS. (1910). Flexion-reflex of the limb, crossed extension-reflex, and reflex stepping and standing. Journal of physiology. 40(1-2), 28-121. Smith, JW (1957). The forces operating at the human ankle joint during standing. Journal of Anatomy, 91, 545-564 Stelmach, GE., Phillips, J., DiFabio, RP., & Teasdale, N. (1989). Age, functional postural reflexes, and voluntary sway. Journal of Gerontology, 44(4), B100–B106.  Taube, W. Schubert, M., Gruber, M., Beck, S., Faist, M., & Gollhofer, A.(2006). Direct corticospinal pathways contribute to neuromuscular control of perturbed stance. Journal of Applied Physiology, 101(2), 420–429.  van der Kooij, H., Jacobs, R., Koopman, B., & Grootenboer, H. (1999). A multisensory integration model of human stance control. Biological Cybernetics, 80(5), 299–308.  van der Kooij, H., & Peterka, RJ (2011). Non-linear stimulus-response behavior of the human stance control system is predicted by optimization of a system with sensory and motor noise. Journal of Computational Neuroscience, 30(3), 759–778.  Winter, DA, Patla, AE, Prince, F., Ishac, M., & Gielo-Perczak,  K. (1998). Stiffness Control of Balance in Quiet Standing.  Wolfson, LI., Whipple, R., Amerman, P., Kaplan, J., & Kleinberg, A. (1985). Gait and balance in the elderly. Two functional capacities that link sensory and motor ability to falls. Clinics in Geriatric Medicine, 1(3), 649–59.  Woollacott, MH., Shumway-Cook, A., & Nashner, LM. (1986). Aging and Posture Control:   62 Changes in Sensory Organization and Muscular Coordination. The International Journal of Aging and Human Development, 23(2), 97–114.                         63 Appendix - Individual EMG onset latencies Table A.1 - Onset latency data for all participants  Onset latencies are given by ms ± SD * One participant had unidentifiable responses and is not included in this table   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0368918/manifest

Comment

Related Items